The brain functional connectivity alterations in traumatic patients with olfactory disorder after low-level laser therapy demonstrated by fMRI

Abstract

Background

Low-level laser therapy (LLLT) has been clinically accepted to accelerate the nerve regeneration process after a nerve injury or transection. We aimed to investigate the neuronal basis and the influence of LLLT on brain functional networks in traumatic patients with olfactory dysfunction.

Methods

Twenty-four Patients with traumatic anosmia/hyposmia were exposed to pleasant olfactory stimuli during a block-designed fMRI session. After a 10-week period, patients as control group and patients who had completed the sessions of LLLT were invited for follow-up testing using the same fMRI protocol. Two-sample t-tests were conducted to explore group differences in activation responding to odorants (p-FDR-corrected <0.05). Differences of functional connectivity were compared between the two groups and the topological features of the olfactory network were calculated. Correlation analysis was performed between graph parameters and TDI score.

Results

Compared to controls, laser-treated patients showed increased activation in the cingulate, rectus gyrus, and some parts of the frontal gyrus. Shorter pathlength (p = 0.047) and increased local efficiency (p = 0.043) within the olfactory network, as well as decreased inter-network connectivity within the whole brain were observed in patients after laser surgery. Moreover, higher clustering and local efficiency were related to higher TDI score, as manifested in increased sensitivity to identify odors.

Conclusions

The results support that low-level laser induces neural reorganization process and make new connections in the olfactory structures. Furthermore, the connectivity parameters may serve as potential biomarkers for traumatic anosmia or hyposmia by revealing the underlying neural mechanisms of LLLT.

Introduction

Olfaction is one of the oldest chemical senses in humans and an important chemical warning system which helps to be cautious in dangerous situations.^1–3 There are many reasons accounting for olfactory dysfunction but the three main causes are sino-nasal diseases, upper respiratory viral infection, and head trauma. ⁴ Following traumatic brain injury (TBI), fragmentation of the olfactory nerve fibers, where they pass through the cribriform plate, is the most common cause of post-traumatic anosmia.^5,6 It is expected that following head trauma the olfactory epithelium regenerates and receptor cells probably send axons toward central part of the nervous system. However, it is conceivably assumed that the lamina cribrosa of the cribriform plate is closed, with formation of scar tissue created after injury. ⁷ In fact, this fibrotic tissue prevents from synaptic contact of the epithelial olfactory neurons with the olfactory bulb neurons. Consequently, the normal olfactory pathway is disrupted; resulting in the smell loss.^8,9 Regeneration of olfactory receptor connections with the olfactory bulb, axonal growth rate, and functional recovery following olfactory nerve transection depend on the degree and type of injury. ¹⁰

Quantitative psychophysical tests are currently using for olfactory assessment in patients complaining smell dysfunction.^11,12 The Sniffin’ Sticks test is a reliably valid test battery firstly developed by Hummel in 1997 for European community (Burghart GmbH, Wedel, Germany). ¹³ This test had been already used to diagnose the presence of olfactory disorder with three subtests including smell threshold, discrimination and identification tests. ¹⁴ Nowadays, advanced neuroimaging techniques such as Functional Magnetic Resonance Imaging (fMRI) by using blood-oxygenation level dependent (BOLD) signal, offer a non-invasive tool to detect brain activity and functional connectivity during odor presentation by olfactometer in normal subjects and patients with smell loss.^15,16 It also enables researchers to objectively investigate the effect of therapeutic options in patients with olfactory dysfunction.^17–19 There have been recent advances in the acquisition and analysis of functional neuroimaging data to better characterize the olfactory-related areas in the brain networks and its relationship with functional recovery. ²⁰

Regarding treatment of olfactory disorders, olfactory training introduced by Hummel et al. in 2009 was a promising treatment for patients with olfactory loss and specially who complained of smell loss after upper respiratory tract infection. ²¹ Functional changes and effective connectivity networks in these patients after olfactory training have been also demonstrated in previous studies.^17,22,23

In recent years, LLLT, known as photobiomodulation (PBM), firstly used in the 1980s, has been already the subject of numerous recent studies; especially in the areas of physical medicine and rehabilitation, as it revealed an immediate protective effect, reducing the formation of edema and scar tissue in the region of injury and increasing axonal metabolism.^24,25 In addition, it has anti-inflammatory properties and increases mitotic activity, resulting in a speedy nerve regeneration process in peripheral nerves repaired by surgery.^26–29 Thus, low intensity radiation with an optimized wavelength can trigger cellular proliferation and differentiation, which would finally result in activation of transcriptional factors. ³⁰ Then expression of olfactory receptors patterns may be altered. ³¹

Among studies recently performed, researchers had already never engrossed in the clinical effects of laser on injured olfactory nerves in patients with post-traumatic olfactory loss. Taking into account the widespread incidence of post-traumatic olfactory loss and its functional disorders, the current study aimed to evaluate the efficacy of LLLT on brain activation responding to odors, as well as functional connectivity changes within the whole-brain and the olfactory network using task-based fMRI experiment. Also, the associations between functional measures and clinical status following laser surgery are measured. According to a priori hypothesis, increasing neural regeneration and functional recovery induced by LLLT in treated patients would cause specific alterations in the functional connectivity within the whole brain and the olfactory networks. Also, it is conceivably expected that following laser surgery, significant activation in the main olfactory structures will be found in treated patients.

Materials and methods

Ethic and subjects

The study design met the requirements of the Declaration of Helsinki and had been approved by the Ethics Committee of the Medical Faculty at the Iran University of Medical sciences (IR.IUMS.FMD.REC.1397.099). All subjects were informed about the aims of the study and gave their written informed consent prior to inclusion. We performed two testing sessions, which are described in details in the following paragraphs.

Pre-scanning session

Patients with post-traumatic anosmia or hyposmia (determined from the TDI scores, see psychophysical measures) who referred to our ENT research center were recruited in the study. Twenty-four patients were divided into two cohort of subjects: 12 patients as control group (mean age 27.25 years; SD 8.04; mean disease duration 8.3 months; SD 5.6) and 12 patients as experimental group (mean age 29.58 years; SD 8.6; mean disease duration 6.4 months; SD 4.01). All the patients who completed all measurements were invited to the fMRI testing session and underwent further analysis. None of the women were pregnant, and none of the participants had diabetes or other significant health problems currently or medical history that may be associated with disorders of olfactory function (e.g., neurodegenerative disorders such as Alzheimer and Parkinson disease). Furthermore, each underwent a standard ENT examination with endoscopy of the nasal cavity to determine the cause of olfactory dysfunction. Thus, individuals with polyps, acute allergic rhinitis, major septum deviations, or superior respiratory tract infection as well as congenital anosmia were excluded from the study. All participants were right-handed, non-smoker and reported no claustrophobia of MRI examinations.

Post-scanning session

After completing first scanning session, the experimental group underwent eight sessions of LLLT during a period of 4–6 weeks. After 10 weeks from beginning the intervention, all patients participated in the second fMRI testing session. Two patients (one of each group) had to be excluded from the data set due to incomplete fMRI measurements and difficulties in follow-up LLLT. All patients in both groups were investigated with psychophysical test and fMRI experiment in two sessions.

Psychophysical measures

The olfactory function was evaluated by the Iranian version of the Sniffin’ Sticks test, taking into consideration three different parameters: smell identification, discrimination, and threshold test. ³² All participants were instructed to refrain from eating and drinking anything except to water for 1 hour prior to testing and each were tested individually, in a quiet and ventilated room. The protocol was based on test battery firstly developed by Hummel et al. ¹³ Sum of all the three subtests gives the TDI score which determine the olfactory status. A TDI score of 16.5 or less presented anosmia and between 16.5 and 30.5 showed hyposmia.

Laser surgical procedure

Photo-stimulation by LLLT, manufactured in our country, did not cause any pain or distress; therefore, it was not necessary to use anesthesia. However, as patients might be frightened and as a convenience for them, before LLLT, patients were asked to lie down on the examination couch and topical anesthesia was applied to bottom of the nostrils for approximately 10 min with cotton pads soaked in 0.5% tetracaine and 0.5% phenylephrine solution. When the nose completely analogized, the laser output power was set to 80 mw in the continuous wave and contact mode. The mean operation time was 20 min for each patient. The laser emitted light at a wavelength of 660 nm with a 1 mm fiber opening diameter. The laser tip was made to slide over the surface of the superior turbinate rapidly and smoothly to prevent cauterization and bleeding. No complications were observed (e.g., major bleeding), and no nasal packing was necessary. Each patient underwent the surgery two sessions per week, during a period of 4–6 weeks (8 sessions totally). Assessment of the olfactory function was also performed 10 weeks after beginning the LLLT using the psychophysical test and fMRI experiment.

Experimental design

Olfactory stimulation was performed using a hand-made multimodal MR-compatible olfactometer. ²³ Functional images were subsequently obtained in a block-design task consisting of two odor exposure blocks (on-period) and two normal breathing blocks (off-period). Two samples of odor (vanillin and rose) during 10 s on-periods, each followed by 30 s of odorless air during the off-period, with continuous air flow (30 L/min) were delivered to both nostrils through a nasal mask placed at the end of a silicon air path. The block was repeated 4 times and total session duration was 320 s (Figure 1). Switching from odor to non-odor conditions provides no visual, tactile, auditory, or thermal cues to the subjects. During fMRI scanning, all patients were asked to breathe normally through the nose and keep calm.

Figure 1.

Olfactory stimulation paradigm. A schematic drawing of block-design task consists of 30 s half-blocks of no-odor and 10 s of odor condition.

MRI acquisition

All MRI data acquisitions were performed on a 3.0 T MR scanner (SIEMENS-Prisma) using a 20-channel head coil. Before acquisition of fMRI data, a high-resolution T1-weighted structural image (magnetization-prepared rapid gradient-echo [MPRAGE]) was acquired (160 sagittal slices, voxel size = 1 mm × 1 mm × 1 mm, repetition time (TR) = 1800 ms, echo time (TE) = 3.53 ms, flip angle = 7°, slice thickness = 1 mm and field of view (FOV) = 256 × 256). Eco-planar images were acquired in an ascending sequence of 45 slices, (voxel size = 3 mm × 3 mm, echo time (TE)/repetition time (TR) = 30/3000 ms, matrix size = 64 × 64, flip angle = 90°, field of view (FOV) = 448 × 448 mm). A total of 109 brain volume sequences were collected during 5.3 min. Slices were aligned parallel to the anterior/posterior commissure line. Three additional volumes were acquired to compensate the delay time of odor presentation by the olfactometer.

Analysis

fMRI data preprocessing and activation analysis

fMRI data preprocessing was carried out using FEAT (FMRI Expert Analysis Tool) Version 6.00, part of FSL (FMRIB’s Software Library, www.fmrib.ox.ac.uk/fsl). ³³ Non-brain tissue was removed from high-resolution images using FSL’s brain extraction tool (BET). ³⁴ The functional data sets were corrected for slice-time differences and spatially smoothed with a Gaussian kernel having a full width at half maximum (FWHM) of 5 mm. The functional data sets were motion-corrected using MCFLIRT.^35,36 Motion correction results indicated that movement during scans was minimal and did not exceed an absolute or relative limit of 0.5 mm for any subject; therefore, no scan was discarded due to movement. The functional images from each time series were normalized to the mean of the series, temporally smoothed using a high pass filter with a cutoff of 80 s (corresponding frequency of 0.0125 Hz). The functional images were then registered to the high-resolution 3D volume using FLIRT (FMRIB’s Linear Image Registration Tool) and, prior to group analysis, warped to the standard anatomical space (Montreal Neurologic Institute, MNI), using the FSL normalization algorithm. ³⁶ Statistical analyses were performed using FEAT and general linear modeling (GLM) algorithms ³⁷ to investigate the contrasts between vanillin and rose stimulus at individual and group levels. Cluster thresholding was performed to clarify those clusters with statistically significant activations (Z >2.3; p false discovery rate-corrected <0.05). ³⁸ We tested group differences using two-sample t tests (p <0 .05 corrected), with the same statistical threshold.

Regions of interest (ROIs) analysis

Further, ROI analysis was performed to emphasize on the pre-reported and critical olfactory regions. The related masks were built using WFU Pickatlas 3.0.5b software. ³⁹ The nine predefined ROIs included piriform cortex [Broadman area 34], entorhinal cortex [Broadman area 28], amygdala, hippocampus, para-hippocampal cortex, orbitofrontal cortex and insula, representing primary and secondary olfactory cortex, respectively. Inferior frontal gyrus and cingulate gyrus were also included as regions associated with odor memory processing and emotional valence. Predefined masks were mapped to each subject’s space using FLIRT ³⁶ and the pre-computed standard to subject’s space matrix during preprocessing. The GLM analysis was performed according to the ROIs.

Functional Connectivity and graph theoretical analysis

Functional connectivity was measured as the Pearson correlation coefficient between the time series of the selected seed region and all other voxels (or ROIs) in the brain. We chose piriform cortex as the seed region, ²² primarily responsible for olfaction. The whole brain network with connectivity matrix of 109 × 109 ROIs was constructed using the mentioned WFU Pickatlas software. Seed-based Correlation Analysis was performed as one of the most common ways to explore functional connectivity within the brain using the CONN toolbox ⁴⁰ (http://www.nitrc.org/projects/conn). After performing the preprocessing steps on functional and structural data, nuisance parameters extracted from the motion correction process were regressed out. Then, the mean time course extracted from the predefined seed region was correlated with the time course of all other ROIs in the whole brain. The results provided separate functional connectivity maps within the seed region correlated to other ROIs time points for the two patient groups. ROI-to-ROI graph metrics analysis was also performed and four network properties were measured to explore functional integration and segregation. ⁴¹ Using an intermediate (K = 0.15) cost threshold level, the global and local efficiency, clustering coefficient and pathlength of each ROI within the network, were computed and averaged across all subjects. Voxel-to-voxel analyses were also carried out to detect functional connectivity within the whole brain and the olfactory network. Group Multivariate parametric analysis (MVPA) estimates, for each seed-voxel, a multivariate representation of the connectivity pattern between this voxel and the entire brain.

Statistical Analysis

Statistical analysis was performed using the Statistical Package for the Social Sciences, version 24.0 (SPSS; IBM, Armonk, New York). Kolmogorov–Smirnov test was used to check normal distribution of data (p-value >0.05). Parametric statistical test (paired-samples t test) was applied to compare olfactory performance scores, fMRI activations and functional connectivity, between baseline and follow-up sessions for each group. To compare TDI and connectivity measures between two groups, parametric independent sample t test was conducted. p values less than 0.05 were considered to be statistically significant.

Results

Clinical characteristics and results of olfactory testing

At the first testing session, all patients of both groups achieved low TDI scores. There was no significant difference between odor scores of two groups (p = 0.630). A comparison of olfactory performance measurements in laser group before and after laser surgery, revealed a significant improvement in the TDI score (p = 0.015) and odor identification test (p = 0.041). This improvement was also obtained in two other smell tests (threshold and discrimination), but the difference was no significant between the two testing sessions. There was also no significant difference in TDI scores between two groups at the second testing session (p = 0.274). Furthermore, control group showed no significant difference compared with their baseline for the TDI score (p = 0.257). Detailed results of olfactory performance measures and the demographic information of the patients are presented in Table 1. A comparison of TDI scores has been given in Table 2.

Table 1.

Demographic data and the results of olfactory performance measures for control and laser-treated groups.

	Control group		Laser group
	Mean (SD)		Mean (SD)
	1st test	2nd test	1st test	2nd test
TDI score	13.03 (4.63)	14.43 (3.94)	14.56 (7.43)	17.84 (7.49)
Odor threshold	2.34 (2.29)	1.68 (1.94)	1.31 (0.53)	2.09 (2.03)
Odor discrimination	7.0 (2.20)	6.87 (2.10)	6.62 (2.72)	7.50 (2.56)
Odor identification	3.75 (2.54)	5.87 (3.97)	6.62 (5.09)	8.37 (4.10)
Age (yrs.)	27.25 (8.04)		29.58 (8.6)
Body mass index (BMI)	22.4 (4.7)		24.5 (3.3)
Disease duration (months)	8.37 (5.6)		6.4 (4.01)

Table 2.

Comparison of TDI scores between and within two groups at the first and second testing sessions.

Between-group comparison		Within-group comparison
First testing session	p = 0.630	Laser group (before and after treatment)	p = 0.015
Second testing session	p = 0.274	Control group (first and second testing session)	p = 0.257

fMRI data

Activation in laser group

Whole brain analysis was performed to explore the evoked activations by vanillin and rose stimulus. The results revealed that sniffing with odor presentation triggered activation in the right supramarginal gyrus, left superior frontal gyrus and left middle occipital gyrus. After treatment, vanillin and rose-sniffing produced activation in the bilateral postcentral gyrus and more robust activations were found in some olfaction-related structures listed for each odor condition in Table 3. ROI analysis of the BOLD signal time course showed that the bilateral medial frontal gyrus had been activated. After treatment, the cingulate gyrus and sub-gyral of frontal lobe responding to odorant stimulation, were significantly activated compared to baseline. Figure 2 shows cluster-thresholded (z > 2.3; p < 0.05) group activation with the defined ROIs before and after treatment.

Table 3.

Location and volume of the activation clusters of olfactory brain structures triggered by odor-sniffing in laser group (post-treatment compared to pre-treatment).

Condition	L/R	Region	MNI coordinates (mm)			Cluster size (voxel)	Z-max	Significance (p-value)
			x	y	z
Vanillin	R	Postcentral gyrus	20	−42	60	700	3.99	<0.05
	L	Postcentral gyrus	−46	−36	58	449	4.17	<0.05
	L	Middle temporal gyrus	−56	−18	−14	432	3.69	<0.05
Rose	L	Fusiform gyrus	−30	−32	−20	2981	4.09	<0.05
	L	Superior frontal gyrus	−20	10	48	1192	4.23	<0.05
	R	Cingulate gyrus	14	−4	50	935	3.64	<0.05
	L	Middle temporal gyrus	−56	−18	−14	558	3.85	<0.05
	R	Cerebellum	12	−48	−48	7081	4.63	<0.05
	R	Thalamus	22	−12	10	1359	3.89	<0.05

Figure 2.

ROI-based activation analysis. The figure shows activation in patients before (a); and after (b) treatment. Activation was found in the bilateral medial frontal gyrus before laser therapy (a). But after treatment, rose stimulation produces significant activation in the cingulate gyrus (b).

Activation in control group

Whole-brain analysis revealed activation in the right medial orbitofrontal, right superior temporal gyrus, and left middle frontal gyrus in the first scanning session. The right supramarginal gyrus showed BOLD signal in both scanning sessions. After a 10-week period compared to baseline, the left cerebellum was activated during the presentation of odorant stimulus. ROI-based analysis revealed no significant difference of activated areas in this group in comparison of two sessions.

Between-group differences

Two-sample t test revealed activation in the left cerebellum and right fusiform gyrus of control patients and bilateral cerebellum of laser group due to both odor stimuli. After treatment, we found significant activation in treated patients compared to controls (Table 4). In ROI analysis, the olfactory parts of frontal gyrus, predicted to be activated after treatment; including the bilateral medial frontal gyrus, left rectus gyrus, right lateral orbitofrontal gyrus and left superior frontal gyrus in contrast to control group (Figure 3). We also observed increased activation with local maxima in the cingulate gyrus responding to rose stimulation in treated group. Activation was largely absent in regions critical to olfactory processing in controls compared to laser group at the baseline and second testing session.

Table 4.

Comparison of location and volume of the activation clusters of olfactory brain structures triggered by odor-sniffing between laser-treated group and controls in post-scan.

Condition	L/R	Region	MNI coordinates (mm)			Cluster size (voxel)	Z-max	Significance (p-value)
			x	y	z
Vanillin	R	Precuneus (parietal)	18	−56	52	895	4.07	<0.05
	L	Postcentral gyrus	−46	−34	58	350	4.55	<0.05
Rose	L	Sub-gyral (temporal gyrus)	36	−50	4	7151	4.41	<0.05
	L	Postcentral gyrus	−46	−33	56	326	4.23	<0.05
	R	Cingulate gyrus	20	−16	42	1282	4.06	<0.05

Figure 3.

ROI-based activation analysis for anosmic patients in laser-treated group after treatment compared with control group shows significant activation in the left superior frontal gyrus, right medial and lateral orbitofrontal gyrus (a), left medial frontal gyrus (b) and left rectus (c) in vanillin-sniffing. The cingulate gyrus reveals activation in response to rose stimulation.

Changes in the olfactory network functional connectivity

Seed-to-voxel bivariate correlation analysis

The piriform cortex (Broadman area 34) [selected on a standard brain using the WFU Pickatlas] was used as the seed. At baseline, laser group exhibited positive correlation from the left precuneus, right para-hippocampal gyrus and lateral occipital cortex toward the piriform seed. After completing the treatment, seed correlation toward olfactory areas was found in the right para-hippocampal gyrus, right amygdala, left hippocampus, right postcentral gyrus, right superior frontal gyrus, left caudate, and anterior cingulate gyrus. In control patients, main olfactory processing areas (bilaterally including the orbitofrontal cortex, temporal pole, amygdala and hippocampus and left para-hippocampal gyrus) were correlated to the piriform in condition of both odor-sniffing. But, after a 10-week period, this group exhibited no significant changes compared to the baseline. Some areas just in the right hemisphere including the right superior frontal gyrus, putamen, precentral gyrus, para-hippocampal gyrus, and the brain-stem cortex correlated to the piriform. Results are thresholded at FDR-corrected cluster-level p < 0.05 (with uncorrected two-sided p < 0.001 height threshold).

ROI-to-ROI correlation analysis

ROI-to-ROI functional connectivity (bivariate correlation measure) was measured as the Pearson correlation coefficient between the seed and a set of 64 ROIs defining the Brodmann areas (p FDR-corrected seed-level <0.05). Although neither group revealed significant connections in vanillin contrast, there were four connections in rose contrast for each group. In control patients, connections from the seed to the olfactory processing areas (left amygdala, T = 5.89; right amygdala, T = 10.89; right hippocampus, T = 8.95; left para-hippocampal gyrus, T = 5.38) were observed (Figure 4 left (b)). Prior to the treatment in laser group, a network including (right para-hippocampal gyrus, T = 4.88; right middle temporal gyrus, T = 5.02; left temporal pole, T = 5.11; left lateral occipital cortex, T = 6.12) was noted (Figure 4 left (d)). After completing the treatment, two connections (left amygdala, T = 6.25; right amygdala, T = 4.46) were found in response to vanillin (Figure 4 right (c´)). Also, one connection from the piriform to a higher-level olfactory structure with an increase of intensity was identified (left hippocampus, T = 8.05) in rose stimulation (Figure 4 right (d´)). The highest decrease of functional connections was obtained for the control patients. After the time interval, no connection was observed in this group (Figure 4 right (a´, b´)).

Figure 4.

ROI-to-ROI functional connectivity with piriform seed area during olfactory stimulation. Left: controls (a, b) and experimental patients (c, d) before intervention at the first session. Right: controls (a′, b′) and experimental patients after laser treatment (c′, d′). Results are thresholded at FDR-corrected p < 0.05. The central dot represents the selected ROI (Broadman area 34). The rest dots capture the statistically significant functionally correlated brain areas.

ROI-to-ROI graph metrics analysis

The entire metrics of ROI-to-ROI functional connectivity values (bivariate correlation measure) were computed for each subject using the Brodmann area ROIs, and thresholded at a fixed network-level cost value (k = 0.15) to define an undirected graph characterizing the entire network of functional connections between positively associated ROIs (ROIs defined from MNI atlas Brodmann areas). After treatment, patients showed a significantly increased local efficiency, specifically in the vanillin contrast (0.677 ± 0.114 vs 0.613 ± 0.058) (p = 0.043) (Figure 5). Also, clustering coefficients increased in the olfactory network of treated patients, but they were not significant compared to control group (0.567 ± 0.101 vs 0.509 ± 0.086) (p > 0.05). Average pathlength decreased for both groups, but it was only significant for laser-treated patients compared with prior to treatment (2.71 ± 0.4407 vs 3.10 ± 0.2416) (p = 0.047)

Figure 5.

ROI-to-ROI graph analysis. Local efficiency (a measure of integrated information within the network) in the olfactory network shows significant difference in the number of connections between two groups; top: laser group before and after treatment, bottom: control patients in pre- and post-scanning sessions in the vanillin contrast. Boxplot shows the local efficiency of each group in two sessions. The cost level thresholded at K = 0.15 and significance level at p-value <0.05.

Correlation between network parameters and TDI Score

In laser group, the local efficiency of the olfactory network was significantly correlated with the TDI score (r² = 0.51, p = 0.046) and remained significant after treatment (r² = 0.51, p = 0.045). This correlation was also observed in the clustering coefficient associated with the TDI score (r² = 0.53, p = 0.039) and remained significant after following up the patients (r² = 0.50, p = 0.049) (Figure 6). No significant relationships were observed for control patients.

Figure 6.

The relationship between TDI score and graph metrices. Scatter plots show linear relationships between TDI score and network parameters for traumatic anosmic patients before and after treatment. R-squared for each parameter is presented. Local efficiency and clustering coefficient show significant and positive correlation with TDI score.

Voxel-to-voxel analysis

Group-MVPA analysis was performed to estimate, for each seed-voxel, a multivariate representation of the connectivity pattern between this voxel and the entire brain. Figure 7 shows the spatial extent of widespread network connectivity in patients prior to treatment, encompassing largely nonolfactory regions in the default mode, sensorimotor, salience, and visual networks. Also, some parts of the frontal gyrus and olfactory-related areas were involved in the connectivity networks. The results showed high degree of hemispheric symmetry and functional connectivity within both the olfactory network and the whole brain far beyond the olfactory areas. After treatment, the most of nonolfactory connectivity networks dispersed and some of the olfactory-related areas-OFC, IFG, para-hippocampal gyrus, frontal pole, and precentral gyrus were retained. Conversely, in control patients, the functional connectivity within the networks from olfactory-related areas-OFC and frontal pole was increased to recruitment of other regions, quite similar to those seen in the laser group before treatment.

Figure 7.

Voxel-to-voxel connectivity analysis shows the cluster sizes recruited in functional networks during odor-sniffing for both patient groups. Top: before (a) and after (b) the treatment in laser group. Below: first (c) and second (d) scan in control group. Bold regions show the recruited cluster sizes functionally connected, for each group and each session (voxel thresholded at P-uncorrected <0.001).

Discussion

In this study, we could demonstrate a significant functional improvement in the central olfactory system during odor-sniffing in laser-treated patients compared to baseline and particularly to control patients. Whole brain analysis of laser group after treatment showed significant activation in some regions receiving secondary olfactory projections, which is similar to those reported in previous observations.^15,42 Also, significant BOLD signal in the cerebellum, as previously mentioned, ¹⁵ may be considered as an evidence of extensive activation of the frontal lobe. Two-sample t test at group level revealed significant activation in laser-treated patients in the right precuneus and left postcentral gyrus, as well as the sub-gyral of temporal lobe and cingulate gyrus in response to vanillin and rose stimuli. Furthermore, ROI analysis revealed more promising results, which means that laser-treated patients compared to control ones showed significant activation in the olfactory structure of frontal gyrus including the bilateral medial frontal gyrus, right lateral orbitofrontal gyrus, and left superior frontal gyrus in response to vanillin. Activation in frontal regions, previously found activated in fMRI studies, represents brain processes linked to olfactory network and suggests that the frontal cortex plays an important role in recognition of olfactory stimuli.^43,44 Also, significant activation founded in the cingulate gyrus in response to rose odorant, predicted to be activated after treatment, can mediate the process of rapid attention to olfactory stimuli. ⁴⁵ The cingulate cortex, known as the limbic lobe, is activated by olfactory stimulation; independent of valence or just responding to pleasant odors. ⁴⁴ We assumed that the close anatomic relations between the olfactory and emotional system may be accounted for the important links found between these two functions in our study. No significant activation, as expected, was observed in olfactory structures in control group compared to the baseline.

In addition, after completing the treatment, based on ROI-to-ROI correlation analysis, we found new connections from the piriform to the primary and secondary olfactory cortices (including the limbic areas: the bilateral amygdala and left hippocampus) in both contrasts. These findings demonstrated that we were able to show a neural reorganization process as well as a reconstruction of the olfactory-specific processing network induced by LLLT, as the olfactory training partially recovered olfactory function in these patients.^21,22 Conversely, decreased activation and the absence of functional connections between the olfactory areas in control patients after a time period are consistent with recent studies in anosmic patients.^17,46

In the current study, graph theory-based network analysis demonstrated a pattern of the dynamics behind functional connectivity changes, as investigated in the previous study of resting-state fMRI. ⁴¹ The intra-network connectivity, expressed by increased local efficiency, within the olfactory network was increased following laser surgery. Furthermore, a shorter average pathlength after treatment showed the ability to rapidly combine pieces of specialized information from distributed brain regions of the olfactory network as well as reduction of brain consumption. ⁴⁷ Some studies reported decreased global efficiency in patients with mild traumatic brain injury. ⁴⁸ Our results of the voxel-based graph analysis in both groups are consistent with the study of anosmic patients conducted by Park et al. ⁴¹ As the brain is a complex network, a decreased average pathlength along with increased intra-network connectivity, expressed by increased local efficiency, within the olfactory network, as well as decreasing tendency of functional connectivity within the whole brain reflect a more organized connectivity network and a less random dynamic interactions responding to external odorant stimulation in patients treated with LLLT. ⁴⁹ We observed no significant differences of global efficiency in each group. However, the increased inter-network connectivity, explained as the increased connectivity in the global brain functional network in control patients compared to baseline, may support the brain plasticity for adapting to the local damage. Recruitment of other brain areas could be a manifestation of the compensatory mechanism as previously supported in many neurodegenerative diseases and the olfactory dysfunction.^22,50 It is also worth noting that two graph measures had a significant correlation with the TDI score, as suggested in another research. ⁴¹ These measures provide imaging biomarkers to objectively evaluate the clinical status with considering the possible connectivity changes within the olfactory network of traumatic patients with anosmia.

In addition to the main effects of the LLLT obviously observed alterations in the connectivity networks, we demonstrated a statistically significant smell function improvement for the TDI score (p = 0.015) and odor identification test (p = 0.041). According to previous studies, it is supposed that the odor detection threshold represents basic olfactory function, while the other two abilities recruit more complex processing of olfactory information and determine higher olfactory function.^51,52 Our results indicate that LLLT improves the higher olfactory function such as identification ability and can logically affect the basic olfactory function primarily, if longer periods of laser surgery be performed. These findings are similar to previous studies revealed an increase in odor identification. ²¹ According to many previous studies conducted on rat olfactory epithelium, the primary sensory elements are required for functional recovery of the olfactory system after damage restoration. Thus, higher olfactory function may require a longer period to recover, at least within 3 months after the lesion. ⁵³ Also, a significant decrease in injury-associated tissue and inflammatory factors within 42 days after injury, despite being significantly better in the mild injury model, suggesting that restoration of the olfactory receptors depends upon the degree and the type of injury.^8,10,54 The morphological and functional effects of laser phototherapy on axonal regeneration have been investigated in different studies on peripheral and facial nerves.^25,30,31 Using diode laser surgery in patients with allergic rhinitis and long-term follow-up after 6 years proved that diode laser is a useful procedure that has a long-lasting effect on allergic rhinitis. ⁵⁵ Thus, it will be notable in future studies to investigate the long-term effects of LLLT, which may shed light on the impact of laser surgery on neural regeneration and higher level of olfactory processing. One limitation of this study is that we recruited quite a small study population. However, we could perform statistical analysis to find relatively significant differences of activation and connectivity network in the task-based fMRI experiment. This is the first demonstration of changing network connectivity within the olfactory network and the whole brain, probably reflecting reorganization mechanisms using LLLT in patients with traumatic anosmia/hyposmia.

Conclusion

The results of this study suggest that 660-nm laser phototherapy as a therapeutic method may enhance the regenerative process of the olfactory nerves and subsequently induce modifications in functional connectivity within the whole brain and the olfactory networks in patients with traumatic olfactory dysfunction. Consistent with the improvement of smell function at the behavioral level, the activation of some basic and higher olfactory structures including the orbitofrontal and the cingulate gyrus was significantly increased. We present a task-related fMRI evidence of traumatic olfactory disorder-induced connectivity changes in the olfactory functional networks. Our findings may acknowledge that connectivity measures serve as potential biomarkers to evaluate the effects of a therapeutic method in patients with traumatic anosmia or hyposmia.

ORCID iD

Seyedeh Fahimeh Hosseini https://orcid.org/0000-0002-3253-2824

Data availability statement

The corresponding authors are ready to share data of this manuscript acquired at (ENT and Head and Neck Research Center and Department, The Five Senses Health Institute, School of Medicine, Iran University of Medical Sciences, and National Brain Mapping Laboratory, Tehran, Iran) upon reasonable request.