Effects of chronic peripheral olfactory loss on functional brain networks

(1)

EFFECTS OF CHRONIC PERIPHERAL OLFACTORY LOSS

ON FUNCTIONAL BRAIN NETWORKS

K. KOLLNDORFER,a,b_{A. JAKAB,}a_{C. A. MUELLER,}c S. TRATTNIGdAND V. SCHO¨PFa,e,f*

a_{Department of Biomedical Imaging and Image-guided} Therapy, Medical University of Vienna, Austria b

Department of Pediatric and Adolescent Medicine, Medical University of Vienna, Austria

c

Department of Otorhinolaryngology, Medical University of Vienna, Austria

d_{High-Field MR Center, Department of Biomedical Imaging} and Image-guided Therapy, Medical University of Vienna, Austria e_{Institute of Psychology, University of Graz, Austria}

f_{BioTechMed, Graz, Austria}

Abstract—The effects of sensory loss on central processing in various sensory systems have already been described. The olfactory system holds the special ability to be activated by a sensorimotor act, without the presentation of an odor. In this study, we investigated brain changes related to chronic peripheral smell loss. We included 11 anosmic patients (eight female, three male; mean age, 43.5 years) with smell loss after an infection of the upper respiratory tract (mean disease duration, 4.64 years) and 14 healthy controls (seven female, seven male; mean age, 30.1 years) in a functional magnetic resonance imaging experiment with a sniffing paradigm. Data were analyzed using group-independent component analysis and functional connectivity analysis. Our results revealed a spatially intact olfactory network in patients, whereas major aberrations due to peripheral loss were observed in functional connec-tivity through a variety of distributed brain areas. This is the first study to show the re-organization caused by the lack of peripheral input. The results of this study indicate that anosmic patients hold the ability to activate an olfaction-related functional network through the sensori-motor component of odor-perception (sniffing). The areas involved were not different from those that emerged in healthy controls. However, functional connectivity appears to be different between the two groups, with a decrease in functional connectivity in the brain in patients with chronic peripheral sensory loss. We can further conclude that the loss of the sense of smell may induce far-reaching effects in the whole brain, which lead to compensatory mechanisms from other sensory systems due to the close

interconnectivity of the olfactory system with other functional networks. _Ó 2015 The Authors. Published by Elsevier Ltd. on behalf of IBRO. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).

Key words: functional connectivity, piriform cortex, olfaction, anosmia, snifﬁng.

INTRODUCTION

Our sensory systems are highly plastic (Hubel and Wiesel, 1970; Goldstone, 1998). This plasticity, which can be observed at both the cellular and cognitive levels (Bende and Nordin, 1997; Cadiou et al., 2014), provides adaptive opportunities to optimize sensory function when learning and encountering new experiences (Gilbert and Sigman, 2007). In contrast to these gains in function, trauma, injury, disease, and sensory deprivation can induce plasticity among sensory systems in a reductive manner (Merabet and Pascual-Leone, 2010).

The olfactory system is extraordinarily plastic, due to mechanisms that have been the subject of extensive investigation (Barkai and Saar, 2001; Mainland et al., 2002; Wilson et al., 2004; Li et al., 2006). For instance, despite acute nasal occlusion, both human olfactory acu-ity and primary cortical odor representations persist at normal levels, likely due to compensatory top-down mechanisms (Wu et al., 2012). Based on previous ﬁnd-ings, in which alterations in functional connectivity were induced by an olfactory training program in anosmic patients (Kollndorfer et al., 2014), we have hypothesized that the extent and mechanisms of olfactory plasticity will hold critical clues about the brain mechanisms responsi-ble for neural function and recovery. This is especially rel-evant given the loss of olfactory function in a great number of neurological conditions (Doty, 2008).

The investigation into chemosensory processing in patients with smell loss is challenging, as they are not able to consciously perceive chemosensory stimuli. Thus, differences in neural activation patterns are expected between anosmic patients and healthy controls. However, the olfactory system has the special characteristic that it can be activated without the application of an odor solely by the primary sensorimotor component for olfaction – sniffing. Previous research has revealed that sniffing is not only a simple breathing technique to deliver the stimulus, but is also important for inducing neural activity in olfactory brain areas

http://dx.doi.org/10.1016/j.neuroscience.2015.09.045

0306-4522/Ó2015 The Authors. Published by Elsevier Ltd. on behalf of IBRO.

This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). *Correspondence to: V. Scho¨pf, Institute of Psychology, University of

Graz, Universita¨tsplatz 2, 8010 Graz, Austria. Tel: +43-(0)316-3808490.

E-mail address:[email protected](V. Scho¨pf). Abbreviations:EPI, echo-planar imaging; FCA, functional connectivity analysis; GIFT, Group ICA for fMRI Toolbox; GLM, general linear models; ICA, independent component analysis; ICs, independent components; NBS, Network-Based Statistics; TDI, Threshold-Detec tion–Identiﬁcation; ROIs, regions of interest; SD, standard deviation.

(2)

(Sobel et al., 1998a,b; Mainland and Sobel, 2006). Animal studies have shown that the sniffing amplitude influences activation in the olfactory bulb (Wesson et al., 2008). Neural activation in primary olfactory areas is not necessar-ily contingent on odor stimulation. Previous literature sug-gests that the piriform cortex (Sobel et al., 1998a) is activated by sniffing without the presentation of an odor.

In this study, we took advantage of olfactory network activation evoked solely by the sensorimotor act of olfaction—sniﬃng (Sobel et al., 1998a,b)—to determine the impact of chronic olfactory sensory loss on olfactory network activity. Similar approaches have been used in previous studies, including exploring motor imagery as an appropriate condition with which to investigate motor networks in patients with upper limb amputation (Diers et al., 2010). We performed fMRI in anosmic patients with long-term smell loss due to an infection of the upper res-piratory tract, and in healthy human subjects. We hypoth-esized that there would be sniﬀ-induced neural activation despite peripheral olfactory loss in anosmic patients. To investigate processing networks and to gain deeper insight into the alterations caused by peripheral sensory loss, data-driven analyses, independent component anal-ysis (ICA), and functional connectivity analanal-ysis (FCA) were performed.

EXPERIMENTAL PROCEDURES

Subjects

Nineteen healthy subjects and 11 patients with smell loss after an upper respiratory tract infection participated in the study. Data from seven anosmic patients, who participated in another study, have been previously presented inKollndorfer et al. (2014). All control subjects had normal olfactory function and all participants had no history of neurological or psychiatric diseases. As depres-sive symptoms are known to interfere with olfactory func-tion (Deems et al., 1991; Pollatos et al., 2007; Croy et al., 2014), the Beck Depression Inventory II (BDI-II; Beck et al., 1996) was performed with all subjects. Only partic-ipants, who did not surpass the cut-oﬀ score of 13, which is deﬁned as the threshold for mild depressive symptoms, were included in the study. Five healthy subjects had to be excluded from the data set. One did not reach nor-mosmic values for olfactory performance, and four had to be excluded due to incomplete fMRI measurements, resulting in a total of 14 healthy subjects included in the analysis. Eleven anosmic patients (eight female, three male; mean age, 43.5 years, standard deviation (SD), 13.4; mean disease duration, 4.64 years, SD, 3.36) and 14 healthy controls (seven female, seven male; mean age, 30.1 years; SD, 6.7) completed all measurements. The only patients included in this study were those diag-nosed with anosmia, the complete loss of olfactory func-tion (Kobal et al., 2000). An otorhinolaryngologist (C.A. M.) examined all patients, and the exam included an endoscopic examination of the nasal cavity, to determine the cause of olfactory dysfunction, and a clinical olfactory performance measure, in order to ensure that all partici-pants met the criteria for anosmia or normosmia. All patients included in the analyses were determined to have

acquired anosmia due to an infection of the upper respira-tory tract. The study was performed in accordance with the Declaration of Helsinki (1964), including current revi-sions and the EC-GCP guidelines, and was approved by the Ethics Committee of the Medical University of Vienna. All subjects were informed about the aim of the study and gave their written, informed consent prior to inclusion. Olfactory performance

Olfactory performance was assessed using the Sniffin’ Sticks test battery (Burghart Instruments, Wedel, Germany). This clinically approved test battery includes three subtests that assess nasal chemosensory function—detection threshold, odor discrimination, and odor identification—using pen-like devices for odor presentation (Kobal et al., 1996, 2000; Hummel et al., 1997). All olfactory performance tests were performed using a standardized computer-controlled test protocol (Hummel et al., 2012) to prevent influence by the investi-gator. The olfactory detection threshold ofn-butanol was assessed using a single-staircase, three-alternative, forced-choice procedure. Next, odor discrimination ability was obtained using 16 triplets of odorants (two pens con-tained the same odorant; the third pen concon-tained an odd odorant). The participants’ task was to detect the odd pen (forced choice). The odor identification task is com-posed of 16 common odors and uses a multiple-choice answering format, with a list of four descriptors for each odor. For the detection threshold, scores can range from 1 to 16, and a score between 0 and 16 can be achieved for the other two subtests. The results of all three subtests were summed to obtain a Threshold-Detection–Identifica tion (TDI) score. We defined anosmia based on clinical definitions (Kobal et al., 2000). Specifically, anosmia was defined by a TDI score of 16 or less and normosmia, normal olfactory function, was defined by a TDI score of at least 31 (Kobal et al., 2000).

Sniﬃng paradigm

The paradigm for the fMRI experiment was presented in a block-design consisting of five sniffing blocks and five normal breathing blocks, with each block containing eight breathing cycles with a block duration of 32 s. During the active sniffing condition, each sniff was characterized by a short and deep intake of breath through the nose. During the baseline condition, subjects were instructed to breathe normally through the nose. Subjects were trained in how to perform this paradigm correctly before each scanning session. No odor was presented during the scanning sessions. For temporal standardization, the subject’s breathing cycles were guided by auditory stimuli, which were presented by speakers integrated in the scanner, throughout the whole experiment. For details on the presented auditory paradigm please see the supplementary materials.

Imaging acquisition

fMRI measurements were performed on a 3-Tesla Trio System (Siemens Medical Solution, Erlangen, Germany)

(3)

using a 32-channel head coil, and optimized, 2D single-shot, gradient-recalled, echo-planar imaging (EPI), which included online EPI distortion correction with point-spread function mapping (Zaitsev et al., 2004). Thirty-six slices (2.7 mm thickness, 0.5 mm gap) were acquired, with a ﬁeld of view (FOV) of 210210 mm and an echo time (TE)/repetition time (TR) of 32/2000 ms. The slices were aligned parallel to the AC–PC line.

Data preprocessing

fMRI data were preprocessed using SPM12b (http:// www.ﬁl.ion.ucl.ac.uk/spm/), implemented in MATLAB (Matlab 7.14.0, Release 2012a, Mathworks Inc., Sherborn, MA, USA), and included motion correction, spatial normalization to an MNI template, and spatial smoothing. Postprocessing included group ICA and FCA, using a seed-region approach.

ICA

In a first step, group ICA was performed to identify sniffing-induced functional networks. ICA is a data-driven methodology that can be used to separate data sets of multivariate signal characteristics into statistically independent components (ICs). Previous studies revealed that ICA is particularly appropriate for analyzing experiments using chemosensory stimulation (Scho¨pf et al., 2011; Frasnelli et al., 2012). Group ICA was per-formed conjointly for the complete study sample, using the Group ICA for fMRI Toolbox (GIFT; http://icatb.source-forge.net; (Calhoun et al., 2001)). The number of ICs was estimated using the minimum description length (MDL) cri-terion (Li et al., 2007), as implemented in GIFT as a default setting. ICA was performed using the Infomax algorithm (Bell and Sejnowski, 1997) after dimension reduction, using principal component analysis (PCA) in two consecu-tive reduction steps. The statistical reliability of estimated ICs was tested using the ICASSO algorithm (Himberg et al., 2004), implemented in GIFT. Using ICASSO, the IC estimation was calculated 20 times, varying the initial conditions of the algorithm, as well as the bootstrapped data sets. In a last step, differences between the two groups were calculated using a two-sample t-test (FWE-corrected,p< 0.05) with SPM12b.

FCA

To explore functional networks in anosmic patients in more detail, a whole-brain FCA was performed. Therefore, additional preprocessing steps were accomplished using the CONN toolbox ( Whitﬁeld-Gabrieli and Nieto-Castan˜on, 2012) (http://www.nitrc. org/projects/conn), implemented in MATLAB, including regression of nuisance parameters, which were extracted from the motion-correction process.

For analyzing functional connectivity between brain areas, the cerebral cortex was classified into 84 regions of interest (ROIs) based on Brodmann’s cytoarchitectonic map (Brodmann areas; BA) of the human brain (Brodmann, 1909; Craddock et al., 2013). As previous studies identified the cerebellum as an impor-tant region with respect to sniffing (Sobel et al., 1998b),

we additionally included the cerebellum into the FCA. For the division of the cerebellum into additional 28 ROIs, the probabilistic atlas SUIT (http://www.icn.ucl.ac.uk/ motorcontrol/imaging/suit.htm; (Diedrichsen et al., 2009) was used, resulting in a total of 112 ROIs in the FCA. Next, the CONN toolbox was used to calculate correla-tions of the mean time-courses between all ROIs at the single-subject level. These correlations were computed for the sniffing condition as well as for the rest condition. In a next step, a whole-brain FCA at the group level was performed to uncover sniffing-related differences in connections between brain areas. Functional connectivity was defined as the Z-transformed Pearson product-moment correlation coefficient between time-courses.

Z-scores were used to construct two connectivity matrices for each subject, representing functional connectivity during rest and task runs. For the group-level analysis, the Network-Based Statistics (NBS) method was used, which is utilized to tackle the multiple comparisons problem as described by Zalesky et al. (2010), and was implemented in the NBS Toolbox for Mat-lab R2012. This approach has the advantage of exploiting the internal connectional structure of brain graphs during the correction step for multiple comparisons, and poten-tially oﬀers a substantial gain in statistical power. Statistical inference with NBS is performed similarly to other, permutation-based tests for general linear models (GLM) (Winkler et al., 2014). In order to quantify the eﬀect of

anosmiaon the task-activated whole-brain functional

con-nectivity, we performed a general linear model-based anal-ysis, using a group (normal subjects vs. anosmic), two-level per-subject (two-way Mixed Effects ANOVA) design. In this repeated measurements GLM, one level repre-sented the functional connectivity at rest, while the other represented the functional connectivity during a task. Sub-ject age and mean frame-wise displacement during the scans were added as covariates to the model, as age was identified as a possible influencing factor in functional connectivity (Damoiseaux et al., 2008). The GLM analysis was run for each network edge, and the level of significance was calculated according to the description inZalesky et al. (2010), anda< 0.05 was accepted as significant.

Statistics

Statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS, Chicago, Illinois, USA), version 20.0. For the olfactory performance scores, mean and SD were calculated. To compare olfactory performance values between anosmic patients and healthy controls, the nonparametric Mann–Whitney test was performed due to the small sample size. Correlations of age and gender with olfactory performance were calculated using Pearson’s correlation. The alpha level for all statistical tests was set ata= 0.05.

RESULTS

Olfactory performance

Anosmic patients achieved signiﬁcantly lower scores in the olfactory performance measures compared to

(4)

healthy controls (for all subtests and the TDI score:

p< 0.001). Detailed results of olfactory performance are presented in Table 1. The two subject groups did not differ significantly in the distribution of gender (v2= 0.649, p= 0.420) or educational background (v2= 7.618, p= 0.079). Healthy controls were statistically significantly younger compared to anosmic patients (T= 3.017, p= 0.009). However, neither age (anosmic patients: r= 0.269, p= 0.423; controls:

r= 0.302, p= 0.294) nor gender (anosmic patients:

r= 0.231, p= 0.494; controls: r= 0.088, p= 0.766) was signiﬁcantly correlated with olfactory performance scores within the two subject groups.

fMRI results

In a first step, a group ICA was performed to test whether the sniffing paradigm activated the olfactory networks, as hypothesized based on the literature. Thus, in a second step, whole-brain FCA was performed to gain a deeper insight into the alterations of functional networks in patients with smell loss. In this study, only sniff-induced functional activation of the cerebral cortex was investigated.

Independent component analysis. Data from all

subjects were submitted to a combined group ICA estimation, resulting in 44 ICs. Group ICA revealed an

olfactory network, comparable to a network that has already been reported in healthy controls after stimulation with chemosensory compounds (Kollndorfer et al., 2015b). The relevant IC was identiﬁed by visual inspection, based on the spatial distribution of the activa-tion patterns. The network involved olfacactiva-tion-related areas such as the piriform and the entorhinal cortices, the amygdala, and the thalamus. When the networks between healthy controls and anosmic patients were compared, using a two-sample t-test (threshold:

T23= 6.886; p< 0.05, FWE-corrected), there were no

signiﬁcant diﬀerences in their spatial extent (seeFig. 1).

Functional connectivity analysis. The measures of the

whole-brain network graph, obtained from a two-way ANOVA, consist of nodes (ROIs) and edges (connections between ROIs). This setting allowed for direct comparisons between the two groups (anosmics and controls) and between active (sniffing) versus baseline conditions. Comparing the whole-brain functional connectivity of anosmic patients with healthy controls (controls > anosmics) uncovered a broad variety of additional connections in healthy controls during the sniffing conditions (see Table 2 and Fig. 2). The majority of the nodes with significantly higher functional connectivity incorporated brain areas known to be responsible for processing olfactory stimuli, including connections involving the anterior prefrontal cortex, the anterior cingulate cortex, the entorhinal cortex and the cerebellum. However, differences in functional connectivity were observed even beyond the common olfactory network, such as connections between the posterior cingulate cortex, the temporopolar area, and the fusiform gyrus, as well as connections from the fusiform gyrus to the primary motor and somatosensory areas.

In contrast, no additional functional connections were identified in anosmic patients compared to healthy controls (anosmics > controls) during the sniffing condition. Interestingly, decreased functional connectivity in anosmic patients was observed only during the sniffing condition, but not at baseline (see Fig. 3).

Fig. 1.Axial mean anatomical images overlaid with one relevant component, resulting from the combined group ICA (p< 0.05, FWE-corrected) for healthy controls. No diﬀerences in spatial extent between the two subject groups were detected (p> 0.05, FWE-corrected). The color bar representst-values. L = Left, R = Right. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.)

Table 1. Olfactory performance measures for healthy controls and anosmic patients Healthy controls Mean (SD) Anosmic patients Mean (SD) p-Value TDI-score 35.06 (1.95) 11.91 (2.10) <0.001 Odor threshold 8.61 (1.68) 1.55 (0.71) <0.001 Odor discrimination 13.07 (1.81) 5.82 (1.54) <0.001 Odor identiﬁcation 13.92 (1.44) 4.73 (2.05) <0.001 Disease duration (in years) 4.64 (3.36)

(5)

DISCUSSION

In the present study, a sniffing paradigm was used to investigate chemosensory network connectivity and spatial characteristics in patients with chronic peripheral smell loss. We showed that although the spatial extent of the olfactory network did not differ between patients and controls, the connectivity within this processing network was decreased for patients compared to healthy controls. This finding could be related to compensatory mechanisms by other eloquent sensory function in response to chronic peripheral loss.

Olfaction is the phylogenetically and ontogenetically oldest sense (Breiphol and Apfelbach, 1986) and stands out among all other human sensory systems, especially in the domain of central processing, in that peripheral input can be processed without a thalamic relay. Further-more, the olfactory network responsible for the sense of smell projects largely ipsilaterally within the brain (Lascano et al., 2010; Iannilli et al., 2013). The spatial organization of the olfactory network is much more dis-persed compared to other senses (for a review see Lundstro¨m et al. (2011)). Secondary and tertiary areas of olfactory processing involve parts of the limbic system, and are thus closely linked to memory and emotional states (Arshamian et al., 2013). Moreover, the olfactory system holds the unique ability to be activated by the sen-sorimotor act of sniﬃng, without the presentation of an odor (Sobel et al., 1998a). Due to this fact, the sense of smell is a powerful model for investigating the eﬀects of peripheral sensory loss on central processing.

Previous studies identified various demographic variables that potentially influence olfactory performance. A frequently discussed topic with respect to olfactory performance is gender effects. In many studies females outperformed males (for a review see Doty and Cameron (2009)). However, the general outper-formance by female over male subjects with regard to overall olfactory function has led to contradictory results (Hummel et al., 2007; Derntl et al., 2013). In this study,

gender was not significantly correlated with olfactory performance, neither in anosmic patients, nor in healthy controls. Another important factor influencing olfactory performance is age. Various studies (Choudhury et al., 2003; Hummel et al., 2007) have shown that age signifi-cantly influences olfactory performance, resulting in poorer performance in elderly subjects. In this study, age was not significantly correlated with olfactory perfor-mance measures within the subject groups. Therefore, we assume that the difference in age between anosmic patients and healthy controls did not significantly influ-ence the results of this study. However, age has also been reported to be an important variable in the investiga-tion into funcinvestiga-tional connectivity networks in the brain (Damoiseaux et al., 2008); therefore, age was included as a covariate in the FCA. Another important factor in the investigation into olfactory dysfunction is the duration of the dysfunction from the onset of the disorder, as atro-phy of olfaction-related brain regions increases with longer disease duration (Rombaux et al., 2009; Peng et al., 2013). However, in our study disease duration was not significantly correlated with olfactory performance measures.

Our study aimed to investigate alterations in functional networks in patients with peripheral sensory loss. Using group ICA, an olfactory network that covers brain areas known to be involved in processing odors (Kollndorfer et al., 2015b), such as the piriform cortex, the entorhinal cortex, and the thalamus, was detected in healthy con-trols, as well as in anosmic patients. This finding supports the hypothesis that the sensorimotor act of sniffing alone, without the presentation of an odor, activates an olfactory network (Sobel et al., 1998a). Furthermore, no significant differences in the spatial extent of the functional network between anosmic patients and healthy controls were detected. Thus, although anosmic patients are unable to perceive odors, the olfactory network can still be activated by the sensorimotor component of odor perception—sniff-ing. In a study bySobel et al. (1998a), nasal occlusion, with a reduced airstream and sniffing volume, led to less Fig. 2.Differential connectivity networks in healthy controls vs. anosmic patients (controls > anosmics). Graph edges represent the statistically different functional connections between the two groups after performing the NBS procedure (Zalesky et al., 2010). Additional connections were only found in healthy controls compared to anosmic patients for the sniffing condition. Node sizes are proportional to the number of statistically different edges, and their color code refers to the cerebral lobes. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

(6)

Table 2.Comparison of FC analysis between healthy controls and anosmics for the sniﬃng paradigm. Following connections were observed in healthy controls, but not in anosmic patients.

Node 1 Node 2 Test statistics

(F)

PrSomatosensory2 (R)

BA.2 (Right). Primary Somatosensory Cortex

Temporopolar (L)

BA.38 (Left). Temporopolar Area

23.5589

VPCingulate (R)

BA.23 (Right). Ventral Posterior Cingulate Cortex

Temporopolar (R)

BA.38 (Right). Temporopolar Area

22.5834

Fusiform (R)

BA.37 (Right). Fusiform gyrus

Somatosensory-assoc (L)

BA.5 (Left). Somatosensory Association Cortex

22.2107

PrSomatosensory3 (L)

BA.3 (Left). Primary Somatosensory Cortex

Fusiform (R)

21.5680

AEntorhinal (R)

BA.34 (Right). Anterior Entorhinal Cortex

20.3924

Insula (L)

BA.13 (Left). Insular Cortex

19.6696

Fusiform (R)

PrMotor (L)

BA.4 (Left). Primary Motor Cortex

18.9144

AntPrefrontal (L)

BA.10 (Left). Anterior Prefrontal Cortex

Insula (L)

17.9650

SVisual (R)

BA.18 (Right). Secondary Visual Cortex

AEntorhinal (R)

17.0066

DPCingulate (R)

BA.31 (Right). Dorsal Posterior Cingulate Cortex

AEntorhinal (R)

16.7367

Insula (L)

Somatosensory-assoc (R)

BA.5 (Right). Somatosensory Association Cortex

16.3898

Left_VIIb

Cerebellum (Left). Lobe VIIb

Left_IX

Cerebellum (Left). Lobe IX

16.2443

Fusiform (L)

BA.37 (Left). Fusiform gyrus

16.1656

VPCingulate (R)

Parahippocampal (L)

BA.36 (Left). Parahippocampal cortex

15.7012

Fusiform (L)

PrMotor (L)

15.6702

AEntorhinal (R)

15.2071

DPCingulate (R)

InfPrefrontal (R)

BA.47 (Right). Inferior Prefrontal Gyrus

15.1278

RsCingulate (R)

BA.29 (Right). Retrosplenial Cingulate Cortex

Supramarginal (R)

BA.40 (Right). Supramarginal Gyrus

14.8536

AsVisual (R)

BA.19 (Right). Associative Visual Cortex

Right_VI

Cerebellum (Right). Lobe VI

14.7796

VPCingulate (L)

BA.23 (Left). Ventral Posterior Cingulate Cortex

AEntorhinal (R)

14.5098

Insula (R)

BA.13 (Right). Insular Cortex

14.4507

Orbitofrontal (R)

BA.11 (Right). Orbitofrontal Cortex

(7)

Table 2(continued)

(F)

Cingulate (L)

BA.30 (Left). Cingulate Cortex

Subcentral (R)

BA.43 (Right). Subcentral Area

14.3925

BA.2 (Right). Primary Somatosensory Cortex

Fusiform (R)

14.0213

SVisual (R)

BA.17 (Right). Primary Visual Cortex

AEntorhinal (L)

BA.34 (Left). Anterior Entorhinal Cortex

13.8958

Temporopolar (L)

13.8839

PrVisual (R)

AEntorhinal (R)

13.5542

Orbitofrontal (R)

PrMotor (L)

13.5244

MidTemporal (R)

BA.21 (Right). Middle Temporal Gyrus

VPCingulate (R)

13.0230

Fusiform (L)

Left_VIIIa

Cerebellum (Left). Lobe VIIIa

12.9660

Temporopolar (L)

PrAuditory (L)

BA.41 (Left). Primary Auditory Cortex

12.9587

AntPrefrontal (R)

BA.10 (Right). Anterior Prefrontal Cortex

AEntorhinal (R)

12.9070

DPCingulate (L)

BA.31 (Left). Dorsal Posterior Cingulate Cortex

AEntorhinal (R)

12.7530

Temporopolar (L)

Right_I_IV

Cerebellum (Right). Lobe I-IV

12.6907

AEntorhinal (R)

Angular (L)

BA.39 (Left). Angular gyrus

12.2926

Orbitofrontal (R)

VACingulate (L)

BA.24 (Left). Ventral Anterior Cingulate Cortex

12.2824

PEntorhinal (R)

BA.28 (Right). Posterior Entorhinal Cortex

Right_IX

Cerebellum (Right). Lobe IX

12.2416

Cingulate (R)

BA.30 (Right). Cingulate Cortex

Vermis_VIIb

Cerebellum (Vermis). Lobe VIIb

12.1661

Left_X

Cerebellum (Left). Lobe X

Vermis_X

Cerebellum (Vermis). Lobe X

12.0338

Cingulate (R)

BA.30 (Right). Cingulate Cortex

Supramarginal (R)

11.9259

PrAuditory (L)

Premotor (L)

BA.6 (Left). Premotor Cortex

11.8057

VACingulate (R)

BA.24 (Right). Ventral Anterior Cingulate Cortex

11.8032

AEntorhinal (R)

Supramarginal (L)

BA.40 (Left). Supramarginal Gyrus

11.5690

PrVisual (R)

Right_VI

Cerebellum (Right). Lobe VI

11.5374

Insula (R)

BA.13 (Right). Insular Cortex

11.5357

VACingulate (L)

Supramarginal (R)

11.2865

Insula (L)

AsVisual (R)

11.2395

VPCingulate (R)

AEntorhinal (R)

11.1687

(8)

Table 2(continued)

(F)

Left_VIIIa

Cerebellum (Left). Lobe VIIIa

Left_IX

Cerebellum (Left). Lobe IX

11.0601

RsCingulate (R)

Left_CrusI

Cerebellum (Left). Lobe VIIa Crus I

11.0054

AntPrefrontal (R)

BA.10 (Right). Anterior Prefrontal Cortex

Insula (L)

10.9260

MidTemporal (R)

BA.21 (Right). Middle Temporal Gyrus

DPCingulate (R)

10.9171

Left_I_IV

Cerebellum (Left). Lobe I-IV

Left_VI

Cerebellum (Vermis). Lobe VI

10.5755

PrVisual (R)

AEntorhinal (L)

BA.34 (Left). Anterior Entorhinal Cortex

10.5067

RsCingulate (R)

Left_CrusII

Cerebellum (Left). Lobe VIIa Crus II

10.4804

Insula (L)

VPCingulate (L)

BA.23 (Left). Ventral Posterior Cingulate Cortex

10.4557

Temporopolar (L)

10.4193

DPCingulate (L)

BA.31 (Left). Dorsal Posterior Cingulate Cortex

IFC-tr (R)

BA.45 (Right). IFC pars triangularis

10.3955

SupTemporal (R)

BA.22 (Right). Superior Temporal Gyrus

10.2866

SVisual (R)

BA.18 (Right). Secondary Visual Cortex

Perirhinal (L)

BA.35 (Left). Perirhinal cortex

10.2796

Temporopolar (L)

10.2758

Right_V

Cerebellum (Right). Lobe V

Right_IX

Cerebellum (Right). Lobe IX

10.2480

Temporopolar (L)

10.2457

RsCingulate (R)

Supramarginal (L)

10.2322

VACingulate (L)

10.2235

PrAuditory (L)

Premotor (L)

BA.6 (Left). Premotor Cortex

10.1624

AsVisual (R)

Supramarginal (L)

10.1256

SupTemporal (L)

BA.22 (Left). Superior Temporal Gyrus

DLPFC (L)

BA.46 (Left). Dorsolateral Prefrontal Cortex

10.0677

InfPrefrontal (R)

10.0556

Orbitofrontal (R)

10.0324

VPCingulate (R)

InfPrefrontal (R)

(9)

functional activity during sniffing. This does not contradict our study, as a mechanical blockage and an intact olfac-tory airway would obviously cause differences in sniffing behavior, resulting in differing neural activation.

When investigating the functional characteristics we detected significant differences in functional connectivity during the performed sniffing paradigm. Functional connections that were more activated to a greater degree by the sniffing paradigm in healthy controls by the sniffing paradigm mainly involved areas responsible for processing chemosensory stimuli, such as the entorhinal cortex or the primary somatosensory cortex. However, increased functional connectivity in healthy controls was also detected in a broad variety of brain areas even beyond the olfaction-related areas, such as the supramarginal cortex or the premotor area. This finding indicates that the peripheral loss of sensory perception may cause reduced functional connectivity not only in the sensory-specific network, but also in global brain networks. Thus, sensory loss has wide implications in functional networks in the brain. One possible explanation for the altered functional connectivity induced by smell loss may be the close interaction between the olfactory pathways, the limbic system and memory function. Due to the high interconnection of the olfactory system with other functional networks, a loss of chemosensory input may induce far-reaching effects in the whole brain. However,

the detailed investigation into this ﬁnding should be part of future studies.

Interestingly, these alterations in functional connectivity were observed only for the sniﬃng condition, whereas no diﬀerences were found for the baseline condition. We therefore assume that a deprived olfactory system induced altered functional connectivity in anosmic patients, which was not limited to olfaction-related areas, but also involved a decrease in connections in the whole brain. A recently published investigation (Kollndorfer et al., 2015a), examining a part of the participants of this study, revealed that altered func-tional networks were observed during chemosensory stimulation. Reduced functional connections were not only observed in the olfactory network, but also in the somatosensory and the integrative networks.

In the last few decades, a broad variety of studies have demonstrated the brain’s capability to reorganize its function after traumatic injuries, stroke, or sensory loss (Bende and Nordin, 1997; Cadiou et al., 2014). Previous research determined that sensory loss often induces functional and structural modifications of the brain (Bola et al., 2014; Qin et al., 2014). Alterations of brain structure, as well as structural connectivity, have been identified in patients with smell loss. In addition, a decrease in gray and white matter has been detected in patients with smell loss (Bitter et al., 2010a). The decrease in brain volume was detected not only for olfac-tory areas, such as the piriform cortex, but also for the anterior cingulate cortex or the anterior insula. It is assumed that volume loss is induced by a lack of sensory input (Bitter et al., 2010a). The findings of our study are in accord with previous research that investigated structural alterations (Bitter et al., 2010b), where a decrease in gray matter in anosmic patients was observed. However, the brain is able not only to reorganize its functional connec-tions, but also to establish new functional and structural connections after severe damage in order to renew lost functions. The training of functions at the behavioral level has already been shown to cause the development of new functional (Vidyasagar et al., 2014) and structural connec-tions (Vahdat et al., 2014). Recently, this has also been shown for olfactory loss (Kollndorfer et al., 2014). In that study, we demonstrated that an olfactory training program (Hummel et al., 2009; Damm et al., 2014) is not only suc-cessful at the behavioral level, but also induces functional brain reorganizational processes. Similar to studies on hearing loss (Liu et al., 2015), resting-state fMRI may also be a promising tool with which to investigate functional networks in smell loss. However, basic research on olfac-tory networks at rest is needed to aid in the interpretation of spatial and temporal characteristics.

A potential limitation of this study is the small sample size of 11 anosmic patients. However, a very strict screening procedure was applied to obtain a homogeneous study sample. We included only patients with smell loss after an infection of the upper respiratory tract to avoid heterogeneity effects from the influence of different causes of smell loss. This is important since smell loss may be the first symptom of a neurodegenerative disorder, such as Alzheimer’s or Fig. 3.The figure represents network connectivity for task (x-axis)

and rest conditions (y-axis) for healthy controls (red dots) and anosmic patients (blue dots). Functional connectivity was defined as the Z-transformed correlation coefficients between time-courses. All values are presented as mean Z-transformed Pearson product-moment correlation coefficient scores of all ROIs during task and rest conditions. Task connectivity was higher in healthy controls com-pared to anosmic patients, as healthy controls achieve higher Z scores on thex-axis. In contrast, no group differences were obtained for the rest condition, as no differences occurred in the distribution at they-axis. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

(10)

Parkinson’s disease (Mesholam et al., 1998; Doty, 2012). Another limiting factor of the present work is the statisti-cally significant difference of age between the two subject groups. However, age was included as a covariate in all analyses, and we, therefore, assume that the differences in age did not influence the results of the present study.

CONCLUSION

In this study, we were able to show that patients with anosmia, after a sinunasal infection still have an intact functional olfactory network, which can be investigated and activated by a sniffing paradigm. However, the results of this study revealed a significant decrease in functional connectivity in anosmic patients. Although this reduced functional connectivity mainly affects olfaction-and sensory-related brain areas, a decrease in functional connections was observed even in global brain networks. Based on the findings of this study, we believe that peripheral sensory loss affects not only sensory-specific functional networks, but also induces far-reaching alterations in neural networks in the whole brain.

Acknowledgments and project funding—KK is supported by an FWF grant (P23205-B09) to VS. KK is also supported by the Austrian National Bank (15356). Special thanks are dedicated to Daniel W. Wesson and Johanna Reichert for a critical review of the paper and valuable comments and discussion throughout the creation of the manuscript. AJ is supported by the European Union FP7 Marie Curie IEF Research grant FABRIC – ‘‘Exploring the Formation and Adaptation of the Brain Connectome,”grant no. 2012-PIEF-GA-33003. We would also like to thank the reviewers for their anonymous support which significantly increased the quality of this manuscript.

REFERENCES

Arshamian A, Iannilli E, Gerber JC, Willander J, Persson J, Seo H-S, Hummel T, Larsson M (2013) The functional neuroanatomy of odor evoked autobiographical memories cued by odors and words. Neuropsychologia 51:123–131.

Barkai E, Saar D (2001) Cellular correlates of olfactory learning in the rat piriform cortex. Rev Neurosci 12:111–120.

Beck ATT, Steer RAA, Brown GKK (1996) Beck Depression Inventory–II (BDI–II). San Antonio, TX: Harcourt Assessment Inc. Bell AJ, Sejnowski TJ (1997) The ‘‘independent components” of

natural scenes are edge ﬁlters. Vision Res 37:3327–3338. Bende M, Nordin S (1997) Perceptual learning in olfaction

professional wine tasters versus controls. Physiol Behav 62:1065–1070.

Bitter T, Bru¨derle J, Gudziol H, Burmeister HP, Gaser C, Guntinas-Lichius O (2010a) Gray and white matter reduction in hyposmic subjects—a voxel-based morphometry study. Brain Res 1347:42–47.

Bitter T, Gudziol H, Burmeister HP, Mentzel H-J, Guntinas-Lichius O, Gaser C (2010b) Anosmia leads to a loss of gray matter in cortical brain areas. Chem Senses 35:407–415.

Bola M, Gall C, Moewes C, Fedorov A, Hinrichs H, Sabel BA (2014) Brain functional connectivity network breakdown and restoration in blindness. Neurology 83:542–551.

Breiphol W, Apfelbach R (1986). In: Breipohl W, Apfelbach R, editors. Ontogeny of Olfaction – Principles of Olfactory Maturation in Vertebrates. Berlin: Springer.

Brodmann K (1909) Beitra¨ge zur histologischen Lokalisation der Grosshirnrinde: VI. Die Cortexgliederung des Menschen. J fu¨r Psychol und Neurol 10:231–246.

Cadiou H, Aoude´ I, Tazir B, Molinas A, Fenech C, Meunier N, Grosmaitre X (2014) Postnatal odorant exposure induces peripheral olfactory plasticity at the cellular level. J Neurosci 34:4857–4870.

Calhoun VD, Adali T, Pearlson GD, Pekar JJ (2001) A method for making group inferences from functional MRI data using

independent component analysis. Hum Brain Mapp

14:140–151.

Choudhury ES, Moberg P, Doty RL (2003) Inﬂuences of age and sex on a microencapsulated odor memory test. Chem Senses 28:799–805.

Craddock RC, Jbabdi S, Yan C-G, Vogelstein JT, Castellanos FX, Di Martino A, Kelly C, Heberlein K, Colcombe S, Milham MP (2013) Imaging human connectomes at the macroscale. Nat Methods 10:524–539.

Croy I, Symmank A, Schellong J, Hummel C, Gerber J, Joraschky P, Hummel T (2014) Olfaction as a marker for depression in humans. J Aﬀect Disord 160:80–86.

Damm M, Pikart LK, Reimann H, Burkert S, Go¨ktas O, Haxel B, Frey S, Charalampakis I, Beule A, Renner B, Hummel T, Hu¨ttenbrink K-B (2014) Olfactory training is helpful in postinfectious olfactory loss: A randomized, controlled, multicenter study. Laryngoscope 124:826–831.

Damoiseaux JS, Beckmann CF, Arigita EJS, Barkhof F, Scheltens P, Stam CJ, Smith SM, Rombouts SARB (2008) Reduced resting-state brain activity in the ‘‘default network” in normal aging. Cereb Cortex 18:1856–1864.

Deems DA, Doty RL, Settle RG, Moore-Gillon V, Shaman P, Mester AF, Kimmelman CP, Brightman VJ, Snow JB (1991) Smell and taste disorders, a study of 750 patients from the University of Pennsylvania Smell and Taste Center. Arch Otolaryngol Head Neck Surg 117:519–528.

Derntl B, Scho¨pf V, Kollndorfer K, Lanzenberger R (2013) Menstrual cycle phase and duration of oral contraception intake aﬀect olfactory perception. Chem Senses 38:67–75.

Diedrichsen J, Balsters JH, Flavell J, Cussans E, Ramnani N (2009) A probabilistic MR atlas of the human cerebellum. Neuroimage 46:39–46.

Diers M, Christmann C, Koeppe C, Ruf M, Flor H (2010) Mirrored, imagined and executed movements diﬀerentially activate sensorimotor cortex in amputees with and without phantom limb pain. Pain 149:296–304.

Doty RL (2008) The olfactory vector hypothesis of neurodegenerative disease: is it viable? Ann Neurol 63:7–15.

Doty RL (2012) Olfaction in Parkinson’s disease and related disorders. Neurobiol Dis 46:527–552.

Doty RL, Cameron EL (2009) Sex diﬀerences and reproductive hormone inﬂuences on human odor perception. Physiol Behav 97:213–228.

Frasnelli J, Lundstro¨m JN, Scho¨pf V, Negoias S, Hummel T, Lepore F (2012) Dual processing streams in chemosensory perception. Front Hum Neurosci 6:288.

Gilbert CD, Sigman M (2007) Brain states: top-down inﬂuences in sensory processing. Neuron 54:677–696.

Goldstone RL (1998) Perceptual learning. Annu Rev Psychol 49:585–612.

Himberg J, Hyva¨rinen A, Esposito F (2004) Validating the independent components of neuroimaging time series via clustering and visualization. Neuroimage 22:1214–1222. Hubel DH, Wiesel TN (1970) The period of susceptibility to the

physiological eﬀects of unilateral eye closure in kittens. J Physiol 206:419–436.

Hummel C, Zucco GM, Iannilli E, Maboshe W, Landis BN, Hummel T (2012) OLAF: standardization of international olfactory tests. Eur Arch Oto-Rhino-Laryngol 269:871–880.

Hummel T, Kobal G, Gudziol H, Mackay-Sim A (2007) Normative data for the ‘‘Sniﬃn’ Sticks” including tests of odor identiﬁcation, odor discrimination, and olfactory thresholds: an upgrade based

(11)

on a group of more than 3,000 subjects. Eur Arch Otorhinolaryngol 264:237–243.

Hummel T, Rissom K, Reden J, Ha¨hner A, Weidenbecher M, Hu¨ttenbrink K-B (2009) Eﬀects of olfactory training in patients with olfactory loss. Laryngoscope 119:496–499.

Hummel T, Sekinger B, Wolf SR, Pauli E, Kobal G (1997) ‘‘Sniﬃn” sticks’: olfactory performance assessed by the combined testing of odor identiﬁcation, odor discrimination and olfactory threshold. Chem Senses 22:39–52.

Iannilli E, Wiens S, Arshamian A, Seo H-S (2013) A spatiotemporal comparison between olfactory and trigeminal event-related potentials. Neuroimage 77:254–261.

Kobal G, Hummel T, Sekinger B, Barz S, Roscher S, Wolf S (1996) ‘‘Sniﬃn’ sticks”: screening of olfactory performance. Rhinology 34:222–226.

Kobal G, Klimek L, Wolfensberger M, Gudziol H, Temmel A, Owen CM, Seeber H, Pauli E, Hummel T (2000) Multicenter investigation of 1,036 subjects using a standardized method for the assessment of olfactory function combining tests of odor identiﬁcation, odor discrimination, and olfactory thresholds. Eur Arch Oto-Rhino-Laryngol 257:205–211.

Kollndorfer K, Fischmeister FPhS, Kowalczyk K, Hoche E, Mueller CA, Trattnig S, Scho¨pf V (2015a) Olfactory training induces changes in regional functional connectivity in patients with long-term smell loss. NeuroImage: Clin 9:401–410.

Kollndorfer K, Kowalczyk K, Frasnelli J, Hoche E, Unger E, Mueller C, Trattnig S, Scho¨pf V (2015b) Same same but diﬀerent. Diﬀerent trigeminal chemoreceptors share the same central pathway. PLoS One 10:e0121091.

Kollndorfer K, Kowalczyk K, Hoche E, Mueller CA, Pollak M, Trattnig S, Scho¨pf V (2014) Recovery of olfactory function induces neuroplasticity eﬀects in patients with smell loss. Neural Plast. Article ID 140419.

Lascano AM, Hummel T, Lacroix J-S, Landis BN, Michel CM (2010) Spatio-temporal dynamics of olfactory processing in the human brain: an event-related source imaging study. Neuroscience 167:700–708.

Li W, Luxenberg E, Parrish T, Gottfried JA (2006) Learning to smell the roses: experience-dependent neural plasticity in human piriform and orbitofrontal cortices. Neuron 52:1097–1108. Li Y-O, Adali T, Calhoun VD (2007) Estimating the number of

independent components for functional magnetic resonance imaging data. Hum Brain Mapp 28:1251–1266.

Liu B, Feng Y, Yang M, Chen J-Y, Li J, Huang Z-C, Zhang L-L (2015) Functional connectivity in patients with sensorineural hearing loss using resting-state MRI. Am J Audiol.

Lundstro¨m JN, Boesveldt S, Albrecht J (2011) Central processing of the chemical senses: an overview. ACS Chem Neurosci 2:5–16. Mainland J, Sobel N (2006) The sniﬀ is part of the olfactory percept.

Chem Senses 31:181–196.

Mainland JD, Bremner EA, Young N, Johnson BN, Khan RM, Bensaﬁ M, Sobel N (2002) Olfactory plasticity: one nostril knows what the other learns. Nature 419:802.

Merabet LB, Pascual-Leone A (2010) Neural reorganization following sensory loss: the opportunity of change. Nat Rev Neurosci 11:44–52.

Mesholam RI, Moberg PJ, Mahr RN, Doty RL (1998) Olfaction in neurodegenerative disease: a meta-analysis of olfactory functioning in Alzheimer’s and Parkinson’s diseases. Arch Neurol 55:84–90.

Peng P, Gu H, Xiao W, Si LF, Wang JF, Wang SK, Zhai RY, Wei YX (2013) A voxel-based morphometry study of anosmic patients. Br J Radiol 86:20130207.

Pollatos O, Albrecht J, Kopietz R, Linn J, Schoepf V, Kleemann AM, Schreder T, Schandry R, Wiesmann M (2007) Reduced olfactory sensitivity in subjects with depressive symptoms. J Aﬀect Disord 102:101–108.

Qin W, Xuan Y, Liu Y, Jiang T, Yu C (2014) Functional Connectivity Density in Congenitally and Late Blind Subjects. Cereb Cortex. Rombaux P, Duprez T, Hummel T (2009) Olfactory bulb volume in the

clinical assessment of olfactory dysfunction. Rhinology 47:3–9. Scho¨pf V, Windischberger C, Robinson S, Kasess CH, Fischmeister

FP, Lanzenberger R, Albrecht J, Kleemann AM, Kopietz R, Wiesmann M, Moser E (2011) Model-free fMRI group analysis using FENICA. Neuroimage 55:185–193.

Sobel N, Prabhakaran V, Desmond JE, Glover GH, Goode RL, Sullivan EV, Gabrieli JD (1998a) Sniﬃng and smelling: separate subsystems in the human olfactory cortex. Nature 392:282–286. Sobel N, Prabhakaran V, Hartley CA, Desmond JE, Zhao Z, Glover

GH, Gabrieli JD, Sullivan EV (1998b) Odorant-induced and sniﬀ-induced activation in the cerebellum of the human. J Neurosci 18:8990–9001.

Vahdat S, Darainy M, Ostry DJ (2014) Structure of plasticity in human sensory and motor networks due to perceptual learning. J Neurosci 34:2451–2463.

Vidyasagar R, Folger SE, Parkes LM (2014) Re-wiring the brain: increased functional connectivity within primary somatosensory cortex following synchronous co-activation. Neuroimage 92:19–26.

Wesson DW, Carey RM, Verhagen JV, Wachowiak M (2008) Rapid encoding and perception of novel odors in the rat. PLoS Biol 6: e82.

Whitﬁeld-Gabrieli S, Nieto-Castan˜on A (2012) Conn: a functional connectivity toolbox for correlated and anticorrelated brain networks. Brain Connect 2:125–141.

Wilson DA, Best AR, Sullivan RM (2004) Plasticity in the olfactory system: lessons for the neurobiology of memory. Neuroscientist 10:513–524.

Winkler AM, Ridgway GR, Webster MA, Smith SM, Nichols TE (2014) Permutation inference for the general linear model. Neuroimage 92:381–397.

Wu KN, Tan BK, Howard JD, Conley DB, Gottfried JA (2012) Olfactory input is critical for sustaining odor quality codes in human orbitofrontal cortex. Nat Neurosci 15:1313–1319. Zaitsev M, Hennig J, Speck O (2004) Point spread function mapping

with parallel imaging techniques and high acceleration factors: fast, robust, and ﬂexible method for echo-planar imaging distortion correction. Magn Reson Med 52:1156–1166.

Zalesky A, Fornito A, Bullmore ET (2010) Network-based statistic: identifying diﬀerences in brain networks. Neuroimage 53:1197–1207.

(Accepted 18 September 2015) (Available online 28 September 2015)