Cortical morphology alterations in non-COVID-19 post-infectious olfactory dysfunction

Abstract

While COVID-19-associated olfactory dysfunction leads to gray matter (GM) structural alterations, it remains unclear whether these structural GM alterations are specific to COVID-19. To address this, we investigated GM changes in patients with post-infectious olfactory dysfunction (PIOD) who tested negative for SARS-CoV-2 infection. This study included 65 patients with PIOD who were SARS-CoV-2 negative and 65 healthy controls. Olfactory function was assessed using the Sniffin’ Sticks test and olfactory event-related potentials (OERPs). T1-weighted MRI scans were analyzed using three complementary approaches: region-based morphometry (RBM) to assess regional GM volume, voxel-based morphometry (VBM) to examine voxel-wise GM density differences between PIOD patients and controls, and surface-based morphometry (SBM) to investigate cortical thickness alterations. Compared to controls, patients with PIOD exhibited significant reductions in GM volume, voxel-based GM density, and cortical thickness. Regions demonstrating GM volume/thickness reductions across these three morphological analyses included the orbitofrontal cortex, insula, fusiform gyrus, superior frontal cortex, and hippocampus. Further region-of-interest (ROI) analyses revealed that reduced cortical thickness in the orbitofrontal cortex was negatively correlated with the duration of olfactory dysfunction and positively correlated with olfactory performance. Our findings indicate brain structural alterations also occurs in SARS-CoV-2-negative PIOD patients.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-32800-6.

Introduction

Olfactory dysfunction (OD) affects approximately 22.2% of the general population¹, with prevalence rising sharply to 47% during the COVID-19 pandemic². A frequent cause of OD is upper respiratory tract infections, leading to post-infectious olfactory dysfunction (PIOD)³. Indeed, beyond SARS-CoV-2 infection, OD remains one of the most common presentations in otolaryngology clinics following upper respiratory viral infections. Notably, PIOD exhibits distinct demographic and clinical features depending on etiology: non-COVID-19 PIOD patients are more likely to be female and older compared to those with COVID-19-related OD^4,5, while COVID-19 patients demonstrate a higher incidence of parosmia and better initial olfactory test results⁶. Furthermore, olfactory function appears to improve more frequently after treatment in non-COVID-19 PIOD patients than in their COVID-19 counterparts⁴.

The pathophysiology of PIOD is not fully elucidated but is thought to involve viral damage to the olfactory epithelium (OE) or direct impairment of the central olfactory processing pathways via the olfactory nerve⁷. Viral replication within the olfactory sustentacular cells and neuroepithelial remodeling of the OE have been well-documented^8,9. In contrast, pathological changes within the central olfactory pathways in PIOD remain less clear.

Strong evidence implicates olfactory bulb (OB) abnormalities in PIOD. MRI studies have observed transient OB oedema in COVID-19 patients and decreased OB volume in patients with persistent OD^10–14. Reduced OB volumes correlated with olfactory function were also demonstrated in PIOD patients prior to the COVID-19 pandemic^15,16. However, literature on gray matter (GM) volume alterations within the primary and secondary olfactory networks in PIOD remains scarce, partly due to the requirement for advanced medical imaging and post-processing techniques. Advances in high-resolution T1-weighted MRI and statistical software now enable detailed brain morphological assessment. GM volume can be quantified using region-based morphometry (RBM), which provides volumes for atlas-defined region of interests (ROIs)¹⁷. More recently, voxel-based morphometry (VBM) and surface-based morphometry (SBM) analyses have been employed to investigate the cortical morphology in OD patients¹⁸. While VBM measures GM volume, SBM provides additional indices of cortical structure, including thickness, surface area, and gyrification. Combining VBM and SBM offers a more comprehensive understanding of neurobiological processes¹⁹.

Recent studies report that COVID-19 patients with OD exhibit smaller GM volumes and thinner cortices in the olfactory-related areas (i.e. insular cortex, parahippocampal gyrus, orbitofrontal cortex, gyrus rectus, left amygdala, caudate, and putamen ) compared to COVID-19 patients with normal olfaction^20,21. A large-scale longitudinal MRI study²² from the UK Biobank comparing pre- and post-COVID-19 brain scans identified cortical atrophy specifically within the orbitofrontal and parahippocampal regions, alongside a greater overall reduction in brain size in COVID-19 cases versus controls. These findings suggest that structural brain changes may underlie the pathophysiology of post-COVID-19 OD^20–23. Nevertheless, this hypothesis remains untested in PIOD resulting from other viruses, especially given the established clinical differences between COVID-19-related and non-COVID-19-related OD.

Therefore, this study aims to investigate whether brain cortical morphological alterations-specifically GM volume and cortical thickness-are present in non-COVID-19 PIOD patients compared to healthy controls, utilizing a multi-morphometric approach combining VBM and SBM analyses. Furthermore, we performed correlations between cortical measurements and clinical presentation to elucidate the relationship between brain structural changes and OD severity.

Results

Demographic and clinical characteristics

This study analyzed 65 PIOD patients without SARS-CoV-2 infection confirmed via real-time reverse-transcriptase polymerase chain reaction (RT-PCR) and 65 controls. Demographic and clinical characteristics are listed in Table 1. Prior to the onset of PIOD, 52 patients (80%) had a documented history of H1N1 influenza. In the remaining 13 cases (20%), OD occurred following an upper respiratory tract infection of unidentified cause, though COVID-19 was ruled out by RT-PCR. The interval between OD onset and imaging (duration) ranged from 3 to 24 months. All patients had persistence of OD, as defined by the World Health Organization criteria for post-COVID-19 condition²⁴. No significant differences were observed between groups in age, sex or years of education (P > 0.05 for all). Cognitive screening (MMSE) confirmed no participants scored below the impairment threshold. Olfactory assessment revealed significantly worse performance in the PIOD group versus controls for both TDI scores (Mann-Whitney U, P < 0.001) and OERPs (P < 0.001 for all components). Volumetric MRI analysis showed no significant between-group differences in TIV (P = 0.09) and GM volume ( P = 0.06).

Table 1.

Demographic, olfactory performance, total intracranial volume, and gray matter volume of the participants.

Parameters	PIOD	Control	P uncorr
Number of participants	65	65	N.A
Age, years (mean ± s.d.(95% CI))	47.5 ± 12.0 (44.4–50.5)	43.4 ± 14.1 (39.9–46.9)	0.08
Sex (male/female)	28 (43.1%) / 37 (56.9%)	30 (46.2%) / 35 (53.8%)	0.29
Years of education (median (IQR))	16 (3)	16 (3)	0.74
MMSE	29.3 ± 0.9	29.3 ± 1.0	0.71
Duration, months (median (IQR))	12 (9)	N.A	N.A
TDI score (median (IQR))	14 (12)	35 (2)	< 0.001
OERPs scores
N1 waveform (mean ± s.d.)
Latency (ms)	497.1 ± 62.2	290.8 ± 23.2	< 0.001
Amplitude (µV)	− 4.9 ± 1.2	− 6.1 ± 1.1	< 0.001
P2 waveform (mean ± s.d.)
Latency (ms)	639.9 ± 76.3	502.2 ± 31.1	< 0.001
Amplitude (µV)	5.1 ± 1.0	9.0 ± 1.0	< 0.001
TIV, mL (mean ± s.d.(95% CI))	1432.1 ± 108.6 (1405.2–1459.0)	1468.2 ± 129.0 (1436.3−1500.2)	0.09
GM volume (mean ± s.d.(95% CI))	631.8 ± 42.2 (621.4−642.3)	649.6 ± 60.9 (634.6−664.7)	0.06

Statistical analyses employed nonparametric tests for variables violating assumptions of normal distribution within groups. Specifically, independent samples t- tests were used to compare age, MMSE scores, OERPs parameters, TIV and GM volume between groups. Group differences in sex distribution were assessed using Chi-square test; For years of education and TDI scores, which were not normally distributed, the Mann-Whitney U tests was applied.

PIOD, post infectious olfactory dysfunction; CI, confidence interval; N.A, not applicable; IQR, inter quartile range; MMSE, mini-mental state examination; OD, olfactory dysfunction; TDI, threshold, discrimination, and identification; OERPs, olfactory event-related potentials; TIV, total intracranial volume; GM, gray matter.

RBM analysis findings

While GM volume was lower in PIOD patients (631.8 ± 42.2 mL) than in controls (649.6 ± 60.9 mL), this difference did not reach statistical significance (P = 0.06). However, ROI analysis using the Neuromorphometrics atlas (CAT12 toolbox) revealed significant GM volume reduction in 28 regions in the PIOD group compared to controls (P < 0.05, FDR-corrected), indicating early-stage atrophy. Figure 1 visually contrasts these regions of reduced GM volume between groups. A summary of the significantly affected regions is provided in Table 2. Notably, no brain regions exhibited increased GM volume in PIOD patients relative to controls.

Fig. 1.

Region-based morphometry (RBM) reveals gray matter volume reduction. Brain regions exhibiting statistically significant volume reduction in PIOD patients compared to controls. Regions are defined using the Neuromorphometrics atlas within the CAT12 toolbox. Color coding reflects the significance level (P < 0.05, FDR corrected).

Table 2.

Differences in regional gray matter volume between PIOD and control groups.

ROI	PIOD group	Control group	P
R Ventral DC	1.01 ± 0.10 (0.98–1.03)	1.08 ± 0.14 (1.05–1.12)	0.008
R superior parietal lobule	8.58 ± 0.89 (8.36–8.80)	8.95 ± 0.99 (8.71–9.19)	0.019
R superior temporal gyrus	5.80 ± 0.47 (5.69–5.92)	6.05 ± 0.75 (5.87–6.24)	0.020
R supramarginal gyrus	6.44 ± 0.57 (6.30–6.58)	6.76 ± 0.76 (6.57–6.94)	0.025
R Pallidum	0.20 ± 0.07 (0.18–0.21)	0.24 ± 0.09 (0.21–0.26)	0.031
R parietal operculum	1.61 ± 0.22 (1.55–1.66)	1.69 ± 0.27 (1.63–1.76)	0.022
R fusiform gyrus	6.48 ± 0.61 (6.33–6.63)	6.77 ± 0.82 (6.57–6.97)	0.018
R inferior temporal gyrus	10.31 ± 0.92 (10.08–10.54)	10.73 ± 1.31 (10.41–11.06)	0.006
R middle temporal gyrus	12.37 ± 1.07 (12.11–12.64)	12.81 ± 1.29 (12.49–13.13)	0.020
R medial orbital gyrus	3.51 ± 0.39 (3.41–3.60)	3.67 ± 0.46 (3.56–3.78)	0.018
R gyrus rectus	1.80 ± 0.21 (1.75–1.85)	1.89 ± 0.22 (1.84–1.95)	0.022
R lingual gyrus	6.51 ± 0.73 (6.33–6.69)	6.83 ± 0.70 (6.66–7.01)	0.033
R frontal pole	3.14 ± 0.36 (3.05–3.23)	3.33 ± 0.58 (3.19–3.47)	0.021
R anterior insula	3.67 ± 0.37 (3.58–3.76)	3.80 ± 0.40 (3.70–3.90)	0.030
R temporal pole	8.38 ± 0.81 (8.18–8.58)	8.70 ± 1.08 (8.43–8.97)	0.028
R anterior orbital gyrus	1.54 ± 0.18 (1.49–1.58)	1.61 ± 0.23 (1.56–1.67)	0.018
L Ventral DC	1.01 ± 0.10 (0.99–1.03)	1.08 ± 0.13 (1.05–1.11)	0.010
L middle cingulate gyrus	3.61 ± 0.35 (3.53–3.70)	3.82 ± 0.49 (3.69–3.93)	0.009
L planum polare	1.66 ± 0.18 (1.61–1.70)	1.74 ± 0.23 (1.68–1.80)	0.022
L medial orbital gyrus	3.50 ± 0.39 (3.40–3.60)	3.71 ± 0.47 (3.59–3.83)	0.017
L inferior occipital gyrus	4.78 ± 0.56 (4.65–4.92)	5.05 ± 0.62 (4.90–5.21)	0.016
L inferior frontal gyrus	2.43 ± 0.30 (2.36–2.50)	2.57 ± 0.33 (2.48–2.65)	0.020
L superior temporal gyrus	5.51 ± 0.55 (5.37–5.64)	5.78 ± 0.72 (5.60–5.95)	0.018
L middle temporal gyrus	11.77 ± 1.06 (11.51–12.04)	12.17 ± 1.33 (11.84–12.50)	0.016
L anterior orbital gyrus	1.42 ± 0.18 (1.37–1.46)	1.49 ± 0.22 (1.44–1.55)	0.026
L central operculum	3.32 ± 0.35 (3.23–3.40)	3.42 ± 0.43 (3.32–3.53)	0.022
L superior frontal gyrus	5.44 ± 0.54 (5.31–5.58)	5.67 ± 0.76 (5.48–5.86)	0.031
L frontal pole	2.92 ± 0.31 (2.84–2.99)	3.05 ± 0.53 (2.92–3.18)	0.041

GM regions exhibiting a significant reduction in volume (P < 0.05, FDR-corrected) between PIOD group and controls are shown. Regions were defined using the neuromorphometrics atlas in CAT12. Volumes (mL) are presented as mean ± s.d.(range).

PIOD, post infectious olfactory dysfunction; ROI, regions of interest; R, Right; L, Left; GM, gray matter; FDR, false discovery rate.

VBM analysis findings

Whole-brain VBM analysis revealed 10 clusters (extent threshold = 50 voxels) of statistically significant GM volume reduction in PIOD group compared to controls (P < 0.05, FWE-corrected). Six clusters were localized to the right hemisphere (superior temporal cortex, fusiform gyrus, rectus, inferior temporal cortex, supramarginal gyrus, superior frontal cortex), while four clusters were identified in the left hemisphere (superior temporal cortex, inferior occipital cortex, superior frontal cortex, hippocampus). Figure 2 provides a graphical representation of these results. The MNI coordinates and peak T-statistics for each cluster are detailed in Table 3. No clusters showing significantly increased GM volume were observed in the PIOD group relative to controls.

Fig. 2.

Gray matter volume loss in PIOD patients. Whole-brain voxel-based morphometry (VBM) analysis revealed statistically significant gray matter volume reductions PIOD patients compared with controls. Clusters are color-coded by significance level (P < 0.05, FWE-corrected; cluster extent threshold = 50 voxels).

Table 3.

Clusters of gray matter voxels with significant volume reduction in the PIOD patients relative to healthy controls-whole brain VBM analysis. (P < 0.05, FWE-corrected; extent threshold = 50 voxels).

Cluster size (voxels)	Peak MNI coordinates (mm)			Hemisphere	Region	T-value	P-value
	x	y	z
2082	42	− 26	5	R	Superior temporal cortex	5.22	0.005
1216	36	− 44	− 12	R	Fusiform gyrus	6.48	< 0.001
487	6	52	− 26	R	Rectus	4.77	0.011
290	51	− 44	− 18	R	Inferior temporal cortex	4.97	0.005
231	52	− 26	33	R	Supramarginal gyrus	4.82	0.006
108	18	45	38	R	Superior frontal cortex	4.77	0.012
972	− 44	− 26	5	L	Superior temporal cortex	4.57	0.041
366	− 51	− 69	− 6	L	Inferior occipital cortex	4.82	0.007
294	− 20	58	4	L	Superior frontal cortex	4.69	0.030
126	− 18	− 23	− 10	L	Hippocampus	4.88	0.007

Regions of the brain were defined using the neuromorphometrics atlas.

PIOD, post infectious olfactory dysfunction; VBM, voxel-based morphometry; FWE, family-wise error; MNI, Montreal Neurological Institute; R, right; L, left.

SBM analysis findings

Whole-brain SBM analysis revealed 13 clusters of statistically significant cortical thinning in the PIOD group relative to controls (P < 0.05, FWE-corrected). Of these, eight clusters localized to the right hemisphere (including the insula, fusiform gyrus, lateral orbitofrontal cortex, precentral gyrus, middle temporal gyrus, calcarine cortex, inferior occipital gyrus, and superior frontal gyrus) and five clusters involved the left hemisphere (insula, lateral orbitofrontal cortex, rectus gyrus, middle temporal gyrus, superior temporal gyrus). Peak coordinates and statistical details for all clusters are provided in Table 4. Figure 3 displays the spatial distribution of these results rendered on the FSaverage surface. No regions of significant cortical thickening were observed in the PIOD group compared to controls.

Table 4.

Clusters brain cortex vertices with significant thickness reduction in the PIOD patients compared to healthy control subjects-whole brain SBM analysis. (P < 0.05, FWE-corrected; extent threshold = 50 voxels).

Cluster size (voxels)	Peak MNI coordinates (mm)			Hemisphere	Region	T-value	P-value
	x	y	z
765	40	− 31	12	R	Insula	5.54	< 0.001
219	28	− 69	− 11	R	Fusiform gyrus	5.39	< 0.001
144	33	35	− 9	R	Lateral orbitofrontal cortex	5.71	< 0.001
118	35	− 25	47	R	Pre-central gyrus	5.01	< 0.001
85	49	− 60	18	R	Middle temporal cortex	4.93	< 0.001
36	16	− 70	11	R	Calcarine cortex	4.86	< 0.001
32	44	− 75	− 7	R	Inferior occipital cortex	4.82	< 0.001
26	21	37	36	R	Superior frontal cortex	4.92	< 0.001
262	− 40	2	11	L	Insula	5.63	< 0.001
188	− 32	34	− 10	L	Lateral Orbitofrontal cortex	5.49	< 0.001
82	− 7	27	− 25	L	Rectus gyrus	5.34	< 0.001
82	− 45	− 64	12	L	Middle temporal cortex	5.02	< 0.001
78	− 42	− 30	10	L	Superior temporal cortex	4.96	< 0.001

Regions of the cerebral cortex were defined using the Desikan-Killiany atlas.

PIOD, post infectious olfactory dysfunction; SBM, surface-based morphometry; FWE, family-wise error; MNI: Montreal Neurological Institute; R, right; L, left.

Fig. 3.

Cortical thinning in PIOD patients. Surface-based morphometry (SBM) reveals significant difference in cortical thickness between PIOD patients and controls. Clusters are color-coded by significance level (P < 0.05, FWE-corrected; cluster extent threshold = 20 vertices) on the FsAverage surface. Views show lateral, ventral and medial cortical surfaces.

Correlations between olfactory function and regional GM volume/cortical thickness

In patients with PIOD, the multiple regression model revealed a significant positive association between TDI score and cortical thickness in the lateral orbitofrontal cortex, as defined by the Desikan-Killiany atlas (right hemisphere: t = 3.404, P = 0.001; left hemisphere: t = 3.305, P = 0.002; Supplementary Table S1), suggesting that better olfactory performance correlates with greater cortical thickness in this region. Interestingly, cortical thickness in the same areas also showed a significant negative correlation with OD duration (right hemisphere: t = − 5.002, P < 0.001; left hemisphere: t = − 3.792, P < 0.001; Supplementary Table S2), indicating that longer disease duration is associated with cortical thinning. The correlation analyses are presented in Fig. 4. The model revealed no significant association between TDI score or OD duration and the mean GM volume within the a priori defined ROIs (Supplementary Table S3).

Fig. 4.

Scatter plot showing correlations between regional cortical thickness and olfactory measurements in PIOD patients. Cortical thickness in the right (red arrow) and left (blue arrow) orbitofrontal cortex is positively correlated with olfactory performance but inversely correlated with the duration of olfactory dysfunction.

Discussion

In this study, we report clinical and structural MRI data from a cohort of PIOD patients who were SARS-CoV-2 negative. Using a multi-morphometric approach, we compared GM structural alterations between PIOD patients and control subjects, and examined their association with olfactory performance and the duration of OD. To our knowledge, this is the first study investigating structural brain changes in PIOD patients without SARS-CoV-2 infection. Our finding revealed significant GM loss and reduced cortical thickness in SARS-CoV-2 negative PIOD patients, and cortical thickness decrease was associated with lower olfactory performance and longer OD duration, suggesting that persistent OD from other viral infection is also associated with brain abnormalities.

oERPs are an electrophysiological technique used to assess the olfactory function and the integrity of the olfactory pathways²⁵. The waveforms generated by oERPs typically show a prominent negative peak (N1) followed by a positive peak (P2). Quantification of their amplitude and latency provides insights into brain activity: amplitude generally reflects the degree of neuronal activation, while latency corresponds to processing speed and the timing of activation²⁵. Our study reconfirmed that PIOD patients exhibit longer latency and lower amplitudes in both the N1 and P2 waves compared to controls²⁶. The oERPs revealed diminished response, indicating impaired olfactory processing. Altogether the structural alteration in specific brain regions may be indicate peripheral damage to the olfactory epithelium, resulting in reduced afferent input and subsequent cortical reorganization²⁷. As our sample was assessed at a late-post-acute stage of OD (mean 12 months), the observed changes also may reflect chronic neuroinflammation originating from olfactory neurons. Either or both of these mechanisms could contribute to persistent OD in PIOD patients.

The RBM analysis revealed statistically significant GM volume reduction in 28 regions in the PIOD group compared to the control group. These regions included the ventral DC, parietal lobule, temporal lobule, orbital gyrus, gyrus rectus, lingual gyrus and insula cortex. This represents the first evidence of GM reduction observed in PIOD patients without SARS-CoV-2 infection. Compared with previous studies of OD patients with SARS-CoV-2 infection^20,21, fewer regions exhibited GM volume reduction. This discrepancy likely arises from distinct pathological mechanisms, preferred infection sites, and host immune responses. SARS-CoV-2 attacks olfactory pathways more aggressively through multiple routes: direct invasion of olfactory sensory neurons and supporting cells; induction of neuroinflammatory processes that disrupt signaling; and systemic inflammation causing neurological symptoms²⁸. This may be why COVID-19 patients exhibit GM volume reduction in more brain regions.

To better understand the neurobiological processes underlying PIOD, this study employed both VBM and SBM to investigate structural brain changes¹⁹. Compared to controls, PIOD patients exhibited statistically significant reductions in GM density within multiple regions, including: the bilateral superior temporal cortex, bilateral superior frontal cortex, right inferior temporal cortex, right fusiform gyrus, right rectus gyrus, right supramarginal gyrus, left inferior occipital cortex, and left hippocampus. Cortical thinning was observed in the: bilateral insula, bilateral lateral orbitofrontal cortex, bilateral middle temporal cortex, right fusiform gyrus, right pre-central gyrus, right calcarine cortex, right inferior occipital cortex, right superior frontal cortex, left rectus gyrus, and left superior temporal cortex. Significant overlap existed between the regions identified by VBM and SBM analyses. These overlapping regions, which also exhibited GM volume reduction found by RBM analyses, are known to be associated with olfactory function impairment.

As a key component of the secondary olfactory cortex, the orbitofrontal cortex serves as the primary neocortical target of the primary olfactory cortex and contributes to olfactory perception²⁹. In addition, the interconnection between olfactory identification and spatial memory is thought to be attributed to the orbitofrontal cortex³⁰. A recent systematic review of structural brain changes in olfaction identified the gyrus rectus and orbitofrontal cortex as major atrophic regions in patients with acquired anosmia¹⁸. In this study, orbitofrontal cortex thickness was positively associated with TDI scores and negatively correlated with OD duration. This finding aligns with reports on post-traumatic³¹ and post-SARS-CoV-2 ²¹ infection anosmia, suggesting a critical role for the orbitofrontal cortex in olfactory performance and odor memory. Multiple neuroimaging modalities investigating COVID-19-related OD consistently identify the orbitofrontal cortex as a key site of impairment. In patients with COVID-19-anosmia, neurochemical analyses³² reveal significantly reduced levels of N-acetyl aspartate, choline, and creatine in this region. Similarly, MRI-based glymphatic system³³ assessments show an increased free water fraction in the orbitofrontal cortex among affected individuals. This may explain why orbitofrontal cortex thickness was reduced in PIOD patients.

Our findings also revealed GM volume/thickness reductions in other olfactory areas, including the insula, fusiform gyrus, superior frontal cortex, and hippocampus. The insula contributes to odor quality coding and plays a crucial role in accurately assessing sensory information quality and discrimination learning³⁴. Additionally, studies using other modalities have corroborated impairments in the insula in individuals with OD. For instance, a diffusion-weighted MRI study³⁵ revealed reduced white matter fiber density in this region, while another reported increased cerebral blood flow³⁶. Previous evidence suggests these regions—the insula, fusiform gyrus, superior frontal cortex, and hippocampus—may play specific roles in olfactory recognition memory^37,38. Thus, GM volume/thickness reductions in these areas may be associated with impaired odor memory.

We also observed GM volume/thickness reductions in regions not classically associated with olfactory processing, such as the superior temporal cortex, supramarginal gyrus, inferior occipital cortex, precentral gyrus, and calcarine cortex in PIOD patients. This finding aligns with prior studies^20,21,31 indicating that, similar to other senses, olfactory abilities rely on both sensory processing and cognitive functions, engaging multiple GM regions beyond traditional olfactory networks³⁹. The study also found that GM alterations affect both hemispheres, with inconsistent laterality patterns. This may be reflected that both hemispheres were involved but with differential contributions, and that the alterations occur at a different pace in PIOD patients¹⁸ .

The pathophysiological mechanisms of PIOD have been better elucidated since the onset of the COVID−19 pandemic. Sustentacular cells have been identified as the principal targets of SARS-CoV−2 infection, whereas olfactory sensory neurons appear to be less directly affected⁹. In contrast, H1N1 influenza primarily damages the olfactory epithelium, and H5N1 demonstrates a propensity for rapid dissemination to the central nervous system⁴⁰. Although parainfluenza virus is not typically associated with brain invasion, it has been shown to infect olfactory sensory neurons and surrounding cells within the olfactory epithelium and bulb, which may contribute to impaired olfactory function⁴¹.

Neuroanatomical changes in patients with PIOD have rarely been studied in relation to the duration of OD. Postma et al.⁴² suggest that the correlation between the structure alteration of olfaction-related brain areas and OD duration may not be directly mediated by the length of OD itself, as olfactory function declines over time. Thus, the impact of OD duration brain morphology should be interpreted with caution.

The following limitations must be considered for interpretation of the present findings. Firstly, This cross-sectional study is limited by the absence of baseline (pre-infection) imaging data, which restricts our ability to determine whether GM structural alterations in patients with PIOD are progressive or modifiable with early intervention. The generalizability of these findings is limited by the single-center design, the referral-based population, and the exclusive focus on post-H1N1, which may cause selection bias. Secondly, although the two groups showed no significant differences in major demographics, years of education, and MMSE scores, we did not identify subtle cognitive differences between PIOD patients and controls subjects. As reported by Douaud et al.²² reported, a subtle pattern of lower cognitive abilities was observed in the participants who developed SARS-CoV-2 infection. It remains unclear whether this patter of reduced cognitive abilities also exists in PIOD patients without SARS-CoV-2 infection. In addition, to assess whether microstructural alterations in gray matter are specific to the infecting virus type, an additional control group comprising PIOD patients with SARS-CoV-2 infection would be required. By the way, all participants denied a history of COVID-19, and serological testing was not performed. Nevertheless, the likelihood of a prior asymptomatic infection is considered very low, owing to the weekly RT-PCR screening implemented for all participants since the start of the pandemic. Future studies should also include a COVID-PIOD comparison group and investigate the alteration of entire olfactory pathway using multimodal structural and functional imaging.

In conclusion, this multi-morphometric MRI analysis provides clear evidence of significant brain structural alterations in PIOD patients without SARS-CoV-2 infection, though less pronounced and widespread than that in PIOD patients with SARS-CoV-2 infection reported in previous studies^20,21. The affected regions, which include insula, fusiform gyrus, superior frontal cortex, and hippocampus, are critical for olfactory recognition memory. The severity of orbitofrontal cortex thinning correlated with olfactory function. This work contributes to a deeper understanding of PIOD pathophysiology.

Materials and methods

Participants

Sixty-five patients with PIOD and 65 sex- and age- matched healthy controls were enrolled in this study. All participants underwent COVID-19 testing, standard otorhinolaryngological evaluations, brain MRI scans, detailed medical history assessments, and psychophysical olfactory function tests at the Olfactory & Taste Disorder Diagnosis and Treatment Center of Beijing Anzhen Hospital, Capital Medical University, between August 2020 and October 2022. Inclusion criteria for PIOD patients: (1) Negative SARS-CoV-2 test via RT-PCR of nasopharyngeal specimens and no history of COVID-19; (2) ≥ 3 months of persistent OD following upper respiratory tract infections; (3) Patent olfactory cleft confirmed by endoscopic examination; (4) OD objectively verified by psychophysical olfactory testing; (5)Age range of 18–70 years, right-handedness (Edinburgh Handedness Inventory), and Mini-Mental State Examination (MMSE) score ≥ 27. Exclusion criteria for PIOD patients: (1) Etiologies confounding OD diagnosis (e.g., neruodegenerative diseases, sinonasal diseases, head trauma); (2) Structural brain lesions on MRI or poor scan quality; (3) History of olfactory impairment. Based on medical history reviews, there were no documented cases of tobacco, alcohol or substance use disorders among the participants. This study was approved by the Ethics Committee of Beijing Anzhen Hospital [approval no. 2021082X] and written informed consent was obtained from all participants. The guidelines of Helsinki Declaration were followed. All methods were performed in accordance with the approved guidelines.

Olfactory assessment

Psychophysical olfactory function was evaluated in all participants using Sniffin’ Sticks (SS) test (Burghart Gmbh, Wedel, Germany). This standardized assessment comprises three components: odor detection threshold (T), odor discrimination (D), and a four-alternative forced-choice odor identification task (I). The composite TDI score (sum of T, D, and I subscores) was calculated for each participant. Consistent with established normative data, hyposmia was defined as a TDI score < 30.75⁴³.

Olfactory event-related potentials (oERPs) were recorded to objectively assess olfactory function. Following established methodology²⁶, oERPs were elicited using phenethyl alcohol delivered via an OM6b olfactometer (Burghart, Germany). Signals were amplified, filtered, and recorded as stable waveforms. Analysis was performed using the EPE evaluation program (Kobal, Erlangen, Germany). oERPs were considered present if the average waveform exhibited a characteristic negative-positive complex with peak amplitudes exceeding ± 2µV. Recorded parameters included: (1) Latency and amplitude of the N1 component (most prominent negative peak); (2) Latency and amplitude of the P2 component (most prominent positive peak).

MRI acquisition

MRI scans were performed with a 3.0 Tesla scanner (Magnetom Trio Tim system, Siemens, Erlangen, Germany) using a standard 12-channel head coil. The high-resolution T1-weighted images covering the whole brain were acquired with a three-dimensional magnetization-prepared rapid gradient-echo (3D-MP RAGE) sequence (repetition time:1900 ms, echo time: 2.52 ms, inversion time:1100ms, flip angle:9°, field of view:250 mm 250 mm, 176 contiguous sagittal slices; voxel size:1 × 1 × 1mm³ ) .

Image processing

Regional GM volume differences between patients with PIOD and controls were assessed using RBM and VBM. SBM was employed to evaluate cortical thickness. All image preprocessing was performed using the CAT12 toolbox (Computational Anatomy Toolbox12.9, r2577, http://www.neuro.uni-jena.de/cat/) implemented in SPM12 (Statistical Parametric Mapping 12, v7771, http://www.fil.ion.ucl.ac.uk/spm) and executed in Matlab R2024b (The Mathworks, Natick, MA, USA), following the CAT12 manual¹⁷. Preprocessing included five steps. (1) Tissue segmentation: high-resolution T1-weighted images were segmented into GM, white matter (WM) and cerebrospinal fluid (CSF). While the tissue segmentation in CAT12 is prior-independent, the procedure was initialized using SPM’s standard Tissue Probability Maps. (2) Spatial Normalization: segmented GM images were spatially normalized to the Montreal Neurological Institute (MNI) space using Geodesic Shooting approach with predefined templates in CAT12. (3) Cortical surface reconstruction (for SBM): cortical thickness was estimated using the projection-based thickness (PBT) method. Central cortical surface for both hemisphere were reconstructed, incorporating topology correction, spherical inflation and spherical registration to the Freesurfer “FsAverage” template in CAT12. (4) Spatial smoothing: Normalized GM images were smoothed with an isotropic Gaussian kernel to reduce inter-subject anatomical variability and improve data normality. Smoothing was performed with kernels of 8 mm and 15 mm full-width at half-maximum for VBM and SBM analyses, respectively. The kernel sizes were chosen to optimize the detection of potential group differences in cortical volume and thickness. (5) Quality control: preprocessed images underwent quality assurance; only image quality exceeding 80% was included in subsequent analyses. Total intracranial volume (TIV), GM volume, WM volume, and CSF volume were calculated for each subject during CAT12 preprocessing. Mean GM volume and mean cortical thickness values were then extracted for each region of interest (ROI) using CAT12’s ROI tools.

Statistical analysis

Statistical analyses were performed using SPSS 22 (IBM, Ehningen, Germany). Data normality was assessed using the Shapiro-Wilk test. Normally distributed data were expressed as mean ± standard deviation and were compared using independent-sample t tests; Non-normally distributed data were reported as the median value [inter quartile range] and were analyzed using Mann-Whitney U tests. Group differences in sex distribution were assessed using Chi-square test. A multiple regression model was used to assess the association between olfactory function and brain morphometry, adjusting for age, disease duration, and sex, with region GM volume/cortical thickness as the dependent variable. Statistical significance was defined as P < 0.05 for all tests.

Voxel-wise comparisons of GM volume (using both RBM and VBM) and cortical thickness (using SBM) between PIOD patients and control groups were conducted using independent two-sample t-test within CAT12 toolbox. For RBM and VBM analyses, TIV, sex and age were included as covariates; for SBM analysis, sex and age were included as covariates. Multiple comparisons were corrected as follows: (1) RBM results were corrected using false discovery rate (FDR; P < 0.05) ; (2) VBM and SBM results were thresholded at the peak level using family-wise error (FWE) correction (P < 0.05) with a cluster extent threshold of > 50 contiguous voxels, applied across the whole brain.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization, L.Y. and X.Y.; methodology, J.L.and Y.G.; software, Z.S. and Y.Z.; validation, X.Y.; writing—original draft preparation, L.Y.; writing—review and editing, X.Y.; funding acquisition, X.Y. All authors have read and agreed to the published version of the manuscript.

Funding

The study was supported by a grant from the Research Foundation of Capital Institute of Pediatrics (grant no. JHYJ-2025-02).

Data availability

The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.