Translate this page into:
Magnetic resonance imaging in COVID-19-associated acute invasive fungal rhinosinusitis – Diagnosis and beyond
How to cite this article: Senapathy G, Putta T, Sistla SK. Magnetic resonance imaging in COVID-19-associated acute invasive fungal rhinosinusitis – Diagnosis and beyond. J Clin Imaging Sci 2023;13:23.
The aim of the study was to evaluate the magnetic resonance imaging (MRI) features of acute invasive fungal rhinosinusitis (AIFRS) at presentation and on follow-up imaging when patients receive treatment with systemic antifungal therapy and surgical debridement.
Material and Methods:
This is a retrospective analysis of imaging data from a cohort of patients diagnosed with AIFRS during the second wave of COVID-19 in single tertiary referral hospital in South India between March 2021 and May 2021 (n = 68). Final diagnosis was made using a composite reference standard which included a combination of MRI findings, clinical presentation, nasal endoscopy and intraoperative findings, and laboratory proof of invasive fungal infection. Analysis included 62 patients with “Definite AIFRS” findings on MRI and another six patients with “Possible AIFRS” findings on MRI and laboratory proof of invasive fungal infection. Follow-up imaging was available in 41 patients.
The most frequent MRI finding was T2 hypointensity in the sinonasal mucosa (94%) followed by mucosal necrosis/loss of contrast-enhancement (92.6%). Extrasinosal inflammation with or without necrosis in the pre-antral fat, retroantral fat, pterygopalatine fossa, and masticator space was seen in 91.1% of the cases. Extrasinosal spread was identified on MRI even when the computed tomography (CT) showed intact bone with normal extrasinosal density. Orbital involvement (72%) was in the form of contiguous spread from either the ethmoid or maxillary sinuses; the most frequent presentation being orbital cellulitis and necrosis, with some cases showing extension to the orbital apex (41%) and inflammation of the optic nerve (32%). A total of 22 patients showed involvement of the cavernous sinuses out of which 10 had sinus thrombosis and five patients had cavernous internal carotid artery involvement. Intracranial extension was seen both in the form of contiguous spread to the pachymeninges over the frontal and temporal lobes (25%) and intra-axial involvement in the form of cerebritis, abscesses, and infarcts (8.8%). Areas of blooming on SWI were noted within the areas of cerebritis and infarcts. Perineural spread of inflammation was seen along the mandibular nerves across foramen ovale in five patients and from the cisternal segment of trigeminal nerve to the root exit zone in pons in three patients. During follow-up, patients with disease progression showed involvement of the bones of skull base, osteomyelitis of the palate, alveolar process of maxilla, and zygoma. Persistent hyperenhancement in the post-operative bed after surgical debridement and resection was noted even in patients with stable disease.
Contrast-enhanced MRI must be performed in all patients with suspected AIFRS as non-contrast MRI fails to demonstrate tissue necrosis and CT fails to demonstrate extrasinosal disease across intact bony walls. Orbital apex, pterygopalatine fossa, and the cavernous sinuses form important pathways for disease spread to the skull base and intracranial compartment. While cerebritis, intracranial abscesses, and infarcts can be seen early in the disease due to the angioinvasive nature, perineural spread and skull base infiltration are seen 3–4 weeks after disease onset. Exaggerated soft-tissue enhancement in the post-operative bed after debridement can be a normal finding and must not be interpreted as disease progression.
The baseline incidence of acute invasive fungal rhinosinusitis (AIFRS) is low and often limited to patients with uncontrolled diabetes and immunocompromised state. The second wave of the COVID-19 pandemic in India, between March and July 2021, saw an unprecedented surge in acute invasive rhino-orbital fungal sinusitis, which prompted the government of India to declare COVID-19-associated mucormycosis (CAM) a notifiable disease under Epidemic Diseases Act in May 2021.
This unfortunate situation presented us with the opportunity to evaluate the magnetic resonance imaging (MRI) features of AIFRS at presentation and on follow-up imaging when patients received treatment with systemic antifungal therapy and surgical debridement. AIFRS is characterized by rapid onset of illness (<4 weeks) with vascular invasion and thrombosis, resulting in infarction and tissue necrosis with a very aggressive clinical course. The early angioinvasive nature of the disease, which accounts for the rapid spread across intact bony walls, is often not demonstrated on computed tomography (CT) as bone destruction is a late finding.[3-5] Contrast-enhanced MRI (CE-MRI) in patients with a high index of clinical suspicion helps detect the early and specific signs of AIFRS, such as mucosal necrosis and perisinosal fat stranding before orbital and intracranial spread, thereby prompting early initiation of treatment.[3-8]
MATERIAL AND METHODS
This is a retrospective analysis of imaging data from a cohort of patients diagnosed with AIFRS in a single tertiary referral hospital as part of CAM. A total number of 68 patients were diagnosed with AIFRS between March 2021 and May 2021 in our center with inhouse MRI.
A composite reference standard was used to arrive at a diagnosis of AIFRS, which included in the study:
Definitive AIFRS on imaging plus one or more of the following:
Clinical presentation highly suggestive of AIFRS
Classic findings of invasive fungal rhinosinusitis on diagnostic nasal endoscopy
Intraoperative appearance of invasive fungal infection and
Laboratory proof of invasive fungal infection.
Possible AIFRS on imaging plus laboratory proof of invasive fungal infection with or without one of the following:
Clinical presentation highly suggestive of AIFRS
Classic findings of invasive fungal rhinosinusitis on diagnostic nasal endoscopy and
Intraoperative appearance of invasive fungal infection.
All the patients had MRI of the orbits, paranasal sinuses (PNSs), and brain. Sixty-five out of the 68 had CE-MRI; three had non-contrast MRI in view of deranged renal function parameters.
Imaging follow-up was available between 1 and 13 months of disease onset in 41 out of the 68 patients.
All the patients underwent MRI on Philips MR Systems Ingenia S, 1.5 Tesla, Release 5.7. MRI brain with additional dedicated sequences for the orbits and PNS were performed. The contrast used was MultiHance (gadobenate dimeglumine, 529 mg/mL) administered with a pressure injector at a dose of 10 mL IV at 2 mL/s flow rate, followed by 15 mL saline chase at 2 mL/s flow rate. In patients where intracranial major vessel involvement was suspected, 3D Time of flight magnetic resonance angiography of the intracranial vessels was additionally obtained.
Magnetic resonance sequences are listed in [Table 1].
|DWI Whole Brain Axial, b-1000, NEX-1, ST 5 mm|
|Whole Brain FLAIR, NEX-1, ST 5 mm|
|SWIp–FAST Axial, ST 2 mm|
|Whole brain axial T1W, ST 5 mm|
|Coronal and axial small FOV T2W_TSE
for PNS and Orbits, ST 3.5 mm
|Coronal and axial small FOV STIR_long
TE for PNS and Orbits, ST 3.6 mm
|Axial small FOV T1_TSE for PNS and Orbits, ST 3.9 mm|
|Oblique sagittal T1W both orbits, ST 3 mm|
|Small FOV axial 3D_T1W_mDIXON for
PNS and Orbits, ST 2 mm
|MR angiogram 3D TOF, ST 1. 4 m|
|Post-contrast axial T1_3D_TFE_FS, ST 1.2 mm|
|Post-contrast small FOV T1 SE Axial, ST 3.5 mm|
|Post-contrast small FOV T1 SE Coronal, Sagittal, ST 5 mm|
All the cases were read independently by two radiologists (SG and TP), both having an experience of more than 10 years in cross sectional imaging, and the final interpretation was based on consensus.
The MR images were evaluated for the presence or absence of findings related to AIFRS:
Inflammation/infiltration was defined as T2 and short tau inversion recovery (STIR) hyperintense signal in the soft tissues with or without enhancement
PNS and nasal cavity: The nasal turbinates, nasal septum, and PNS were assessed for the presence of mucosal thickening, their T2 signal intensity, and necrosis.
Extrasinosal spread in the form of inflammation ± necrosis was documented in the following areas:
Pre-antral fat, retroantral fat, pterygopalatine fossa, masticator space, temporal fossa, bony walls of the PNS, other bones including alveolus, palate, zygoma, and the skull base.
Orbit: cellulitis was documented when there was fat stranding in the orbital fat ± abscess, increased bulk with T2 hyperintense signal, and exaggerated enhancement in the extraocular muscles.
Optic nerve: Optic nerve thickening, T2/ fluid attenuated inversion recovery (FLAIR)/ diffusion-weighted imaging hyperintense signal, or exaggerated enhancement along the nerve were marked as inflammation. Marked diffusion restriction with or without necrosis was considered as optic nerve ischemia when associated with complete loss of vision. Orbital apices involvement was also separately documented.
Cavernous sinuses: The presence of thickening, T2 hyperintensity, or enhancement along the lining dura was documented as involvement; thrombosis was defined as non-enhancement within the sinus cavity.
Major intracranial vessels were assessed for the presence of vessel wall thickening and enhancement (documented as vasculitis) with or without thrombosis.
Intracranial extra-axial involvement: Pachymeninges were assessed for both contiguous and non-contiguous thickening and enhancement.
Intracranial intra-axial involvement in the form of cerebritis, abscess, and infarcts was noted. The susceptibility weighted imaging (SWI) sequences were assessed for the presence of blooming.
Perineural spread of disease: Abnormal T2/STIR hyperintense signal along the nerve ± enhancement.
The cases were categorized as “Possible AIFRS” or “Definite AIFRS” based on CE-MRI findings as follows:
Definitive AIFRS on CE-MRI = At least one of the two features was unequivocally present: mucosal necrosis and extrasinosal involvement. T2 hypointense signal in the mucosa of the sinus and/or turbinates may or may not be seen.
Possible AIFRS on CE-MRI = Definite T2 hypointense signal in the mucosa of the sinus and/or turbinates+ No or doubtful mucosal necrosis + No or doubtful extrasinosal involvement.
The radiologists were blinded to the clinical presentation, nasal endoscopy findings, intraoperative findings, and laboratory investigations at the time, the MRI scans were reviewed. Laboratory proof of causative organism (direct microscopy, histopathology with special stains, fungal culture, and molecular diagnostics), when available was documented by the other study participant.
In patients who had limited resection of the turbinates or sinuses before they presented to our center for MRI, the mucosa in the post-operative bed was assessed only for necrosis while the other areas were assessed for both inflammation and necrosis. Patients who underwent more extensive resection before presentation at our center were excluded from our study.
A total of 68 patients were included in our study, all of whom had either concurrent COVID-19 pneumonia or were diagnosed with COVID-19 infection within 45 days before presentation. [Table 2] summarizes the demographics, comorbidity, and imaging profile of the study patients and [Table 3] lists the causative fungal organism.
|Median age (range), in years||52 (24–79)|
|Detected at presentation||24|
|History not available||5|
|Concurrent or recent COVID-19 infection||68|
|History of steroid use for COVID-19 infection|
|History not available||35|
|Other immunocompromised conditions (post-transplant)||2|
|Magnetic resonance imaging|
|With contrast (CE-MRI)||65|
|Follow-up/Outcome at June 2022#|
|Alive; Regression on imaging||14|
|Alive; Stableresidual disease on imaging||4|
|Alive; Progression on imaging||19|
|Lost to follow-up||27|
|Pathogen||No. of patients|
|No fungal growth/fungal elements detecteda||8|
|No in-house laboratory tests performed||5|
The most common clinical presentation was facial pain or numbness in the pre-maxillary region of a few days’ duration, unilateral to begin with, and often associated with bloody nasal discharge. Many of these progressed to have retro-orbital pain and proptosis in a few days.
Sixty-five out of the 68 patients underwent CE-MRI as per institutional protocol, while three underwent non-contrast MRI in view of deranged renal function parameters.
Imaging follow-up was available between 1 and 13 months from disease onset in 41 out of 68 patients.
Definitive AIFRS findings were found in 62/68 cases. The other six cases were categorized as “Possible AIFRS” based on MRI findings; all these cases had laboratory proof of invasive fungal infection.
PNS involvement on MRI
Most cases showed involvement of more than one sinus, with the most frequent combination being maxillary + ethmoid sinuses [Table 4]. The earliest and most frequent MRI finding was T2 hypointense signal in the thickened PNS mucosa (94%); the second most frequent finding was necrosis (92.6%) [Figure 1 and Table 5].
|Sinuses involveda||Number (%)|
|Parameter||Number of patients (%)|
|T2 hypointense signal in the thickened sinonasal mucosa||64 (94.1)|
|Sinonasal necrosis||63 (92.6)|
|Turbinates||37 (60.6) (n=61)a|
|Nasal septum||11 (17.7) (n=62)b|
|Paranasal sinus wall||58 (89.2) (n=65)c|
|Extrasinosal soft-tissue involvement|
|Pre and/or retroantral soft tissue||62 (91.1)|
|Pterygopalatine fossa||56 (82.3)|
|Masticator space||41 (60.3)|
|Orbital apex involvement|
|Optic nerve involvement|
|Cavernous sinus involvement|
|Sinuses inflamed (n=136)||14 (10.3)|
|Sinuses thrombosed (n=136)||10 (7.3)|
|Extra-axial intracranial involvement|
|(Pachymeningeal thickening and enhancement)||17 (25.0)|
|Intra-axial intracranial involvement|
|Major vascular involvement|
|(thrombosis, vasculitis)||6 (8.8)|
Extrasinosal involvement on MRI [Table 5]
The pre-antral and retroantral fat involvement was mostly in the form of fat stranding, followed by obliteration of the fat with inflammatory and/or necrotic tissue. Pre-antral abscess in the form of central necrotic collection with diffusion restriction and peripheral rim enhancement was seen in six out of 68 patients [Figure 2]. Retroantral fat involvement was often associated with extension into the pterygopalatine fossa [Figures 2-4]
Orbital cellulitis (72%) was seen more commonly in the extraconal and in some cases, also in the intraconal compartment. The medial and inferior recti were the most frequently involved extraocular muscles, in continuity with the adjacent PNS [Figures 5-7]. Orbital necrosis was seen in six patients [Figures 5 and 8].
Optic nerve involvement, when present, was always seen along with orbital cellulitis. Four patients had optic nerve ischemia out of which two were frank necrosis of the nerve seen contiguous with other orbital soft-tissue necrosis [Figures 5 and 8].
Conical deformity of the globe with anteroposterior elongation and posterior tenting, indicating the presence of tension orbit, was seen in three patients [Figure 5]. Two patients had globe destruction with deformity [Figure 5].
All the patients with cavernous sinus involvement had either orbital apex, posterior ethmoid, or pterygopalatine fossa involvement [Figures 6, 8 and 9]. Five of these patients showed wall thickening and enhancement of the cavernous internal carotid artery (ICA) [Figure 6]; out of which three had ICA thrombosis [Figure 8]. One of the patients with pre-clival extension of necrotic tissue had basilar artery thrombosis.
Involvement of the bones of skull base was seen in 30% of the patients [Figure 8].
Pachymeningeal thickening and enhancement were seen initially in contiguity with either the involved ethmoid sinuses across the cribriform plate, orbital roof, frontal bones or the pterygopalatine fossa, and later progressing to diffuse pachymeningeal enhancement in some of the cases [Figures 8 and 9].
A total of nine out of 68 patients developed cerebritis of which seven were contiguous with the involved pachymeninges and two were seen in distant parenchyma [Figures 6 and 9]. In all these cases, the involved brain parenchyma showed T2, FLAIR hyperintense signal ± patchy enhancement; six of these cases also showed areas of blooming [Figure 9].
All the five patients with cavernous ICA involvement and the one patient with basilar artery involvement showed cerebral infarcts [Figure 9]. Two other patients with cavernous sinus involvement developed ICA thrombosis and infarcts on follow-up imaging.
Five patients had cerebral abscess, out of which one was at presentation to the hospital [Figure 8]. The other four had cerebritis at presentation which progressed to abscess on follow-up imaging, despite initiation of treatment [Figure 9].
Perineural spread of inflammation along the mandibular nerves traversing across the foramen ovale was seen in five patients [Figure 4]; one patient had retrograde spread from the left optic nerve to optic tract [Figure 8]. Three other patients had retrograde spread of inflammation along the trigeminal nerve in the cisternal segment with infarcts or cerebritis in Pons at the root exit zone [Figure 6].
Few of the patients with extensive maxillary sinus and masticator space involvement also developed osteomyelitis of the palate (n = 7), the alveolar process of maxilla (n = 8), and zygoma (n = 13) during follow-up [Figure 7].
Some extent of surgical resection was seen in all the patients with follow-up studies, ranging from limited resection in the form of uncinectomy, turbinectomy, and ethmoidectomy to the more extensive forms such as orbital exenteration and maxillectomy. Most of these patients continued to show increased soft-tissue enhancement in the post-operative bed lasting for a few weeks after debridement [Figure 10]. Eight patients had follow-up imaging available between 8 and 13 months after disease onset and two of these patients succumbed due to disease progression with intracranial abscesses and infarcts. The remaining were clinically free from the disease but had come for reconstructive facial surgeries due to the deformities cause by resections and erosions during active disease. Varying degrees of osteonecrosis in the craniofacial skeleton were noted in these patients on CT done for planning the reconstructive surgeries [Figure 10].
Fungal rhinosinusitis is broadly divided into invasive and non-invasive and the invasive type is further classified into three subtypes: Acute invasive, chronic invasive, and chronic granulomatous invasive forms.[11-13] Invasive fungal sinusitis is defined by the presence of fungal hyphae within the mucosa, submucosa, bone, or blood vessels of the PNSs. AIFRS occurs in immunocompromised patients, predominantly in those belonging to one of the following two categories: One group is patients with diabetic ketoacidosis or uncontrolled diabetes. The other is those with an immunocompromised state with neutropenia due to causes such as chemotherapy, bone marrow transplant, steroid therapy, or immunosuppressive therapy for organ transplant.[3,11-14]
Secondary infections have been documented in patients affected with COVID-19 infection.[15,16] While a majority of these have been bacterial and fungal pulmonary infections, the COVID-19 pandemic also saw an unprecedented incidence of AIFRS cases in epidemic proportions.[1,9,15-19]
Although there are no findings specific to COVID-19-related AIFRS as opposed to non COVID-19 related AIFRS, this unfortunate situation presented us with the opportunity to document the temporal evolution of the CE-MRI findings in a large number of patients in a short time span. It also helped us identify the pattern of early and delayed craniofacial and intracranial complications in patients on treatment. The high index of clinical suspicion in this situation had also prompted early CE-MRI in many patients, which had also enabled us to document the very early and subtle changes on MR and we believe that this data can in future help in the early diagnosis of AIFRS in the vulnerable population much before orbital and intracranial spread.
The infection often arises in the nasal cavity, spreading to the PNS.[1,3,14] Spread to the orbits, pterygopalatine fossa, and intracranially is through direct invasion as well as by hematogenous spread in the form of vascular invasion and thrombosis, which accounts for the rapid and fulminant course of the disease.
Plain CT scan of the PNSs initially shows non-specific hypodense mucosal thickening but, due to spread across intact bony walls, fails to demonstrate the early and specific signs of AIFRS such as mucosal necrosis and extrasinosal fat stranding [Figure 4].[3-8] Hence, all our patients referred for imaging with a high index of clinical suspicion of AIFRS based on history and clinical examination were subjected to CE-MRI rather than CT, in line with our institutional protocol.
T2 hypointense mucosal signal in the affected PNSs and turbinates is due to the T2 shortening effect of the fungal elements. Although T1 hyperintensity ± T2 hypointensity of the sinus contents is also a feature of fungal sinusitis, it may also be seen with chronic and non-invasive forms of fungal sinusitis and when there are blood products within the sinus cavity. Hence, this finding was deliberately not given weightage in our study.
Orbital involvement was seen as an extension from ethmoid sinusitis more often than maxillary sinusitis, possibly because of the porous lamina papyracea facilitating extrasinosal spread into the orbit. All the patients with optic nerve necrosis, extensive orbital necrosis, or globe destruction underwent orbital exenteration.
Extension of the disease into the pterygopalatine fossa is by direct invasion from the maxillary sinus and through the inferior orbital fissure from the orbit, forming a pathway for further spread to masticator space, temporal fossa, and the cavernous sinuses, as seen in our study. Some of these cases demonstrated frank necrosis in the form of sheet like non enhancing devitalized tissue, sometimes extending to the bones of the skull base and the adjacent pachymeninges. Apart from the bony walls of the sinus cavities, involvement of the other facial bones such as the hard palate, alveolar process of the maxilla, and zygoma was seen beyond 3–4 weeks of onset.
Meningitis and cerebral abscess are the most commonly described central nervous system (CNS) manifestations of invasive fungal infections. Vascular complications such as vasculitis, mycotic aneurysm formation, cerebral hemorrhage, and ischemic infarction are the distant complications resulting from the hematogenous spread of fungal CNS infections due to their angioinvasive features. In our study, cerebral infarcts/cerebritis were often seen at non-contiguous distant sites and showed blooming. Due to direct angioinvasive nature, the infarcts can harbor fungal elements. This, along with vasculitis, may explain blooming on SWI sequences. The ophthalmic and ethmoid arteries are the common vascular channels for intracranial spread, apart from the ICA when the cavernous sinus is involved. A few of these infarcts/fungal cerebritis lesions progressed to abscess formation on follow-up imaging despite treatment.
Perineural spread of fungus through direct infiltration from the cavernous sinus to the brainstem has been described in literature.[22,23] This was seen as trigeminal nerve involvement in the cisternal segment and its root exit zone in Pons in our study in the form of T2/FLAIR hyperintensity and restricted diffusion. Involvement of the branches of trigeminal nerve in foramen ovale and foramen spinosum was a more common finding; one patient had retrograde spread along the left optic nerve to the optic tract; these were mostly observed after 3–4 weeks of disease onset on follow-up imaging in patients with disease progression.
On follow-up, most of the patients continued to show increased soft-tissue enhancement in the post-operative bed for a few weeks after debridement, even when there was unequivocal clinical improvement. We therefore believe that this finding in isolation cannot be considered as persistent disease. On follow-up imaging, presence of active or progressive disease may be identified when there is inflammation or necrosis at a new site.
CE-MRI must be performed in all cases with suspected acute angioinvasive fungal sinusitis as CT fails to demonstrate early extra sinosal disease spread across intact bony walls and non-contrast study fails to demonstrate tissue necrosis.
Orbital apex, pterygopalatine fossa, and the cavernous sinuses form important pathways for disease spread to the skull base and intracranial compartment.
While development of intracranial abscesses, infarcts, and cerebritis can be seen early in the disease due to angioinvasive nature, development of perineural spread and skull base infiltrations are seen 3–4 weeks after disease onset.
Exaggerated soft-tissue enhancement in the post-operative bed lasting a few weeks after debridement can be a normal finding and must not be interpreted as disease progression unless there is interval development of necrosis or inflammation at a new site.
Declaration of patient consent
Patient’s consent not required as patient’s identity is not disclosed or compromised.
Conflicts of interest
There are no conflicts of interest.
Financial support and sponsorship
- Perineural spread of rhinocerebral mucormycosis. AJNR Am J Neuroradiol. 1996;17:114-6.
- [Google Scholar]