Keywords
lack of contrast enhancement - MRI - diffusion restriction - fungal load - coagulative
necrosis
Introduction
As the second wave of coronavirus disease 2019 (COVID-19) wreaked havoc in India beginning
in April 2021, a potentially devastating fungal infection began to occur in its wake.
While patients started to recover from COVID-19 pneumonia, a drastic surge was seen
in the incidence of acute invasive fungal rhinosinusitis (also known as mucormycosis)
among such patients contributing to added morbidity and mortality. The rapid rise
in the incidence of COVID-associated mucormycosis (CAM) led to it being declared as
a notifiable disease in many parts of India. Similar outbreaks of saprophytic fungal
infections have been recorded in the past following natural disasters in India and
other countries, the Indian Ocean tsunami in 2004 being noteworthy.[1]
The sudden deluge in the cases of CAM has been attributed to preexisting uncontrolled
diabetes or diabetes worsened by COVID-19, indiscriminate use of steroids in the management
of COVID-19, oxygen therapy, and high serum ferritin levels.[2] Clinical diagnosis is difficult as symptoms often overlap with those of nonspecific
bacterial and viral rhinosinusitis. Imaging plays a crucial role in early diagnosis
of this aggressive disease so that prompt treatment is initiated to reduce the associated
morbidity and mortality.
Magnetic resonance imaging (MRI) has been considered a sensitive modality to detect
mucormycosis; however, overall MRI appearance of the disease entity and its correlation
with histopathological findings and disease outcome is poorly understood. The aim
of our study is to assess the signal intensity on T1- (T1WI) and T2-weighted images
(T2WI), enhancement pattern and diffusion characteristics in CAM and to correlate
these with histopathological findings and disease outcome.
Materials and Methods
The institutional ethical committee of Maulana Azad Medical College approved this
prospective study vide IEC number (F.1/IEC/MAMC/(84/02/2021/No.396). Twenty-eight
consecutive patients with clinical suspicion of CAM underwent MRI in the institution
between May and July 2021. As per the institutional policy, MRI was done only for
patients who were reverse transcription polymerase chain reaction (RT-PCR) negative
for COVID at the time of presentation. Of the 28 patients, 3 had undergone surgical
debridement/functional endoscopic sinus surgery prior to imaging and were not included
in the study. Three of the remaining 25 patients could not be administered gadolinium-based
intravenous contrast due to impaired renal function (estimated glomerular filtration
rate < 30 mL/min/1.73m2) and were excluded. Two patients did not have any features of mucormycosis on MRI
and their nasal endoscopic findings were also normal. These patients were not taken
up for debridement and were also excluded. The remaining 20 patients formed the final
study group. All 20 patients underwent debridement within 48 to 72 hours after MRI
and the diagnosis was confirmed on histopathology. They were followed up at 6 weeks
and again at 10 weeks from the diagnosis, and outcome was recorded in terms of duration
of hospital stay (more than 6 weeks from confirmation of CAM) and patient survival.
Imaging Technique
MR examination was done on a 3T MR system (MagentomSkyra; Siemens, Erlangen, Germany)
in supine position using a dedicated 32-channel phased array head coil. A standardized
imaging protocol was followed for all patients and included the following sequences
in multiple planes:
-
Spin-echo T1WIs (TR: 450–634 ms, TE: 7.1–9.4 ms) with an intersection gap of 1 to
2 mm, slice thickness of 3 mm, and matrix of 256 × 256.
-
T2-weighted turbo spin-echo images (TR: 3200–5900 ms, TE: 75–103 ms) with an intersection
gap of 1 to 2 mm, slice thickness of 3 mm, and matrix of 256 × 256.
-
Fat-suppressed T2-weighted fast spin-echo images (TR: 4430–6610 ms, TE: 79 ms) with
an intersection gap of 1 to 2 mm, slice thickness of 3 mm, and matrix of 256 × 256.
-
Contrast-enhanced fat-suppressed T1WIs (TR: 450–1010 ms, TE: 7.1–11 ms) with an intersection
gap of 1 to 2 mm, slice thickness of 3 mm, and matrix of 256 × 256 after intravenous
administration of gadopentetate dimeglumine (Magnascan, Unijules Life Sciences Ltd,
India) at a dose of 0.1 to 0.2 mmol/kg.
-
Diffusion-weighted images were obtained using multishot echo planar RESOLVE (readout
segmentation of long variable echo-trains) diffusion-weighted sequence in the axial
plane at b values of 0 and 1000 s/mm2 followed by generation of apparent diffusion coefficient (ADC) maps.
Analysis of Clinical and Imaging Data
Clinical data for patients was recorded with special emphasis on duration between
diagnosis of COVID-19 and CAM and underlying risk factors including diabetes mellitus,
steroid use, and oxygen therapy. Images were analyzed by two radiologists with 20
and 12 years of experience in head and neck radiology. Differences in opinion were
resolved by consensus.
-
Predominant signal characteristics of the involved region were analyzed on T1WI and
T2WI and were categorized as hypointense, isointense, and hyperintense relative to
cerebral gray matter.
-
On postcontrast images, predominant enhancement pattern of the involved sinonasal
region was classified as homogeneous, heterogeneous, or lack of contrast enhancement
(LOC). LOC was defined as a nonenhancing lesion of minimum size 5 × 5 mm on contrast-enhanced
T1-weighted fat-suppressed images with preservation of intervening bone/fat. When
there was no LOC in the lesion, the lesion was categorized into homogeneous or heterogeneous
enhancement pattern.
-
Presence or absence of diffusion restriction in the region of the predominant enhancement
pattern was noted. Lesions having a bright signal on diffusion-weighted images (b value = 1000 s/mm2) and low signal on ADC maps were considered to have restricted diffusion.
Histopathological Correlation
Histopathological specimens were analyzed by a pathologist with 25 years of experience
for the presence of mucosal, osseous, vascular, and neural invasion, presence of coagulative
necrosis, granulomatous inflammatory reaction, and vascular granulation tissue. Type
of fungal organism seen and extent of fungal hyphae were also noted, the latter being
categorized as +1, +2, and +3 (+1: present in 1–2 fields, +2: visible easily, +3:
large in number).
Statistical Analysis
The collected data was entered in MS Excel and analysis was done using SPSS version
21.0. Qualitative data was presented as percentages for categorical variables. Statistical
differences between the categorical variables for two comparable groups was measured
using chi-square test, while for more than two groups, Fischer's exact test was employed.
Predictive accuracy of each independent factor was measured using sensitivity, specificity,
positive predictive value (PPV), negative predictive value, and accuracy. A p-value of less than 0.05 was considered significant.
Results
A total of 20 patients were included in the study of which 13 were men and 7 were
women. The age of the subjects ranged from 35 to 76 years, with a mean age of 51.5
years. In 6 patients, bilateral involvement was seen and all but 2 patients showed
extrasinonasal involvement.
Out of the 20 patients, 15 had RT-PCR-confirmed severe acute respiratory syndrome
COVID-19 infection within the 4 weeks prior to testing positive for CAM. In the remaining
5 patients, there was a high degree of clinical suspicion for recent COVID-19 in view
of history of typical symptoms (such as sore throat, cough, anosmia) or a prior chest
radiograph showing bilateral ground-glass infiltrates. Prominent risk factors for
CAM among patients included in the study group were diabetes (found in 16 of the 20
study subjects, i.e., 80%), history of oxygen support in the form of either invasive
or noninvasive ventilation (in all subjects), and treatment with systemic corticosteroids
for COVID-19 (in 10 patients, i.e., 50%). The most common steroid regimes followed
were intravenous methylprednisolone 0.5 to 1 mg/kg or intravenous dexamethasone 0.1
to 0.2 mg/kg (moderate COVID) and intravenous methylprednisolone 1 to 2 mg/kg in two
divided doses or intravenous dexamethasone 0.2 to 0.4 mg/kg (severe COVID) usually
for a duration of 5 to 10 days.[2]
On T2-weighted MR images, signal pattern was found to be variable with lesions appearing
hyperintense (45%), hypointense (30%), and isointense (25%) ([Table 1]). Out of all 20 cases of CAM included in our study, 90% were hypointense on T1WIs
and only 2 patients showed iso- to hyperintense signal.
Table 1
Correlation of enhancement pattern with other imaging variables, histopathological
findings, and patient outcome
|
Variables
|
LOC, n = 13 (%)
|
Non-LOC, n = 7 (%)
|
p-Value
|
|
Diffusion restriction
|
|
Present
|
13 (100)
|
5 (71.4)
|
0.110
|
|
Absent
|
0 (0)
|
2 (28.6)
|
|
T2 signal intensity
|
|
Hyperintense
|
5 (38.5)
|
4 (57.1)
|
0.523
|
|
Hypointense
|
5 (38.5)
|
1 (14.3)
|
|
Isointense
|
3 (23)
|
2 (28.6)
|
|
Coagulative necrosis
|
|
Present
|
12 (92.3)
|
2 (28.6)
|
0.007[a]
|
|
Absent
|
1 (7.7)
|
5 (71.4)
|
|
Granulomatous reaction
|
|
Present
|
7 (53.8)
|
5 (71.4)
|
0.392
|
|
Absent
|
6 (46.2)
|
2 (28.6)
|
|
Extent of fungus
|
|
+ 1/+ 2
|
3
|
5 (71.4)
|
0.047[a]
|
|
+ 3
|
10
|
2 (28.6)
|
|
Outcome – Duration of hospital stay
|
|
< 6 wk
|
4 (30.8)
|
7 (100)
|
0.004[a]
|
|
> 6 wk
|
9 (69.2)
|
0 (0)
|
|
Outcome – Survival
|
|
Survived
|
10 (76.9)
|
7 (100)
|
0.251
|
|
Deceased
|
3 (23.1)
|
0
|
Abbreviation: LOC, lack of contrast enhancement.
a Statistically significant.
The enhancement patterns on T1 fat-saturated postcontrast images are summarized in
[Fig. 1]. The most common pattern of enhancement in the diseased area was found to be LOC,
seen in 13 cases (65%). The most common sites depicting LOC were the maxillary and
ethmoid sinuses, each being involved in 8 out of 13 patients (61.5%).
Fig. 1 Magnetic resonance (MR) enhancement patterns in coronavirus disease (COVID)-associated
mucormycosis (CAM).
As shown in [Table 2], out of the 20 cases analyzed, 18 showed diffusion restriction (90%). In addition,
LOC was 100% predictive of restricted diffusion.
Table 2
Correlation of diffusion restriction with histopathological findings and patient outcome
|
Variables
|
Diffusion restriction +
n = 18 (%)
|
Diffusion restriction –
n = 2 (%)
|
p-Value
|
|
Coagulative necrosis
|
|
Present
|
13 (72.2)
|
1 (50)
|
0.521
|
|
Absent
|
5 (27.8)
|
1 (50)
|
|
Granulomatous reaction
|
|
Present
|
10 (55.6)
|
1 (50)
|
0.710
|
|
Absent
|
8 (44.4)
|
1 (50)
|
|
Extent of fungus
|
|
+ 1
|
0 (0)
|
1 (50)
|
0.007[a]
|
|
+ 2
|
7 (38.9)
|
0 (0)
|
|
+ 3
|
11 (61.1)
|
1 (50)
|
|
Outcome – Duration of hospital stay
|
|
< 6 wk
|
9 (50)
|
2 (100)
|
0.289
|
|
> 6 wk
|
9 (50)
|
0 (0)
|
|
Outcome – Survival
|
|
Survived
|
15 (83.3)
|
2 (100)
|
0.716
|
|
Deceased
|
3 (16.6)
|
0 (0)
|
a Statistically significant.
Of the 13 patients showing LOC, 12 showed presence of coagulative necrosis on histopathology
(92.3%), whereas only one each of the homogeneously and heterogeneously enhancing
lesions showed coagulative necrosis. This was found to be statistically significant
(p-value = 0.007), as coagulative necrosis was not demonstrable in 71.4% of lesions
showing enhancement patterns other than LOC ([Fig. 2]).
Fig. 2 Coagulative necrosis in lack of contrast enhancement (LOC) and non-LOC lesions.
No significant difference in the presence of granulomatous reaction was seen between
lesions showing LOC, homogeneous, or heterogeneous enhancement. Six cases showed simultaneous
presence of coagulative necrosis and granulomatous reaction, all of which showed LOC
pattern of enhancement.
Of the 13 patients, 10 with LOC as the predominant enhancement pattern had +3 score
for extent of fungal organism (76.9%), and presence of higher fungal load in lesions
with LOC was statistically significant (p-value = 0.047). All 18 patients with restricted diffusion had a histopathological
fungal score of +2 (7/18, 38.9%) or +3 (11/18, 61.1%), whereas none of the restricting
lesions had a score of +1. This was also found to be statistically significant (p-value = 0.007).
Coagulative necrosis was seen in 13 of the 18 patients showing diffusion restriction
([Table 2]); however, no statistically significant relationship could be derived.
All 9 patients with prolonged hospital stay (> 6 weeks) showed LOC as the predominant
enhancement pattern and 3 of these patients expired during follow-up. Patients with
homogeneously or heterogeneously enhancing lesions had a hospital stay shorter than
6 weeks. No deaths were recorded in this group. The relationship between predominant
enhancement pattern and duration of hospital stay was found to be statistically significant
with p-value = 0.004.
Nine of the 18 patients with presence of restricted diffusion had a short duration
of hospital stay and an equal number had prolonged stay. However, only 2 of the 20
patients showed absence of restricted diffusion and both had a short duration of hospital
admission.
Discussion
Mucormycosis is an acute invasive fungal infection seen in immunocompromised individuals,
caused by saprophytic organisms belonging to the class Zygomycetes and genera Rhizopus,
Rhizomucor, Absidia, and Mucor.[3] Although many different sites of involvement have been seen, rhino-orbital-cerebral
disease is the most common form. Clinical symptoms often include facial pain and swelling,
blurring of vision, and bloody nasal discharge with crusting.[4]
Although the definitive diagnosis is histopathological,[5] MRI has been considered the imaging modality of choice for evaluation of mucormycosis.[3]
[6]
[7]
[8] It detects subtle changes in extrasinonasal soft tissues and has been found to be
a good correlate with disease extent as seen on surgery.[9]
Postcontrast enhancement patterns on MRI in mucormycosis have been classified into
three types, namely LOC, homogeneous enhancement, and heterogeneous enhancement[10] ([Figs. 3],[4],[5]). In 2010, Safder et al[11] first described the devitalized nonenhancing sinonasal mucosa as the “Black Turbinate”
sign on MRI in two cases of rhinocerebral mucormycosis, which was later reiterated
by Taylor et al.[12] Seo et al[13] also concluded that the presence of cervicofacial tissue infarction resulting from
angioinvasive nature of these fungi is seen as a region with LOC in the sinonasal
tract or its vicinity. According to the study published by Choi et al[10] in 2018, LOC was found to be the most common pattern of enhancement (47.8%) in their
study group of 23 subjects. Our findings were in concurrence with their study as 65%
of our study subjects also demonstrated LOC.
Fig. 3 (A–E) Lack of contrast enhancement. Axial contrast-enhanced fat-saturated T1-weighted
image (A) shows lack of contrast enhancement (LOC) in the soft tissue involving the left nasolacrimal
region and anterior ethmoid air cells (white arrow) with preserved in-between bone
(black arrow). The area of LOC shows predominantly hyperintense signal on T2-weighted
image (white arrowhead in B) and restricted diffusion on axial diffusion-weighted image (DWI) (double-headed
arrow in C), and corresponding apparent diffusion coefficient (ADC) map (curved arrow in D). Histology shows coagulative necrotic background and broad aseptate hyphae with
right angle branching of Mucor (black arrow in E). Small fragments of hyphae (black arrow heads) are also seen (hematoxylin and eosin
×600).
Fig. 4 (A–E) Homogeneous enhancement pattern. Axial contrast-enhanced fat-saturated T1-weighted
image (A) shows homogeneous enhancement in the right sphenoid sinus (black arrow) and right
cavernous sinus (black arrowhead). Left cavernous sinus is also bulky (white curved
arrow). Isointense signal is seen in this region on T2-weighted image (white arrow
and white arrowhead in B) with restricted diffusion on axial diffusion-weighted image (DWI) image (white arrows
in C) and corresponding apparent diffusion coefficient (ADC) map (black arrows in D). Bilateral internal carotid artery (ICA) flow voids are maintained (* in B). Histology shows long ribbon-like septate hyphae with right angle branching (black
arrowheads in E) (+3 fungal load) in the background of acute inflammatory reaction (hematoxylin and
eosin ×400).
Fig. 5 (A–E) Heterogeneous enhancement pattern. Axial contrast-enhanced fat-saturated T1-weighted
image (A) shows heterogeneous enhancement in the right maxillary sinus (white arrow) which
shows predominantly hyperintense signal on T2-weighted image (white arrowhead in B). The region does not depict any focus of restricted diffusion on corresponding diffusion-weighted
image (DWI) (C) and apparent diffusion coefficient (ADC) map (D). Histology depicts epithelioid cell granuloma (black arrowhead) admixed with multinucleate
giant cells and few fungal hyphae (+1 fungal load) (hematoxylin and eosin ×400).
In our study, the lesions showed variable signal intensity on T2WIs (9 hyperintense,
6 hypointense, and 5 isointense), whereas 18 out of 20 cases showed hypointense signal
on T1WI. Available literature suggests that low signal intensity on T1WI and an even
lower signal intensity on T2WI is usually seen in the involved areas.[14]
[15] In a study undertaken by Zinreich et al,[14] 6 patients with pathologically proven fungal sinusitis underwent MRI, all of whom
demonstrated markedly hypointense signal on T2WI. Similarly, Choi et al[10] demonstrated that homogeneously and heterogeneously enhancing lesions showed exclusively
low signal intensity on T2WI. Hypointensity on T1WI and T2WI in fungal disease has
been attributed to the accumulation of hemosiderin and paramagnetic materials such
as calcium, iron, magnesium, and manganese.[14]
[16] However, we did not see significant hypointensity on T2WI in our study, and no correlation
of T2 signal was seen with enhancement pattern. Ilica et al[7] stated that differences in T2 signal may be found due to variable degree of inflammatory
and fibrotic reaction, consistency of fluid present, fungal load, and mineral deposition
in the involved areas. Hyperintensity on T2WI may also possibly be the result of a
greater degree of necrosis, as described by Sreshta et al.[17]
Coagulative necrosis is the typical form of cell injury seen in ischemic/infarcted
tissues. In our study, 12 of the 13 patients with LOC showed coagulative necrosis
on histopathology whereas it was seen in only 2 of the 7 patients with other enhancement
patterns. This was found to be statistically significant (Fisher's exact test p-value = 0.007), with LOC having a positive likelihood ratio of 5.14 and a PPV of
92.31% for the presence of coagulative necrosis. Of these 13 subjects, 10 also depicted
a high fungal burden on histopathology with a subjective score of +3, thus establishing
a significant relationship between the presence of LOC and fungal load (p-value = 0.047). Choi et al[10] also concluded that LOC lesions are associated with coagulative necrosis and a high
fungal burden. This is also in agreement with the study performed by Safder et al,[11] who demonstrated that areas of nonenhancement in the sinonasal region contained
angioinvasive fungal hyphae on pathological correlation.
Restricted diffusion was seen in all our patients with LOC. The presence of coagulative
necrosis in restricting lesions was also prominent and seen in 13/18 patients (72.2%;
p-value = 0.521). These findings point to a common underlying pathogenesis of LOC,
restricted diffusion, and coagulative necrosis, although significant correlation between
diffusion restriction and coagulative necrosis could not be established. This may
be due to the small data set, and evaluation with further studies is warranted.
Our results indicate statistically significant correlation between presence of diffusion
restriction and fungal hyphal burden. We used a subjective method for categorizing
the extent of fungal hyphae visible on light microscopy and all lesions showing restricted
diffusion had a high fungal load, that is, score of +2 or +3 (p-value = 0.007). To the best of our knowledge, no literature is available till date
that correlates these two variables.
Most of the available literature regarding prognostic factors in mucormycosis deals
with clinical and laboratory parameters with limited studies addressing the significance
of imaging features in predicting morbidity and mortality. Underlying diseases, accompanying
neutropenia, C-reactive protein levels, symptom duration, involvement of nasal septum,
and the presence of facial swelling have been seen to influence the prognosis of mucormycosis.[18] We analyzed our study data for any probable association of imaging features with
disease prognosis/outcome and found that enhancement pattern may act as a predictor
for prolonged hospital stay (p-value = 0.004). All patients with hospital stay longer than 6 weeks showed LOC, whereas
all patients with homogeneous or heterogeneous enhancement were discharged within
6 weeks. Although mortality was seen exclusively in the LOC group, this was not found
to be statistically significant (p-value = 0.251). Enhancement pattern and diffusion restriction did not correlate with
the final patient outcome in terms of survival; however, the role of imaging in predicting
survival may have been downplayed by the overall low mortality in our study population.
In the study by Choi et al,[10] LOC was found to be the sole independent prognostic factor for disease-specific
mortality, probably due to a high fungal load and rapidly progressive coagulative
necrosis. On analysis of pre- and postoperative enhanced MR images, Kim et al[19] concluded that extrasinonasal presence of nonenhancing lesions resulted in poor
disease-specific survival as compared to individuals having only intrasinonasal involvement.
Limitations of our study included a small sample size with presence of concurrent
diabetes as a risk factor in 80% of the subjects which may represent a significant
bias. Second, we saw a relatively favorable prognosis in CAM patients (85% survival),
which is better when compared with survival statistics of mucormycosis in the pre-COVID
era.[20] This may have been either due to a better immune status of our patients who had
no other significant cause of immunosuppression apart from recent COVID-19 infection
or early intervention owing to a high index of suspicion during the pandemic. Third,
duration of hospital stay may have additionally been affected by other complications
secondary to COVID-19, complete assessment of which was not practically possible.
Conclusion
The importance of MRI as a tool for diagnosis of mucormycosis cannot be overstated.
In our study, we have described the MRI findings in CAM, highlighting the imaging
features which may act as surrogate markers for the presence of coagulative necrosis
and fungal burden. We found that LOC was seen to correlate with presence of coagulative
necrosis and high fungal load, while presence of diffusion restriction correlated
with a high fungal burden. Although LOC also correlated with prolonged hospital stay,
the case series is too limited to provide prediction of the patients' outcomes and
it may act as a preliminary study upon which further studies can be based.
