CT manifestations of pulmonary mucormycosis: Influence of risk factor, symptom duration, and glycemic control

Abstract

Background:

Pulmonary mucormycosis (PM) imaging has been characterised primarily in patients with hematologic malignancies. Imaging patterns may differ in diabetes mellitus-associated PM (DAPM), the most common risk factor globally, due to different immunologic profiles and disease progression. The impact of glycemic control on CT findings in DAPM is unknown.

Methods:

We retrospectively analysed chest CT scans of patients with PM, compared DAPM with PM in other risk factors (PMOR), and evaluated imaging features across glycated haemoglobin (HbA1c) categories. The secondary objective was to determine associations between imaging features and mortality in DAPM. Imaging patterns recorded by two radiologists were compared between study groups, and multivariable logistic regression analysis of mortality was performed.

Results:

Among 193 patients (161 DAPM, 32 PMOR), consolidation (97%) and cavitation (84%) were the most frequent. DAPM presented significantly later (median 30 vs. 7 days; P < 0.001) and showed distinct features: smaller nodules (62%, <10 mm) with centrilobular distribution (37.4%), while PMOR had larger nodules (10–30 mm, 55.0%) with random distribution (75.0%). Lymphadenopathy was more common in DAPM (26.0% vs. 9.4%, P = 0.043). Among DAPM, Bird’s nest sign increased with HbA1c severity: 7.7% (HbA1c ≤7%), 33.3% (HbA1c 7-9%), and 43.3% (HbA1c>9%), P = 0.030. Overall, 12-week mortality was 47.9%. Although the reversed halo sign and bird’s nest sign were associated with mortality on univariable analysis, neither was significant after adjusting for clinical factors.

Conclusion:

DAPM presents later and exhibits distinct imaging features compared to PMOR. Bird’s nest sign is more frequent with HbA1c >9%. No imaging features were independently associated with mortality.

INTRODUCTION

Mucormycosis is an aggressive, angio-invasive fungal infection caused by fungi of the order Mucorales.[1] Pulmonary mucormycosis (PM) is the second most common form of mucormycosis and carries a high mortality rate.[2] The epidemiology of mucormycosis differs markedly between developed and developing countries. In high-income countries, haematological malignancy is the dominant risk factor, whereas diabetes mellitus (DM) accounts for most cases in low- and middle-income countries (LMICs).[1,3,4,5] Despite DM being the leading risk factor globally, PM manifestations in diabetic patients remain less well characterised than in neutropenic hosts.[6,7] Uncontrolled DM is associated with particularly high risk,[8] and the disease follows an aggressive course with mortality rates comparable to other risk factors.[2,7]

Treatment options for PM are limited, making early diagnosis essential.[5,9] Computed tomography (CT) plays a pivotal role in raising clinical suspicion for PM and guiding diagnostic evaluation.[5,10,11,12,13] Earlier studies comparing PM with invasive pulmonary aspergillosis (IPA) identified multiple nodules and pleural effusion as key distinguishing features of PM.[14] However, these studies primarily included patients with haematological malignancies, and the imaging appearance of PM in non-neutropenic hosts, particularly those with DM, remains poorly described.[5,10,14,15] Studies in neutropenic patients have demonstrated temporal evolution of PM on serial CT, progressing from early manifestations (nodules or consolidation with halo sign) to advanced features (reversed halo sign, cavitation, air-crescent sign) correlating with neutrophil recovery.[11] In contrast, patients with DM have relatively preserved neutrophil function and may not be under active surveillance, potentially leading to presentation at more advanced disease stages. Whether this difference in host immune status and clinical context translates to distinct imaging patterns at presentation between DAPM and neutropenic hosts remains unclear.

The COVID-19 pandemic led to increased recognition of PM and generated data on imaging manifestations, predominantly in COVID-19-associated cases.[3,6,12,16,17] However, comprehensive data on CT manifestations of DM-associated PM (DAPM) in non-COVID-19 settings remain limited.[7,18,19,20] While studies have compared IPA and PM,[14,21] large datasets characterizing DAPM or evaluating the impact of glycemic control (glycated haemoglobin, HbA1c) on imaging manifestations are lacking.

We hypothesised that CT manifestations would differ between DAPM and PMOR, and that imaging features would vary across glycemic control categories in DAPM. Therefore, the objectives of this study were to compare CT findings between (1) DAPM and PM from other risk factors (PMOR), and (2) categories of glycemic control in DAPM (HbA1c ≤7%, 7–9%, >9%). The secondary objective was to identify imaging features independently associated with 12-week mortality in DAPM.

METHODS

Study design and participants

We retrospectively reviewed our database to identify subjects aged 12 years or older with PM diagnosed between January 2014 and December 2024 at a tertiary care referral centre in northern India. The Institute Ethics Committee approved the study protocol (IEC-INT/2023/STUDY-976) and granted a consent waiver, given the study’s retrospective nature and the use of anonymised participant data. Clinical data from some patients were previously reported.[18,22,23,24]

We included subjects with proven or probable PM who had at least one CT chest done (including non-contrast CT, contrast-enhanced CT, CT bronchial angiography, and CT pulmonary angiography) at presentation. We excluded cases with (1) suboptimal images for interpretation due to motion artifact, or (3) possible PM.

Diagnostic criteria

We diagnosed PM per the updated European Organization for Research and Treatment of Cancer/Mycoses Study Group (EORTC/MSG) criteria, modified to include DM as a host factor.[1,25] Proven PM was diagnosed in patients with clinical-radiologic suspicion when aseptate hyphae were demonstrated on tissue samples or when microscopy showed aseptate hyphae or culture grew Mucorales from a normally sterile site (e.g., pleural fluid or fine-needle aspiration from the lung). Probable PM was diagnosed when microscopy revealed aseptate hyphae or culture grew Mucorales from bronchoalveolar lavage fluid (BALF) or sputum in a patient with suspected invasive mold disease. We labelled COVID-19-associated PM (CAPM) cases according to the published Delphi consensus criteria.[26]

Patient categorisation

We categorised patients into two groups based on underlying risk factors: (1) diabetes mellitus-associated PM, and (2) PM from other risk factors, which included solid organ transplant recipients, haematological malignancies, and other immunocompromising conditions. When multiple risk factors were present, we categorised patients hierarchically (e.g., a patient with organ transplant-related diabetes mellitus was categorised as a solid organ transplant).

Radiological assessment

Two thoracic radiologists (NP, MG) independently reviewed all images from Digital Imaging and Communications in Medicine (DICOM) files stored on the Picture Archiving and Communication system (PACS) or from digitised images from CT chest films (10 mm sections). Discrepancies were resolved by joint consensus review. The radiologists were blinded to patient outcomes but not to clinical history. We classified radiologic signs according to the definitions of the Fleischner Society and the EORTC/MSG Imaging Working Group.[27,28]

We recorded the following features: nodules (further categorised by size: <10 mm, 10–30 mm and >30 mm), consolidation, reverse halo sign, hypodense sign, air crescent sign, halo sign and significant lymphadenopathy (≥10 mm short-axis diameter). Large nodules (10–30 mm) were considered as early features, whereas cavitation or lymphadenopathy were considered features of chronicity.[11] Detailed definitions of each radiologic sign are provided in Table S1.

Table S1.

Terminologies used in the current study

Terminology		Definition used
Nodules		Well-defined or ill-defined round or irregular opacities in the lungs measuring ≤3 cm in average diameter
Small Nodules		Nodules measuring ≤1 cm in average diameter
Consolidation		Homogeneous increase in pulmonary parenchymal attenuation obscuring the margins of vessels and airway walls, which may show air bronchogram, hypodense area, cavitation, or GGOs within it. Mass-like consolidation (i.e focal, rounded opacity) in the lung >3 cm diameter is also included in the definition
Cavitation		Gas-filled spaces seen in a nodule or consolidation
Reverse halo sign		Focal area of GGO surrounded by a complete or incomplete consolidation ring
Hypodense sign		Central lower attenuation area seen in a nodule or consolidation
Halo sign		Rim of GGO seen surrounding a nodule or consolidation.
Bird nest sign		Septations within cavitation surrounded by GGOs and a complete or incomplete consolidation ring or GGOs ring
Air-crescent sign		Crescent-shaped cavitation in a nodule or consolidation
Mediastinal lymphadenopathy		Any of the mediastinal lymph nodes measuring >1 cm in short axis diameter.

GGO=ground glass opacities

Clinical data collection

We extracted the following information from patient medical records using a standardised case-record form: (1) age, (2) sex, (3) underlying risk factors (diabetes mellitus, solid organ transplant recipient, haematological malignancies, COVID-19), (4) duration of symptoms prior to presentation, (5) presence or absence of haemoptysis, (6) severe haemoptysis (defined as haemoptysis leading to significant airway obstruction, abnormal gas exchange, hemodynamic instability, death, or required hospitalisation, blood transfusion, bronchial artery embolisation or surgery for management), (7) HbA1C values at presentation, (8) treatment approach (medical only, surgical only, or combined medical-surgical management), and (9) all-cause mortality at 12 weeks from diagnosis.

All patients were managed according to institutional protocols aligned with international and national guidelines.[1,17,26] Management included optimisation of underlying risk factors (glycemic control and reversal of immunosuppression when feasible), antifungal therapy (intravenous amphotericin B or oral posaconazole), and surgery when indicated. Surgical decisions were made by a multidisciplinary team, comprising pulmonary physicians and thoracic surgeons, and were tailored to each patient’s individual needs.

Statistical analysis

We performed data analysis using R (version 4.3.1) and RStudio.[29] We used the tidyverse, gtsummary and mice packages for data analysis.[30,31] Continuous data are presented as medians (1^st, 3^rd quartile), and categorical data as frequencies (proportions). We compared differences between DAPM and PMOR groups using the Chi-squared test (or Fisher’s exact test when expected cell counts were <5) for categorical variables and the Wilcoxon rank-sum test for continuous variables. As an exploratory analysis, the distribution of symptom duration at the time of evaluation was examined in relation to selected CT imaging features using ridgeline plots. Symptom duration was defined as the number of days from symptom onset to CT evaluation. For each imaging feature, analyses were restricted to patients with the feature. We visualised distributions using ridgeline plots based on kernel density estimates to illustrate the shape and spread of symptom duration for each feature.

We performed multivariable binary logistic regression analysis to identify factors independently associated with 12-week mortality in DAPM. Variables were selected based on biological plausibility or univariable analysis with P < 0.10. We assessed all variables for missing data and evaluated the patterns of missingness. Missing data were handled using multiple imputations by chained equations (MICE) via the mice package. We generated 50 imputed datasets (m = 50) with 10 iterations each to ensure convergence of the imputation algorithm. Parameter estimates from the imputed analyses were combined using Rubin’s rules to yield pooled odds ratios (ORs), 95% confidence intervals (CIs), and P values. A complete case analysis was performed as a sensitivity analysis. Statistical significance was defined as two-tailed P < 0.05.

RESULTS

Of 211 subjects with PM diagnosed during the study period, 18 were excluded due to suboptimal CT or clinical data, leaving 193 participants for analysis [Table 1]. Of these, 161 had DAPM, and 32 had PMOR [Table 1]. The PMOR group included 26 renal transplant recipients, 2 with haematological malignancies, and 4 with other immunosuppressive conditions. COVID-19 was documented in 30 subjects (29 DAPM, one renal transplant recipient).

Table 1.

Clinical and imaging features of PM in Diabetes mellitus vs other predisposing factors

Characteristic		Overall, (n=193)1		DAPM, (n=161)1		PMOR, (n=32)1		P 2
Demographics
Age, years		52 (42–60)		53 (45–60)		40 (32–50)		<0.001
Male sex		144/193 (74.6%)		118/161 (73.3%)		26/32 (81.3%)		0.34
Diagnosis								<0.001
Proven		82/193 (42.5%)		56/161 (34.8%)		26/32 (81.3%)
Probable		111/193 (57.5%)		105/161 (65.2%)		6/32 (18.8%)
Risk factors and outcome
Predisposing factor								<0.001
Diabetes mellitus		161/193 (83.4%)		161/161 (100%)		-
HbA1c		11.3 (8.9–13.1)		11.3 (8.9–13.1)		-
Renal transplantation		26/193 (13.5%)		–		26/32 (81.3%)
Transplant duration (months)		6.5 (4.0–11.0)		–		6.5 (4.0–11.0)
Haematological malignanciesa		2/193 (1.0%)		–		2/32 (6.2%)
Othersb		4/193 (2.1%)		–		4/32 (12.5%)
COVID-19		30/193 (15.5%)		29/161 (18.0%)		1/32 (3.1%)		0.033
Clinical features
Duration of symptoms, daysc		22 (12–40)		30 (15–45)		7 (4–10)		<0.001
Symptom duration ≤10 days		32/129 (24.8%)		11/101 (10.9%)		21/28 (75%)		<0.001
Haemoptysis		77/171 (45.0%)		71/141 (50.4%)		6/30 (20.0%)		0.002
Severe haemoptysis		33/168 (19.6%)		28/139 (20.1%)		5/29 (17.2%)		0.72
Treatment and outcome
Combined surgical-medical management		34/193 (17.6%)		22/161 (13.7%)		12/32 (37.5%)		0.001
In-hospital mortality		68/176 (38.6%)		58/148 (39.2%)		10/28 (35.7%)		0.73
12-week mortality		78/163 (47.9%)		66/138 (47.8%)		12/25 (48.0%)		0.99
CT findings
Consolidation		185/191 (96.9%)		155/159 (97.5%)		30/32 (93.8%)		0.33
Single		96/185 (51.9%)		78/155 (50.35%)		18/30 (60.0%)
Multiple		89/185 (48.1%)		77/155 (49.7%)		12/30 (40.0%)
Number of consolidations								0.49
1		93/185 (50.3%)		75/155 (48.4%)		18/30 (60.0%)
2		48/185 (26.0%)		38/155 (24.5%)		10/30 (33.3%)
3		18/185 (9.7%)		18/155 (11.6%)		0
4		9/185 (4.9%)		8/155 (5.2%)		1/30 (3.3%)
≥5		17/185 (9.2%)		16/155 (10.3%)		1/30 (3.3%)
Size of the smallest consolidation, cm		6.0 (4.0–8.0)		6.0 (4.0–8.0)		6.00 (4.2–8.0)		0.84
Size of the largest consolidation, cm		8.0 (6.7–10.0)		8.0 (7.0–10.0)		7.2 (6.0–8.5)		0.036
Reversed halo sign		109/191 (57.1%)		95/159 (59.8%)		14/32 (43.8%)		0.10
Cavitation		160/191 (83.8%)		135/159 (84.9%)		25/32 (78.1%)		0.34
Bird’s nest sign		75/191 (39.3%)		67/159 (42.1%)		8/32 (25.0%)		0.070
CT Hypodense Sign		91/191 (47.6%)		77/159 (48.4%)		14/32 (43.8%)		0.63
Halo sign		16/191 (8.4%)		12/159 (7.6%)		4/32 (12.5%)		0.32
Air-fluid level		78/190 (41.1%)		68/158 (43.0%)		10/32 (31.3%)		0.22
Air crescent sign		18/190 (9.5%)		13/159 (8.2%)		5/31 (16.1%)		0.18
Chest wall involvement		6/98 (6.1%)		6/81 (7.4%)		0		0.59
Pseudoaneurysm		11/72 (15.3%)		11/67 (16.4%)		0		>0.99
Significant lymphadenopathy		44/190 (23.2%)		41/158 (26.0%)		3/32 (9.4%)		0.043
Pleural effusion								0.82
Absent		114/190 (60.0%)		94/158 (59.5%)		20/32 (62.5%)
Unilateral		63/190 (33.2%)		52/158 (32.9%)		11/32 (34.4%)
Bilateral		13/190 (6.8%)		12/158 (7.6%)		1/32 (3.1%)
Pneumothorax		21/191 (10.1%)		19/159 (12.0%)		2/32 (6.0%)		0.50
Ground glass opacities		36/190 (18.9%)		31/158 (19.6%)		5/32 (15.6%)		0.60
Significant nodule		119/191 (62.3%)		99/159 (62.3%)		20/32 (62.5%)		0.98
Predominant nodule								0.001
1–10 mm		69/119 (58.0%)		61/99 (61.6%)		8/20 (40.0%)
10–30 mm		28/119 (23.5%)		17/99 (17.2%)		11/20 (55.0%)
Both equally predominant		22/119 (18.5%)		21/99 (21.2%)		1/20 (5.0%)
Nodule distribution								0.002
Centrilobular		40/119 (33.6%)		37/99 (37.4%)		3/20 (15.0%)
Random		47/119 (39.5%)		32/99 (32.3%)		15/20 (75.0%)
Mixed		32/119 (26.9%)		30/99 (30.3%)		2/20 (10.0%)

¹Median (Q1-Q3); n/N (%); ²Wilcoxon rank sum test; Pearson’s Chi-squared test; Fisher’s exact test. ^aAcute lymphoblastic leukaemia, Chronic myeloid leukaemia blast crisis; ^bImmunosuppression for autoimmune diseases in three, and no identifiable risk factor in one; ^cData available for 101/161 and 28/32 subjects with DAPM and PMOR respectively

Delayed presentation and distinct imaging findings in DAPM

Individuals with DAPM had a significantly longer symptom duration at presentation compared with PMOR (30 vs. 7 days; P < 0.001). Subjects with DAPM were older (median 53 vs. 40 years, P < 0.001), and more frequently reported haemoptysis (50.4% vs. 20.0%, P = 0.002) than PMOR, although the frequency of severe haemoptysis was comparable between groups. Surgery was performed more frequently in PMOR (37.5% vs. 13.7%, P = 0.001). Overall, 12-week mortality was 47.9%, with no difference between groups (P = 0.99).

CT findings [Figures 1 and 2] were reported from DICOM (n = 86) and digitised images (n = 107). The most common findings were consolidation (185/191, 96.9%) and cavitation (82.9%), but were not different across groups [Table 1]. Pseudoaneurysm (n = 11) and chest wall involvement (n = 6) were observed exclusively in DAPM [Figures 3 and 4]. Other common findings were RHS (57%), the bird’s nest sign (40%), CT hypodense sign (47%) and the halo sign (8%). These features did not differ significantly between DAPM and PMOR.

Figure 1.

CT chest lung window sections demonstrate the most common findings of pulmonary mucormycosis, namely consolidation (arrows in images a-c), and cavitation (dashed arrows in c-e) in different patients. Note is also made of the associated reverse halo sign in areas of consolidation in (b) and air-fluid levels in areas of cavitation in (f)

Figure 2.

CT chest sections in different patients demonstrating the various signs seen in pulmonary mucormycosis (a) Reverse halo sign (black arrow), (b) CT hypodense sign (white arrow), (c) Air-fluid level sign (black arrowhead), (d) Bird’s nest sign (black dashed arrow), (e) Air-crescent sign (white arrowhead), (f) Halo sign (white dashed arrow)

Figure 3.

CT chest images showing pseudoaneurysms in patients with diabetes associated pulmonary mucormycosis (a) Coronal section in mediastinal window and (b) Volume rendered technique demonstrating pseudoaneurysm of pulmonary artery segmental branch in left lower lobe (arrows) seen in a patient, and (c, d) Axial sections in mediastinal window demonstrating pulmonary artery segmental branches pseudoaneurysms (arrows) in right lower lobe in two different patients

Figure 4.

CT chest images (a and c) Lung window and (b and d) Mediastinal window in four different patients demonstrating the chest wall invasion in patients with diabetes associated pulmonary mucormycosis (arrows)

The median size of the largest consolidation was greater in DAPM (8.0 vs. 7.2 cm, P = 0.036), as was the frequency of significant lymphadenopathy (26.0% vs. 9.4%, P = 0.043). Nodules were present in 62% of subjects overall. Among those with nodules, the predominant size and distribution patterns differed between groups (P = 0.001 and P = 0.002, respectively). DAPM predominantly showed smaller nodules (1–10 mm in 61.6%) with a centrilobular distribution (37.4%), whereas PMOR more frequently had nodules of 10–30 mm (55.0%) with a random distribution (75.0%).

Subjects with CAPM more frequently had ground-glass opacities (43.3% vs. 14.4%, P < 0.001) and pseudoaneurysms (45.5% vs. 9.8%, P = 0.009) than non-COVID-19 PM [Table S2].

Table S2.

Clinical and imaging features of COVID-19-associated pulmonary mucormycosis (CAPM) vs non-COVID-19 pulmonary mucormycosis (PM)

Characteristic		Non-COVID PM, (n=163)1		CAPM, (n=30)1		P 2
Age, years		51.00 (42.00–60.00)		55.00 (45.00–61.00)		0.19
Male sex		118/163 (72.4%)		26/30 (86.7%)		0.10
HbA1c		11.90 (9.30–13.40)		9.30 (8.10–11.90)		0.035
Predisposing factors						0.033
Diabetes mellitus		132/163 (81.0%)		29/30 (96.7%)
Others		31/163 (19.0%)		1/30 (3.3%)
Any haemoptysis		58/141 (41.1%)		19/30 (63.3%)		0.026
Severe haemoptysis		28/139 (20.1%)		5/29 (17.2%)		0.72
Combined medical surgical management		27/163 (16.6%)		7/30 (23.3%)		0.37
In-hospital mortality		56/146 (38.4%)		12/30 (40.0%)		0.87
12-week mortality		65/133 (48.9%)		13/30 (43.3%)		0.58
CT appearance
Consolidation		157/161 (97.5%)		28/30 (93.3%)		0.24
Single		78/157 (49.7%)		18/28 (64.3%)
Multiple		79/157 (50.3%)		10/28 (35.7%)
Reversed halo sign		95/161 (59.0%)		14/30 (46.7%)		0.21
Cavitation		135/161 (83.9%)		25/30 (83.3%)		>0.99
Bird nest sign		64/161 (39.8%)		11/30 (36.7%)		0.75
CT hypodense sign		77/161 (47.8%)		14/30 (46.7%)		0.91
Halo sign		13/161 (8.1%)		3/30 (10.0%)		0.72
Air-fluid level		66/161 (41.0%)		12/29 (41.4%)		0.97
Air crescent sign		15/160 (9.4%)		3/30 (10.0%)		>0.99
Pseudoaneurysm		6/61 (9.8%)		5/11 (45.5%)		0.009
Significant lymphadenopathy		40/160 (25.0%)		4/30 (13.3%)		0.16
Pleural effusion						0.40
Absent		99/160 (61.9%)		15/30 (50.0%)
Unilateral		51/160 (31.9)		12/30 (40.0%)
Bilateral		10/160 (6.3%)		3/30 (10.0%)
Pneumothorax		5/161 (9.3%)		6/30 (20.0%)		0.22
Chest wall involvement		5/86 (5.8%)		1/12 (8.3)		0.55
Ground glass opacities		23/160 (14.4%)		13/30 (43.3%)		<0.001
Predominant nodule						0.17
1–10 mm		54/97 (55.7%)		15/22 (68.2%)
10–30 mm		22/97 (22.7%)		6/22 (27.3%)
Both equally predominant		21/97 (21.7%)		1/22 (4.5%)
Nodule distribution						0.55
Centrilobular		31/97 (32.0%)		9/22 (40.9%)
Random		38/97 (39.2%)		9/22 (40.9%)
Mixed		28/97 (28.9%)		4/22 (18.2%)

¹Median (Q1-Q3); n/N (%). ²Wilcoxon rank sum test; Pearson’s Chi-squared test; Fisher’s exact test

Bird’s nest sign was more frequent with poor glycemic control

HbA1C data were available for 132/161 (82%) DAPM subjects [Table 2]. The frequency of bird’s nest sign increased across HbA1c categories: 7.7% (HbA1c ≤7%), 33.3% (HbA1c 7–9%), and 43.3% (HbA1c >9%), P = 0.030 [Table 2]. Haemoptysis also varied across categories (P = 0.007), but no other imaging or clinical features differed significantly.

Table 2.

Clinical and radiological characteristics of participants with diabetes mellitus and pulmonary mucormycosis stratified glycated haemoglobin categories

Characteristic		HbA1C ≤7%, (n=13)1		HbA1C 7–9%, (n=21)1		HbA1C >9%, (n=98)1		P 2
Demographics
Age, years		54 (45–59)		59 (50–65)		52 (43–60)		0.084
Male sex		11/13 (84.6%)		17/21 (81.0%)		69/98 (70.4%)		0.45
Diagnosis								0.67
Proven		3/13 (23.1%)		8/21 (38.1%)		31/98 (31.6%)
Probable		10/13 (76.9%)		13/21 (61.9%)		67/98 (68.4%)
Clinical features
Duration of symptoms, days		30 (30–40)		30 (25–60)		25 (15–40)		0.028
Any haemoptysis		7/12 (58.3%)		16/20 (80.0%)		35/84 (41.7%)		0.007
Severe haemoptysis		0		6/20 (30.0%)		16/82 (19.5%)		0.11
HbA1c, %		6.0 (5.5–6.5)		8.1 (7.8–8.6)		12.3 (10.7–13.7)		<0.001
COVID-19		4/13 (30.8%)		7/21 (33.3%)		14/98 (14.3%)		0.060
Treatment and outcome
Combined medical-surgical management		2/13 (15.4%)		5/21 (23.8%)		11/98 (11.2%)		0.22
Hospital mortality		4/12 (33.3%)		5/19 (26.32)		38/96 (39.6%)		0.52
12-week mortality		4/12 (33.3%)		8/19 (42.1%)		42/88 (47.7%)		0.61
CT findings
Consolidation		13/13 (100.0%)		21/21 (100.0%)		93/97 (95.9%)		>0.99
Single		7/13 (53.9%)		8/21 (38.1%)		46/93 (49.5%)
Multiple		6/13 (46.2%)		13/21 (61.9%)		47/93 (50.5%)
Size of the largest consolidation, cm		8.0 (5.0–10.0)		7.0 (6.0–9.0)		8.0 (7.0–10.0)		0.30
Reversed halo sign		5/13 (38.5%)		9/21 (42.9%)		60/97 (61.9%)		0.11
Cavitation		10/13 (76.9%)		19/21 (90.5%)		83/97 (85.6%)		0.53
Bird nest sign		1/13 (7.7%)		7/21 (33.3%)		42/97 (43.3%)		0.030
CT hypodense sign		4/13 (30.8%)		11/21 (52.4%)		52/97 (53.6%)		0.30
Halo sign		0		1/21 (4.8%)		7/97 (7.2%)		>0.99
Air-fluid level		6/13 (46.2%)		12/21 (57.1%)		36/96 (37.5%)		0.24
Air crescent sign		2/13 (15.4%)		1/21 (4.8%)		8/97 (8.3%)		0.57
Pseudoaneurysm		0		1/8 (12.5%)		7/46 (15.2%)		>0.99
Significant lymphadenopathy		1/13 (7.7%)		5/21 (23.8%)		28/96 (29.2%)		0.27
Pleural effusion								0.88
Absent		8/13 (61.5%)		14/21 (66.7%)		56/96 (58.3%)
Unilateral		4/13 (30.8%)		5/21 (23.8%)		33/96 (34.4%)
Bilateral		1/13 (7.7%)		2/21 (9.5%)		7/96 (7.3%)
Pneumothorax		0		0		13/97 (13.4%)		0.61
Chest wall involvement		0		1/12 (8.3%)		3/48 (6.3%)		>0.99
Ground glass opacities		1/13 (7.7%)		5/21 (23.8%)		19/97 (19.6%)		0.61
Significant nodule		7/13 (53.9%)		11/21 (52.4%)		67/97 (69.1%)		0.23
Predominant nodule								0.81
1–10 mm		3/7 (42.9%)		6/11 (54.6%)		41/67 (61.2%)
10–30 mm		2/7 (28.6%)		2/11 (18.2%)		11/67 (16.4%)
Both equally predominant		2/7 (28.6%)		3/11 (27.3%)		15/67 (22.4%)
Nodule distribution								0.56
Centrilobular		3/7 (42.9%)		4/11 (36.4%)		25/67 (37.3%)
Random		3/7 (42.9%)		5/11 (45.5%)		18/67 (26.9%)
Mixed		1/7 (14.3%)		2/11 (18.2%)		24/67 (35.8%)

HbA1c=glycated haemoglobin, CT=computed tomography. ¹Data presented as median (interquartile range) for continuous variables and n (%) or n/N (%) for categorical variables, ²Kruskal-Wallis rank sum test for continuous variables; Pearson’s Chi-squared test or Fisher’s exact test for categorical variables

Temporal relationship between symptom duration and imaging features

Analysis of symptom duration in relation to individual CT features revealed temporal patterns [Figure 5]. Early features, such as large nodules ≥10 mm, clustered toward shorter symptom durations, while lymphadenopathy (median duration, 30 days) appeared predominantly in patients with longer disease courses. However, substantial overlap in distributions indicates that individual imaging features cannot reliably determine disease duration, and multiple factors likely influence when specific signs develop.

Figure 5.

Ridgeline plots showing the distribution of symptom duration at the time of evaluation among patients in whom selected CT imaging features were present. Ridges depict smoothed distributions of symptom duration among patients with the indicated CT feature. Of the 125 patients with the necessary details available, the number of patients exhibiting each feature were as follows: large nodules ≥10 mm (n = 19), consolidation (n = 121), reversed halo sign (n = 73), bird’s nest sign (n = 57), halo sign (n = 12), cavitation (n = 108), and lymphadenopathy (n = 26). Black dots indicate the median value for each sign, and the dashed vertical line denotes a symptom duration of 10 days

Mortality analysis

Twelve-week mortality data were available for 138/161 (85.7%) DAPM subjects, of whom 66 (47.8%) died. On univariable analysis [Table S3], non-survivors had higher HbA1c (12.3% vs. 11.1%, P = 0.042), received less surgical intervention (7.6% vs. 22.2%, P = 0.017), and more frequently had RHS (68% vs. 49%; P = 0.03), bird nest sign (50.8% vs. 31%; P = 0.02), and larger consolidations (9.0 vs. 8.0 cm, P = 0.029).

Table S3.

Clinical and imaging characteristics of survivors and non-survivors (at 12 weeks) in subjects with diabetes mellitus and pulmonary mucormycosis

Characteristic		Alive, (n=72)1		Dead, (n=66)1		P 2
Demographics
Age, years		52.5 (44.5–59.5)		52.0 (42.0–60.0)		0.89
Male sex		51/72 (70.8%)		49/66 (74.2%)		0.65
Diagnosis						0.14
Proven		27/72 (37.5%)		17/66 (25.8%)
Probable		45/72 (62.5%)		49/66 (74.2%)
Clinical features
Duration of symptoms, days		30 (15–45)		25 (15–40)		0.49
Symptom duration ≤10 days		4/51 (7.8%)		6/37 (16.2%)		0.31
Any haemoptysis		32/62 (51.6%)		28/58 (48.3%)		0.71
Severe haemoptysis		8/61 (13.1%)		15/57 (26.3%)		0.070
HbA1c, %		11.1 (8.5–12.6)		12.3 (9.3–13.7)		0.042
COVID-19		16/72 (22.2%)		13/66 (19.7%)		0.72
Combined medical-surgical treatment		16/72 (22.2%)		5/66 (7. 6%)		0.017
CT findings
Consolidation		68/71 (95.8%)		64/65 (98.5%)		0.62
Single		38/68 (55.9%)		26/64 (40.6%)
Multiple		30/68 (44.1%)		38/64 (59.4%)
Size of the largest consolidation		8.00 (6.60–9.00)		9.00 (7.00–10.30)		0.029
Reversed halo sign		35/71 (49.3%)		44/65 (67.7%)		0.030
Cavitation		60/71 (84.5%)		54/65 (83.1%)		0.82
Bird nest sign		22/71 (31.0%)		33/65 (50.8%)		0.019
CT hypodense sign		39/71 (54.9%)		27/65 (41.5%)		0.12
Halo sign		3/71 (4.2%)		7/65 (10.8%)		0.19
Air-fluid level		27/70 (38.6%)		28/65 (43.1%)		0.59
Air crescent sign		3/71 (4.3%)		6/65 (9.2%)		0.31
Pseudoaneurysm		4/29 (13.8%)		6/29 (20.7%)		0.49
Significant lymphadenopathy		15/71 (21.1%)		16/64 (25.0%)		0.59
Pleural effusion						0.12
Absent		47/71 (66.2%)		32/64 (50.0%)
Unilateral		20/71 (28.2%)		24/64 (37.5%)
Bilateral		4/71 (5.6%)		8/64 (12.5%)
Pneumothorax		6/71 (8.5%)		11/65 (16.0%)		0.46
Chest wall involvement		1/35 (2.9%)		3/37 (8.1%)		0.61
Ground glass opacities		12/71 (16.9%)		18/65 (27.7%)		0.13
Significant nodule		43/71 (60.6%)		42/65 (64.6%)		0.63
Predominant nodule						0.36
1–10 mm		28/43 (65.1%)		21/42 (50.0%)
10–30 mm		7/43 (16.3%)		9/42 (21.4%)
Both equally predominant		8/43 (18.6%)		12/42 (28.6%)
Nodule distribution						0.95
Centrilobular		16/43 (37.2%)		15/42 (35.7%)
Random		15/43 (34.9%)		14/42 (33.3%)
Mixed		12/43 (27.9%)		13/42 (31.0%)

¹Median (Q1-Q3); n/N (%); ²Wilcoxon rank sum test; Pearson’s Chi-squared test; Fisher’s exact test

In multivariable analysis, multiple consolidations and the bird’s nest sign were not independently associated with 12-week mortality, after adjusting for age, sex, HbA1c and surgical management [Table 3]. A complete case analysis (Table S4) also revealed no independent associations with mortality.

Table 3.

Adjusted odds of 12-week mortality among subjects with pulmonary mucormycosis in diabetes mellitus (multiple imputations, n=50)

Characteristic		Odds Ratio (95% Confidence interval)		P
Age, years		0.99 (0.96–1.03)		0.7
Sex
Male		Reference
Female		0.68 (0.29–1.59)		0.4
Type of consolidation
Single		Reference
Multiple		1.71 (0.81–3.61)		0.2
Bird’s nest sign
No		Reference
Yes		1.66 (0.77–3.58)		0.2
Glycated haemoglobin, %		1.13 (0.98–1.31)		0.088
Combined medical-surgical management
No		Reference
Yes		0.47 (0.15–1.50)		0.2

Table S4.

Adjusted odds of 12-week mortality among subjects with pulmonary mucormycosis in diabetes mellitus (complete case analysis)

Characteristic		Odds ratio (95% confidence interval)		P
Age, years		0.99 (0.96–1.02)		0.5
Sex
Male		Reference
Female		0.85 (0.34–2.08)		0.7
Bird’s nest sign
No		Reference
Yes		1.23 (0.54–2.79)		0.6
Type of consolidation
Single		Reference
Multiple		1.53 (0.71–3.37)		0.3
Glycated haemoglobin, %		1.13 (0.98–1.31)		0.10
Combined medical-surgical management
No		Reference
Yes		0.61 (0.17–1.96)		0.4

DISCUSSION

We found consolidation (97%) and cavitation (84%) to be the most common imaging findings in PM, occurring with similar frequency in both DAPM and PMOR. However, important differences emerged: DAPM demonstrated significantly longer symptom duration, higher rates of lymphadenopathy, predominantly small nodules with centrilobular distribution and exclusive occurrence of pseudoaneurysms and chest wall involvement in our cohort. The bird’s nest sign was more frequent in patients with HbA1c >9%, suggesting that glycemic control may influence imaging manifestations. Despite associations between certain CT features and mortality on univariable analysis, none were independently associated after multivariable adjustment.

CT plays a crucial role in the diagnostic evaluation of PM. Most studies describing CT in PM are small (30–50 patients), with a recent study by Coste et al.[5] (n = 114) being the notable exception.[5,10,11,14,15,16,18,20,21,32,33,34,35] Unfortunately, DM was underrepresented in the latter study (4%). Previous studies [Table 4] reported a variable proportion of consolidation and cavitation. For example, Jung et al.[20] reported consolidation in 71% and cavitation in 33% of patients, while a recent study of 34 subjects with haematological malignancy and PM noted cavitation in only 20.6%.[21] These discrepancies may reflect differences in patient populations, underlying risk factors, and the timing of imaging. The high frequency of both consolidation and cavitation in our cohort is related to the predominance of DAPM (83.4%), which presented significantly later than PMOR.

Table 4.

Summary of studies reporting on the key CT features (mentioned as percentages) of pulmonary mucormycosis (proven or probable)

References	Country	n	Key risk factor	DM (%)	Consolidation (%)	Cavity (%)	Nodules (%)	RHS (%)	Halo (%)	BNS (%)	Air crescent (%)	CT hypodense sign (%)	Pleural effusion (%)	Vascular abnormalities (%)	Lymphadenopathyb (%)	Chest wall invasion (%)
McAdams et al.[32]	United States	32a	HM	31a	65.6	40	15	NR	5	NR	13.7	NR	28.1	3.1	3.1	3.1
Jung et al.[20]	South Korea	24	HM	46	71	33	42h 29i	54	40	33	13	42	33	NR	NR	NR
Hammer et al.[10]	United States	30	HD	NR	57	10	67	60	53	NR	NR	NR	NR	NR	NR	NR
Peng et al.[34]	China	24	DM	71	83	33	50	4	12.5	NR	NR	NR	50	NR	29	4
Muthu et al.[18]	India	27	DM	44	48	48	11.1	14.8	0	NR	NR	NR	NR	NR	3.7	NR
Coste et al.[5]	France	114	HM	4	58	23	NR	26	24	NR	NR	NR	52	26g	NR	NR
Muthu et al.[16]	India	84	DM, COVID-19	69	13.1	66.7	2.4d	16.7e	2.4	16.7e	NR	NR	22.6	4.8f	NR	1.2
Yi et al.[35]	China	53	HD	NR	85j	3	85j	8	24	NR	NR	NR	23	NR	NR	NR
Jung et al.[21]	South Korea	34	HM	44	91	20.6	61.8	47.1	58.8	NR	14.7	55.9	NR	NR	NR	NR
Index study	India	193	DM	83	95.9	82.9	62	57	8	40	8.2	47	39.4	5.7f	22.8	3.1

BNS=Bird nest sign, CT=computed tomography, DM=Diabetes mellitus, HD=haematological diseases including haematological malignancies, aplastic anaemia and others, HM=haematological malignancies or hematopoietic stem cell transplants, RHS=reversed halo sign, NR=Not reported. The table presents the values of each observation as % of the total number of participants included (N) unless otherwise mentioned. ^aOf the 32 reported in the study, 29 had clinical details and 20 had CT. 9/29 were diabetes mellitus. One patient each had a pseudoaneurysm, lymphadenopathy, and chest wall invasion, ^bSignificant lymphadenopathy as reported by the studies. In the index study we defined significant lymphadenopathy if the short axis diameter was ≥10 mm, ^cPercentage includes cases with nodules >1 in number, ^dPercentage includes cases with nodules >10 in number, ^eIncludes cases with either RHS or BNS, ^f,gReported vascular abnormalities includes ^faneurysms only or ^ganeurysm/occlusions, ^hNodules ≥10mm and ⁱ<10mm were separately in this study, ^jIncludes cases with either nodules or consolidation

The temporal evolution of CT findings in PM has been best characterised in patients with neutropenia. Nam et al.[11] demonstrated that PM in immunocompromised hosts typically progresses through sequential morphologic patterns: an initial nodule/mass or consolidation with a halo sign, followed by a reversed halo sign, central necrosis, and finally an air-crescent sign, with these changes correlating with neutrophil recovery. The air-crescent sign, representing retraction of infarcted necrotic tissue surrounded by recovering neutrophils, typically appears as a late finding at the time of neutrophil recovery. In contrast to the acute presentation in organ transplant recipients or neutropenic hosts under close follow-up or active surveillance, patients with DAPM presented significantly later, allowing more time for disease evolution.

Ridgeline plots examining symptom duration distributions across CT features demonstrated temporal clustering patterns [Figure 5]. Features considered early in disease evolution (halo sign, large nodules) showed median symptom durations of 10–15 days, while features associated with chronicity (lymphadenopathy, bird’s nest sign, reversed halo sign) had median durations of 20–30 days. Cavitation and consolidation, the most common findings, showed broad distributions spanning the entire range of symptom durations, consistent with their occurrence across disease stages. The substantial overlap in distributions across all features suggests that while temporal trends exist, individual imaging signs cannot precisely determine disease duration. Presentation timing likely reflects multiple interacting factors, including host immunity (the indolent nature of poorly controlled diabetes mellitus), fungal burden and healthcare access. Prospective studies documenting the temporal sequence from symptom onset through serial imaging in DAPM patients are needed to confirm whether disease evolution truly differs from that observed in neutropenic hosts.

Multiple lung nodules were previously described as a key differentiating feature of PM from IPA in subjects with haematological malignancies.[14] Although nodules were commonly found in our PM cohort (62%), their characteristics differed substantially between risk factor groups. In PMOR, the nodules were generally larger (10–30 mm in 55%) and had a random distribution (75%), consistent with prior descriptions in neutropenic hosts. In contrast, most nodules in DAPM were small (1–10 mm in 62%) and had a centrilobular distribution (37%), a pattern not previously emphasised in the literature. This centrilobular pattern may reflect bronchiolar spread or aspiration of fungal elements, or the higher prevalence of haemoptysis at presentation (50.4% vs. 20.0% in PMOR), suggesting aspiration of blood. A previous small study of PM, primarily in patients with DM, similarly reported haemoptysis in >40% of participants.[34]

Significant lymphadenopathy (≥10 mm short axis) occurred more frequently in DAPM than PMOR (26.0% vs. 9.4%, P = 0.043). The significantly longer symptom duration in DAPM may explain the higher frequency of lymphadenopathy [Figure 5], which likely represents a chronic inflammatory response developing over the extended pre-diagnosis period. PMOR patients were likely diagnosed and imaged earlier, before substantial lymph node enlargement had developed. However, the lymph nodes in DAPM were generally modest in size and did not warrant diagnostic sampling.[18] The combination of cavitary lesions, centrilobular nodules and lymphadenopathy in DAPM raises critical diagnostic challenges in tuberculosis-endemic regions like India. This constellation of findings overlaps significantly with pulmonary tuberculosis (PTB), potentially leading to misdiagnosis and delayed antifungal therapy. The presence of reversed halo sign (RHS), observed in 59.8% of our DAPM cases, further complicates differentiation, as RHS has been described in both PM and PTB, including a ‘nodular’ form where the ring of consolidation is studded with small nodules.[36,37,38] In resource-limited settings where both conditions are prevalent, clinicians must maintain high suspicion for PM in diabetic patients with appropriate clinical and radiological features, even when imaging suggests PTB.

Beyond PTB, invasive pulmonary aspergillosis (IPA) represents another important differential diagnosis, particularly in neutropenic individuals. Airway-invasive features, such as centrilobular nodules, peribronchial consolidations and bronchial wall thickening, are more characteristic of IPA; whereas, the RHS is reported more frequently in PM than in IPA.[20,21,26] Future studies directly comparing imaging features of DAPM, culture-confirmed PTB, and IPA in diabetic patients would be valuable for refining diagnostic algorithms in high-burden settings.

Pulmonary artery pseudoaneurysms occurred in 15% of evaluated DAPM patients, with all 11 identified pseudoaneurysms occurring exclusively in the DAPM group. While angioinvasion is a pathologic hallmark of mucormycosis, the apparent exclusive occurrence of pseudoaneurysms in DAPM warrants careful interpretation.[39] Pseudoaneurysm formation may represent a more advanced stage of angioinvasive disease, potentially developing over time as fungal elements progressively invade and weaken vessel walls. PMOR patients were predominantly younger renal transplant recipients under active post-transplant surveillance, which facilitated early imaging, more aggressive diagnostic evaluation, and more frequent surgical management (37.5% vs. 13.7%). The exclusive occurrence of pseudoaneurysms in DAPM (15% of evaluated patients) is likely explained by the significantly longer symptom duration, allowing progressive angioinvasive disease to evolve from acute vascular invasion to chronic pseudoaneurysm formation. Similarly, chest wall involvement occurred in 6 of 98 evaluable patients, all of whom were in the DAPM group; however, limited imaging protocols precluded a comprehensive assessment in many patients.

The association between poorly controlled diabetes (HbA1c >9%) and increased frequency of bird’s nest sign (43% vs. 8–33% in lower HbA1c categories) is a novel finding. The bird’s nest sign represents a cavity containing internal nodular or linear opacities, thought to reflect necrotic debris, fungal elements, or devitalised tissue. The higher frequency in poorly controlled diabetes may indicate chronicity of disease, though the biological mechanisms require further investigation. Interestingly, haemoptysis also varied significantly across HbA1c categories (P = 0.007), suggesting that glycemic control influences not only imaging appearance but also clinical presentation.

Despite univariable associations between certain CT features (RHS, bird’s nest sign, larger consolidations) and 12-week mortality, none remained independently predictive after adjustment for clinical factors, including age, sex, HbA1c and surgical management. The lack of independent imaging predictors in our DAPM cohort suggests that underlying metabolic control, comorbidities, and treatment approach, rather than imaging appearance alone, drive survival. Combined medical-surgical management showed a trend toward lower mortality, suggesting that aggressive intervention may improve outcomes in selected patients.

Our study has several limitations. The retrospective design and single-centre setting may limit generalizability, as our tertiary care centre likely sees more advanced or complicated cases than community settings. We analysed only baseline CT examinations, precluding insights into the temporal evolution of findings. Although we documented symptom duration, we did not perform serial imaging to observe the actual temporal evolution of individual lesions. Imaging protocols varied, with some patients undergoing thin-section (1 mm) CT and others standard (10 mm) sections, which may have affected the detection of small nodules; however, the section thickness distribution did not differ significantly between DAPM and PMOR. Assessment of pseudoaneurysms and chest wall involvement was limited by imaging protocol, as not all patients underwent CT angiography or contrast-enhanced studies with adequate coverage. In addition, isolated tracheal or bronchial mucormycosis remains diagnostically challenging and may lack the characteristic radiological features described in this study, potentially leading to under-recognition of these presentations.[39,40,41,42] The strengths include the most extensive series of DAPM reported to date (n = 161), providing robust data on imaging manifestations in this understudied population. The inclusion of a comparison group (PMOR) and systematic analysis across glycemic control categories offers novel insights into how host factors influence imaging appearance. Documentation of symptom duration and exploratory temporal analysis [Figure 5] provide novel insights into disease evolution patterns.

CONCLUSION

Consolidation, cavitation and RHS are common in PM regardless of the underlying risk factor. However, DAPM presents significantly later, demonstrating distinct features, including predominantly small centrilobular nodules, a higher frequency of lymphadenopathy, and exclusive occurrence of pseudoaneurysms in our cohort. The association between HbA1c >9% and the bird’s nest sign suggests that glycemic control may influence imaging manifestations. Although certain CT features were associated with mortality on univariable analysis, none remained independently predictive after adjustment for clinical factors.

Data availability statement

The datasets generated and analysed during the current study are available from the corresponding author upon reasonable request. Individual patient data have been de-identified to protect participant privacy in accordance with institutional ethics requirements.

Ethics committee approval

IEC-INT/2023/STUDY-976.

Author contributions

NP- involved in radiology interpretation, data collection, drafted and revised the manuscript

MG- involved in radiology interpretation, data collection, and revised the manuscript for intellectual content

RA- involved in patient management, statistical analysis, supervised, and revised the manuscript

UD, ISS, SMR, NG, AB, HS, RR, SP, SM, AC, and VM were involved in patient management and manuscript writing. VM conceived the idea.

Conflicts of interest

There are no conflicts of interest to declare.

Funding Statement

Nil.