Key words ultrasound computed tomography (US/CT) - x-ray - invasive fungal infections - chest
- pneumonia
Introduction
Opportunistic fungal infections of the lung are a common complication in immunocompromised
patients and are related to a high morbidity and mortality [1 ]
[2 ]. Therefore, early and accurate diagnosis of invasive fungal infection (IFI) is important
to improve patient survival [3 ]
[4 ]. When neutropenia lasts more than 7 days, fungal infections may occur (aspergillosis,
candidiasis, mucormycosis) [5 ]
[6 ]
[7 ]. The clinical diagnosis of fungal pulmonary infection is based on host factors (i. e.,
immune status, neutropenic fever), clinical signs (unexplained fever despite broad-spectrum
antibiotics), microbiological evidence of infection, and specific patterns on computed
tomography (CT) imaging [1 ]
[8 ]
[9 ]
[10 ]. Clinical signs and symptoms related to IFI (i. e. cough, pleural pain, or hemoptysis)
are nonspecific and need to be followed up by appropriate diagnostic procedures as
part of an integrated care pathway consisting of daily physical examination, microscopy,
and mycological culture of all samples, serum, and bronchoalveolar lavage (BAL), Galactomannan
(GAL) antigen assay, serum (1–3)-β-D-Glucan (BDG), and high-resolution CT (HRCT) [11 ]
[12 ]
[13 ]
[14 ]. CT plays an important role in the diagnosis and management of patients with pulmonary
fungal infections due to its ability to depict disease at an early stage [15 ].
Chest CT has provided important clues regarding the clinical suspicion and early treatment
of invasive mold pneumonia in immunocompromised patients [16 ]. CT findings in fungal infections such as multiple nodules, ground glass opacities,
and (reverse) halo sign are due to parenchymal invasion of the fungus hyphae and/or
parenchymal infarctions due to invasion of the pulmonary vessels and are thus seen
earlier [15 ]. Legouge et al. reported that the reverse halo sign was present in 15 (94 %) of
16 patients with pulmonary mucormycosis during the first week of the disease, and
the authors proposed that the reverse halo sign could be a sensitive sign for pulmonary
mucormycosis, especially early in the disease [17 ]. However, this could not be confirmed by other authors, making the reverse halo
sign relatively nonspecific [18 ].
The aim of this study was 1) the systematic CT pattern mapping of all invasive fungal
infections of the lung in a current consecutive cohort, 2) the depiction of predictive
CT findings associated with a positive (regression) or negative outcome (progression
or death), 3) to assess the influence of the immune status of patients (neutropenia,
high-dose steroid therapy, organ transplant, other) and the infective agent on the
CT patterns.
Materials and Methods
Patient population
Chest CT images of 92 consecutive patients with invasive fungal infection were included
in this retrospective multicenter study. Informed consent was waived due to the retrospective
nature of the study. All adult patients at the three institutions (tertiary referral
hospitals) who met the criteria for mold infection – Aspergillosis (IPA, Invasive
Pulmonary Aspergillus), non-Aspergillus invasive mold infection (NAIMI, such as Mucormycosis
agents), Cryptococcosis (CRY) – according to the 2008 EORTC/MSG criteria as well as
patients with Pneumocystis jirovecii Pneumonia (PCP) (proven with immunofluorescence
only) were enrolled [10 ].
The inclusion criteria were a proven or probable fungal infection and a chest CT within
one week of the date of diagnosis. We excluded patients with a lower level of diagnostic
certainty (possible infection) and those with mixed pulmonary infections (i. e., bacterial
or viral co-infections) from our analysis. 85 patients met the criteria ([Fig. 1 ]): 39, 33, and 13 patients were recruited at centers 1, 2, and 3, respectively. 63
patients with proven invasive pulmonary fungus and 22 with probable invasive pulmonary
fungus were included.
Fig. 1 Patient flowchart.Abb. 1 Flussdiagramm der eingeschlossenen Patienten.
The study population was categorized by type of fungal infection, and four subgroups
depending on immune status were built: 1) Neutropenia: neutropenia at the time of
CT examination regardless of underlying disease (n = 44). Etiologies for neutropenia
were acute myeloid leukemia (AML) (n = 14), chronic lymphocytic leukemia (CLL) (n = 8),
non-Hodgkin’s lymphoma (NHL) (n = 5), acute lymphocytic leukemia (ALL) (n = 4), chemotherapy,
and other rare disease like multiple myeloma, plasmacytoma, and Hodgkin’s disease
(n = 13). 2) High-dose steroid therapy applied in cases of malignancies, organ transplant,
and different rheumatologic or autoimmune diseases (n = 36). 3) Organ transplant recipients
(OTR) (n = 11, with n = 7 kidney and n = 4 lung transplant). 4) All other underlying
diseases not fitting into one of the previous categories suffering from invasive fungal
pneumonia such as HIV infection (n = 3), malignancy (n = 1), COPD (n = 5), tuberculosis
infection (n = 1), and autoimmune disease (n = 2). If a patient matched two immune
status groups, the pulmonary findings counted in both groups.
Patients with IANA infection were subdivided into the 4 clinical outcome groups: 1)
29 with positive outcomes, 2) 4 with progressive disease, 3) 14 with death due to
other causes and 4) 6 with death due to infection; the remaining 7 patients were lost
to follow-up.
Imaging acquisition and interpretation
Chest CT studies at the time of initial diagnosis of mold infection were evaluated.
Images were acquired with 16, 64, and 128 row CT scanners using a tube voltage of
80–120 kV and automatic tube current modulation (with reference mAs between 80 and
120 mAs or similar noise levels). The slice reconstruction thickness was 1 or 2 mm.
IV contrast agent was applied in 36 of the 85 CT scans (e. g., for additional assessment
of pulmonary embolism or empyema). The following CT models were used (the number of
patients per scanner is indicated): Somatom Definition Flash (n = 28), Somatom Definition
AS+ (n = 13), Somatom Definition Edge (n = 13), Somatom Sensation 64 (n = 6) by Siemens
(Siemens Healthineers, Erlangen, Germany); GE Light Speed VCT (n = 12), GE Discovery
STE (n = 6), GE Bright Speed S (n = 2) by General Electrics (GE, Milwaukee, WI, USA);
Toshiba Aquillion (n = 3) by Canon (Canon Medical Systems Corporation, Otawara, Tochigi,
Japan) and Philipps Brilliance 64, (n = 2) by Philips (Philips, Best, The Netherlands).
Images were anonymized and randomized before radiological reading. All rating was
done on a digital Picture Archiving System (PACS) workstation. All cases were read
centrally by two independent teams of radiologists, with each team consisting of a
resident and a senior consultant thoracic radiologist with five and eight years of
experience, respectively. Discordant results between the two rating teams were resolved
by consensus read-out amongst the four radiologists. The readers were blinded to the
patient demographics and clinical outcomes. Lung patterns were classified as follows:
consolidation (±bronchogram, ±enhancement, ±halo sign, wedge-shaped peripheral infarction);
ground glass opacity (diffuse, patchy or mosaic); nodules (< 5 mm, 5–10 mm, > 10 mm,
halo sign, lobulated, spiculated, cavitary ±air crescent sign, calcified); interstitial
patterns (reticulation, centrilobular, perilymphatic nodules); and bronchial findings
(wall thickening, bronchiectasis, tree-in-bud, bullae, and cysts (see table in appendix))
[19 ]. Each of the five lobes was analyzed separately. The axial distribution was noted
as central predominant (inner 1/3 of the lungs), peripheral predominant (outer 2/3
of the lungs), or diffuse and ±subpleural sparing. Angioinvasivity was defined as
the presence of infarction and the halo or crescent sign. The probability of invasive
mold infection was estimated by the readers as < 25 %, < 50 %, < 75 %, and over 75 %
according to the presence of nodules with the halo sign in the upper lobes and the
previously mentioned signs of angioinvasivity. Furthermore, additional findings such
as lymphadenopathy (mediastinal or hilar nodes > 1 cm) and pleural effusions were
noted.
Clinical findings/diagnostic workup
Proof of invasive pulmonary fungus was established based on the identification of
fungal hyphae from BAL, needle aspiration, or in biopsy specimen. Probable IPA was
established with a combination of one host factor (< 500 neutrophils for more than
10 days, persistent fever above 38 °C, graft-vs.-host disease or use of steroids for
longer than 3 weeks) and one microbiological criterion (i. e., mold detection in culture
from BAL, sputum, or sinus aspirate specimens) and one major (or two minor) clinical
criteria [9 ]. The major factors included CT finding of halo sign, air crescent sign, cavity in
consolidation, whereas minor factors were the presence of cough, chest pain, hemoptysis,
dyspnea, pleural rub, pleural effusion, or consolidation [9 ]. Neutropenic patients with invasive fungal infections were defined as those patients
who received chemotherapy for acute leukemia, lymphoma, or aplastic anemia while having
an absolute neutrophil count < 500 cells/mm3 within 30 days prior to diagnosis of IFI [10 ].
The outcome after 6 weeks was classified according to Herbrecht et al. [20 ] and additional categories were implemented (death due to invasive pulmonary fungus
or other cause). Complete response was defined by the resolution of all clinical signs
and symptoms and more than 90 percent of the lesions due to invasive aspergillosis
that were visible on CT. Partial response was defined by clinical improvement and
> 50 % improvement in findings on radiology. Stable response was defined by the absence
of change from baseline or an improvement of < 50 %. Failure of therapy was defined
by worsening disease based on imaging or fatal outcome. For the statistical analysis,
the patients were divided into four groups: 1) positive outcome (complete, partial,
and stable response); 2) progressive disease 3) death due to causes other than infection;
and 4) death due to infection.
Statistical methods
IPA and NAIMI were pooled together into the group of invasive Aspergillus and non-Aspergillus
infections (= IANA). The prevalence and distribution of the chest CT patterns of IANA,
PCP, and Cryptococcus were analyzed individually. For these three fungal agents, a
separate pattern analysis was performed according to the patient’s immune status.
In addition, patterns were cross-referenced with a positive versus a negative outcome
(outcome 1 vs. 2–4). As PCP and Cryptococcus always had a favorable outcome, their
outcome was not subdivided. A separate pattern analysis was performed for all included
IANA cases versus only the proven IANA cases (probable infections were excluded).
The Chi square test or Fisher exact test was used when appropriate. The pattern and
distribution were analyzed per patient and per lobe. A Bonferroni correction was made
according to the number of patterns examined. For the analysis among the different
subgroups of immune status, a correction factor of 3 was applied for the main patterns
(consolidations, GGO, and nodules). All tests of significance were two-tailed, and
a p-value of < 0.05 was considered to indicate statistical significance. Calculations
were performed with MedCalc version 16.4.3 (MedCalc Software, Ostend, Belgium).
Results
Patient characteristics
The mean age of the 85 included patients (54 males) was 60 ± 15 years (range: 20 to
82 years). The total number of patients with IANA, PCP, and Cryptococcus was 60, 22,
and 3, respectively ([Table 1 ]). Immune status categorization revealed the following numbers per group 1) neutropenia
(n = 44), 2) steroid therapy (n = 36), 3) OTR (n = 11, 7 kidney and 4 lung). 4) other
diseases (n = 17).
Table 1
Clinical findings of the study populations.Tab. 1 Klinische Befunde der Studienpopulation.
IANA
all
IPA
NAIMI
PCP
CRY
age: mean
[range]
59.1
[18.3–89.3]
60.8
[18.3–89.3]
49.8
[20.3–61.1]
58.2
[22.2–89.3]
47.2
[42.1–53.7]
n, total
85
53
7
22
3
n, male
54
33
3
16
2
n, female
31
20
4
6
1
risk factors for IFI
1) neutropenia
44
29
7
8
0
14
9
3
2
0
4
3
1
0
0
8
4
1
3
0
5
3
1
1
0
13
10
1
2
0
26
20
5
1
0)
2) steroids
36
24
2
9
1
3) solid organ transplant
11
7
0
3
1
4) other
17
9
0
6
2
diagnostic certainty
IFI diagnosis proven
67
35
7
22
3
IFI diagnosis probable
18
18
0
0
0
no specific fungal therapy
61
36
2
20
3
specific fungal therapy
24
17
5
2
0
fluconazole
11
11
0
0
0
voriconazole
3
2
1
0
0
posaconazole
8
3
4
1
0
caspofungin
1
1
0
0
0
bactrim forte
1
0
0
1
0
IANA = invasive Aspergillus and non-Aspergillus infection, IPA = invasive pulmonary
aspergillosis, NAMI = non-Aspergillus invasive mold infection, PCP = Pneumocystis
jirovecii Pneumonia, CRY = Cryptococcus; AML = acute myeloid leukemia, ALL = acute
lymphatic leukemia, CLL = chronic lymphatic leukemia, NHL = non-Hodgkin lymphoma,
other = other hematologic disease, solid organ transplantation = status post-transplantation
of kidney or lung, IFI = invasive fungal infection.
Stratified according to immune status – per patient
Consolidations
Overall, the consolidation pattern was seen in 65.9 % of cases in the neutropenic
group, followed by the steroid group with 44.4 %, OTR with 27.3 %, and the group of
various diseases with 52.9 % ([Fig. 5 ]). OTR vs. neutropenia was significant with a p-value of 0.045. In patients under
steroid therapy, patients suffering from PCP demonstrated a significantly lower incidence
of consolidations (11.1 %) compared to IANA patients (57.7 %, p-value = 0.045).
Fig. 5 Prevalence of radiological patterns.Abb. 5 Prävalenz der radiologischen Muster.
GGO
There was no significant difference in GGO incidence among the various groups of immune
status (54.5–76.5 %, [Fig. 5 ]). However, patients with PCP demonstrated generally more GGO than patients witch
IANA in the neutropenic group: 100 % vs. 55.6 % (p-value = 0.049).
Nodules
There was a high incidence of pulmonary nodules. Each immune status group showed at
least one nodule in > 72.7 % of the patients. In the neutropenia group, the invasive
fungi patients had significantly more nodules (94.4 %) compared to the PCP group (50.0 %,
p = 0.002). In the other groups, the results were similar but did not reach statistical
significance. Cavernous nodules were found in 11.4 %, 19.4 %, 18.2 %, and 23.5 % of
the patients in the neutropenia, steroid, OTR, and various diseases groups, respectively
(p > 0.05). Generally, the frequency of large cavernous nodules was higher at 7.3 %.
Intermediate and small cavernous nodules were present in only 3.3 % and 2.3 %, respectively
(appendix table). The typical crescent sign in angioinvasive fungal pneumonia was
only found in 2 IANA patients (3.3 %), and both had a favorable outcome. The halo
sign was observed in 56.7 %, 27.3 %, and 0 % of cases of IANA, PCP, and CRY, respectively.
Lymph nodes, pleural effusion
The incidence of lymphadenopathy was generally low (< 18 %) among the different groups.
Also, the incidence of pleural effusion was only < 25 %. Both findings did not differ
significantly among the groups. Detailed results for all imaging patterns are demonstrated
in the appendix table stratified by infective agent and lobar location.
Stratified according to immune status – per lobe
There was no significant difference among the immune status groups regarding lobe
predilection ([Table 2 ]). The signature sign for invasive pulmonary fungal infection was the solid nodule
with the halo sign. In general, this nodule was found more often in the upper lobes
(p < 0.001). The lower incidence for this nodule for organ transplant recipients is
not significant.
Table 2
Prevalence of solid nodules with positive halo sign per lobe.Tab. 2 Prävalenz von soliden Knoten mit positivem Halo-Zeichen pro Lappen.
immune status
neutropenia
steroids
organ TX
other
right upper lobe
43.2 %
33.3 %
18.2 %
23.5 %
middle lobe
22.7 %
22.2 %
9.1 %
5.9 %
right lower lobe
29.5 %
19.4 %
9.1 %
5.9 %
left upper lobe
36.4 %
36.1 %
9.1 %
17.6 %
left lower lobe
22.7 %
22.2 %
9.1 %
5.9 %
Prevalence is demonstrated as percentage per lobe and per immune status.
Stratified according to infective agent
Lung pattern frequency in IANA patients
IANA patients demonstrated significantly more nodules (93.3 %) than GGO (58.3 %) and
consolidations (56.7 %) (p < 0.001). Solid nodules with the halo sign and ground glass
nodules (GGN) were more common than cavitary nodules ([Fig. 2 ], [3 ]): 32.7 %, 28.3 %, and 25.3 %, versus 9.0 % (p < 0.001). In patients with IANA infection,
4.7 % of the lung lobes showed a wedge-shaped peripheral consolidation (infarction,
[Fig. 4 ]) compared to 0 % in PCP (p = 0.1).
Fig. 2 Patient examples of invasive Aspergillosis and Pneumocystis jiroveci. A 66-year-old
female patient with invasive Aspergillosis is illustrated in the top row A, B . Multiple solid, ground glass, and cavernous nodules (arrow) are visible. She died
due to the mold infection. In the lower row C, D , CT images of a 58-year-old male patient with Pneumocystis jiroveci (PCP) infection
are shown. The subpleural sparing of the mainly ground glass changes was only found
in PCP cases.Abb. 2 Invasive Aspergillose und Pneumocystis-jiroveci-Infektion. Die obere Reihe A, B zeigt Bilder einer 66-jährigen Frau mit invasiver Aspergillose. Multiple Knoten,
teils solide, teils milchglasartig, oder auch eingeschmolzene Knoten (Pfeil) sind
sichtbar. Sie starb an den Folgen der Pilzinfektion. In der unteren Reihe C, D werden CT-Bilder eines 58-jährigen männlichen Patienten mit Pneumocystis-jiroveci
(PCP) -Infektion gezeigt. Die subpleuralen Aussparungen der milchglasartigen Lungenverdichtungen
waren nur bei PCP-Patienten zu sehen.
Fig. 3 59-year-old female patient suffering from chronic lymphatic leukemia and Mucor pneumonia.
Axial CT slice showing a single nodule with a halo sign in the right upper lobe.Abb. 3 59-jährige Patientin mit chronisch lymphatischer Leukämie und Mucor-Pneumonie. Die
axiale CT-Schicht zeigt einen einzelnen soliden Pilzknoten mit peripherem Milchglassaum
(Halo-Zeichen) im rechten Oberlappen.
Fig. 4 Angioinvasive Aspergillus infection of the lung in a patient with acute myeloid leukemia.
The right upper lobe demonstrates a hemorrhagic peripheral infarction (wedge shape)
and a posterior nodule with a halo sign due to invasion of the pulmonary vessels.Abb. 4 Angioinvasive Aspergillus-Pneumonie bei akuter myeloischer Leukämie. Dreieckiger
peripherer Lungeninfarkt und dorsaler Knoten mit Halo-Zeichen, verursacht durch die
Gefäßinfiltration der Pilze.
Lung pattern frequency in PCP patients
Patients with PCP showed GGO in 95.5 % of cases ([Fig. 2 ]), nodules in 59.1 %, and consolidations in 45.5 % (p < 0.004). GGO was thus present
significantly more often in patients with PCP compared to IANA and CRY (95.5 %, 58.3 %,
and 33.3 %, respectively, p < 0.009). Subpleural sparing was found only in patients
with PCP (22.7 %, [Fig. 2 ]).
Lung pattern frequency in CRY patients
All CRY patients demonstrated nodules, while 33 % had consolidations and 33 % had
GGO (nodules vs. consolidations/GGO: p = 0.083).
Pattern prevalence according to the clinical outcome (IANA patients only)
Patients who died from the IANA infection showed GGO in 83.3 % of the cases (corresponding
to a 31.6 % higher prevalence than patients with a positive outcome; p-value = 0.44),
and a diffuse pattern distribution was more frequent (in 33.3 %) compared to patients
with positive outcomes (in 0 %, p = 0.005). In a per lobe analysis, we found that
IANA patients with negative outcomes (progression or death) demonstrated significantly
more nodules with the halo sign. 42.5 % of the lobes were affected compared to 15.9 %
of the lobes in patients with a positive outcome (p < 0.0001). Patients with a negative
outcome demonstrated especially smaller nodules with the halo sign. Small nodules
(< 5 mm), intermediate nodules (5–10 mm), and macronodules (> 10 mm) with the halo
sign affected 20.8 %, 19.2 %, and 15.8 % of the lobes in patients with a negative
outcome compared to 6.9 %, 5.5 % and 8.3 % of the lobes in patients with a positive
outcome (p = 0.007, 0.005 and 0.62). Patients who died from fungal infection demonstrated
the highest prevalence of intermediate nodules with the halo sign: 26.7 % (p = 0.002
compared to positive outcomes). Furthermore, patients with positive outcomes showed
more pulmonary infarctions (peripheral, wedge-shaped consolidations) than patients
with negative outcomes (8.3 % vs. 0.8 %, p = 0.0466).
The frequencies for the presence of each lung pattern per patient are shown in [Fig. 6 ] and per lobe in [Fig. 7 ].
Fig. 6 Presence of each pattern in patients with IANA, PCP, and CRY in a per patient analysis.Abb. 6 Präsenz der jeweiligen CT-Muster in Patienten mit IANA, PCP und CRY in einer patientenbasierten
Analyse.
Fig. 7 Presence of each pattern in patients with IANA, PCP, and CRY is shown per lobe.Abb. 7 Präsenz der verschiedenen Muster in Patienten mit IANA, PCP und CRY in einer lappenbasierten
Analyse.
Angioinvasivity, probability of fungal infection, and rare patterns
The probability of fungal pneumonia estimated by the reading radiologists based on
previously described imaging criteria was observed with equal distribution in 26.4 %,
26.4 %, 25.3 % and 21.9 % of patients in the neutropenia, steroid, OTR, and various
diseases groups, respectively, although signs of angioinvasivity were found in 31.0 %.
Overall, the radiologists judged the probability of fungal infection in the IANA patients
as high (> 50 %) and for the PCP and CRY cases as low (< 50 %, p = 0.023). Subpleural
sparing, solitary nodules, or the crescent sign was seen in 5.75 %, 10.34 %, and 0.66 %
of the lobes, respectively. Other observed lung patterns that are not typically known
for fungal infections were reticulation, bronchiectasis, and centrilobular bullae
in 17.93 %, 11.26 % and 6.90 %, respectively, per lobe.
Distribution within the lungs
There was a trend for fungal infections affecting the upper lobes more than the lower
lobes (p = 0.0852). The following patterns demonstrated a significantly higher involvement
of the upper lobes in IANA patients: mosaic GGO, nodules with halo sign (> 10 mm),
and cavitary nodules (> 10 mm) ([Fig. 7 ]). We observed fungal lung changes significantly more often in the periphery (central
vs. peripheral predominant, p = 0.0035).
CT pattern dependency on the fungus test confidence levels (all versus proven IANA
infections)
The prevalence of consolidations, GGO, and nodules for all IANA cases (proven and
probable) was 56.7 %, 58.3 %, and 93.3 %, respectively, and dropped insignificantly
to 55.8 %, 48.1 %, and 86.5 %, respectively, for the proven infections only (all p-values
> 0.285). Detailed data regarding prevalence and dependencies on immune status are
given in [Table 3 ]. Also, the per lobe analysis and the results for the different outcomes did not
demonstrate differences in CT pattern prevalence for different test confidence levels.
Table 3
Incidences of radiological patterns dependent on the fungus test confidence level.Tab. 3 Inzidenzen der radiologischen Muster in Abhängigkeit von der diagnostischen Sicherheit
des Pilztests.
consolidation
GGO
nodules
immune status
neutropenia
steroids
organ TX
other
all
neutropenia
steroids
organ TX
other
all
neutropenia
steroids
organ TX
other
all
proven and probable IA/NAIMI1
66.7 %
57.7 %
42.9 %
33.3 %
56.7 %
55.6 %
57.7 %
42.9 %3
66.7 %
58.3 %
94.4 %
88.5 %
85.7 %
88.9 %
91.3 %
proven IA/NAIMI2
69.6 %
52.9 %
25.0 %
37.5 %
55.8 %
52.2 %
47.1 %
0 %4
62.5 %
48.1 %
91.3 %
82.4 %
75.0 %
87.5 %
86.5 %
all p-values: all1 vs. proven2 > 0.285
3 n = 3 of 7
4 n = 0 of 4
IA: invasive aspergillosis; NAIMI: Non-aspergillus invasive mold infection; GGO: ground
glass opacity; TX: transplantation.
Discussion
The data from this multicenter study show that nodules, GGO, and consolidations are
common CT findings in all mold infections. Thus, there was no sign that was unique
for one specific pathogen. However, nodules were more predominant in IANA and CRY
and GGO was most often correlated with PCP. Solid nodules with the halo sign < 10 mm
were associated with a worse outcome in our population.
Our data are consistent with a study from Xu et al. who compared imaging findings
in IPA patients with and without hematological malignancies [21 ]. While in non-hematological patients, the airway-invasive form, manifesting ground-glass
opacity and consolidation or mass formation, was more common, while in patients with
hematological malignancy, the angioinvasive form with macronodules (74.5 %) and halo
signs (50 %) was more frequent. In the present study, the most common radiologic findings
in patients with IANA were nodules (93 %). Nodular lesions were the most common finding
on CT in other studies too, and their absence argues against the likelihood of IPA
[22 ]. For example, nodules were seen in 74 % of neutropenic patients in a study by Militio
et al. [23 ]. For most types of nodules (GGN, solid nodules with the halo sign), we observed
a higher frequency of small (approximately 20 %) and intermediate size (10 %) nodules
than macronodules (2 %). The only exceptions were cavernous nodules with the highest
frequency of macronodules (7 %). This might be explained by: a) our definition of
size criteria (macronodules were defined as nodules > 10 mm, compared to other studies
defining macronodules as > 3–5 mm); and b) our inclusion criteria, because we evaluated
early CT scans within the first week of diagnosis, and it is known that nodule size
increases during the first 10 days of infection [24 ].
Whereas it is widely acknowledged that the halo sign is seen frequently in patients
with early-stage IPA, not all authors in the literature have reported the same high
frequency [18 ]. In fact, the incidence of the halo sign among neutropenic patients with hematological
malignancies has varied widely, ranging from 25 % to 95 % [8 ]
[25 ]
[26 ]
[27 ]
[28 ]
[29 ]
[30 ]
[31 ]. The halo sign was observed in 56.7 %, 27.3 %, and 0 % of cases of IANA, CRY, and
PCP, respectively. Although nonspecific, in severely neutropenic patients, the halo
sign is highly suggestive of IFI [32 ]
[33 ]. In addition, the halo sign appears to occur less frequently in patients without
neutropenia who have pulmonary aspergillosis than in those with neutropenia, particularly
when they are receiving corticosteroid therapy [34 ], according to our results ([Table 2 ]).
After the initiation of antifungal therapy, fragments of infarcted lung may separate
from the adjacent parenchyma (pulmonary sequestra), resulting in a cavity with an
air crescent (air crescent sign) [32 ]. As this takes some time to develop in the course of aspergillosis [35 ], we only observed two cases of air crescent sign at this early CT exam time point.
As the air crescent sign is a sign of healing [28 ]
[35 ], it is not surprising that those two patients had a favorable outcome. Other CT
findings described in the literature are pleura-based, wedge-shaped areas of consolidation
[25 ]
[36 ]. These findings correspond to hemorrhagic infarcts, and we observed them in the
IANA group in up to 5 % of cases.
We observed a significantly higher involvement of the upper lobes in IANA patients
for mosaic GGO, nodules with the halo sign, and cavitary nodules. Similar findings
were noted by Xu et al. [21 ]. In our study, the nodular pattern was registered significantly more often in the
upper lobes. In contrast, Militio et al. showed an equal distribution between the
different lobes of the lungs, with nodular lesions in 81 % of the upper lobes, 81 %
of the lower lobes, and 63 % of the middle lobe or lingula [23 ]. Similar to their study, we observed nodular lesions to be predominantly peripheral
and multiple.
Our findings suggest that certain radiologic findings such as infarctions are associated
with a favorable outcome and others like GGO with a poor clinical outcome after 6
weeks. In contrast to the literature [37 ], we cannot confirm the favorable outcome in patients with the halo sign in our cohort.
Nodules with the halo sign (< 1 cm) were seen in 20 % of patients who died due to
the fungal infection and in 15 % of patients with a negative outcome after 6 weeks
as compared to 6 % of patients with a positive outcome. Additionally, larger nodules
with the halo sign were more frequent in patients with a negative outcome, but the
difference was not significant. Park et al. [22 ] also suggested that macronodules and multiple infarct-shaped consolidations have
prognostic implications in patients with IPA. In a study from Singh et al., lung transplant
recipients with IPA had better outcomes if nodular lesions were present [38 ].
Geltner et al. [39 ] have shown that the most common infectious agent in transplant recipients is Aspergillus
fumigatus (67 %). This is in concordance with our results, which showed 65 % of OTR
patients with IANA, 27 % with PCP and 9 % with Cryptococcosis. Hekimoglu et al. [40 ] evaluated CT patterns in a group of 15 patients post-solid organ transplantation
with proven diagnosis of invasive pulmonary aspergillosis. The most frequently observed
patterns in their population were nodules and GGO which is in concordance with the
results of our study as well as with data from Qin et al. [41 ]. However, Qin et al. observed nodules in 64 % of cases and masses in 36 %.
Park et al. studied CT patterns in two patient groups and found similar findings in
patients with neutropenia (n = 60) and non-neutropenic organ transplant recipients
(n = 60) suffering from IPA [22 ]. In their study, consolidations or masses, halo signs, and an angioinvasive form
were observed less often in non-neutropenic transplant recipients than in neutropenic
patients (56 %, 26 % and 32 % vs. 78 %, 55 %, 60 % with p = 0.01, p = 0.002 and p = 0.003,
respectively). This is consistent with our observations, although our groups were
a little different with consolidations, GGO, and nodules being present in 67 %, 56 %,
and 94 % of IPA patients with neutropenia and in 43 %, 43 %, and 86 % of OTR patients.
It is also known that cavitary formation only happens late in patients with neutropenia
when aplasia is resolving. It is an early sign in patients undergoing steroid treatments.
This was shown with 11 % of neutropenic patients with cavernous nodule transformation
and 19 % of patients undergoing steroid therapy. Although the follow-up of such patients
was not included in the scope of the present study, exemplary CT exams of some patients
have shown the development of caverns as a sign of immune reconstitution.
Our study has several limitations. First, our population contains a mixed group of
different fungal species (IANA n = 60, PCP n = 22, and CRY n = 3). This reflects the
daily routine for clinical radiologists when they get a CT order with the suspicion
of opportunistic infection but without confirmation or specific diagnosis. The reading
radiologist thus needs to know the criteria for invasive infections to enable further
diagnostic and therapeutic steps. Due to the strict Bonferroni correction of factor
33 (for all patterns) in the per lobe analysis, the differences had to be huge to
be significant between different immune status groups. Some of the important patterns
with marginally significant differences may have suffered from the inclusion of rather
unimportant patterns. However, for the sake of completeness, we preferred to include
all patterns. Furthermore, due to the retrospective cross-sectional design of our
study, potential confounders were not controlled and follow-up of imaging patterns
was not within the scope of this study and can be investigated in the future.
Conclusion: Patients with IANA showed a higher prevalence of nodules and a lower prevalence
of ground glass opacities than patients with PCP. All patients with PCP and CRY had
a favorable outcome. In patients with IANA, nodules with the halo sign were associated
with an adverse outcome. Patients with neutropenia generally showed more consolidations,
but consolidations were not associated with an adverse outcome.
Chest CT gives important clues for the clinical suspicion and early treatment of invasive
mold pneumonia in immunocompromised patients
This study with a systematic analysis of imaging patterns improves the understanding
of CT patterns in invasive pulmonary mold infections.
Smaller (< 5 mm) and intermediate (5–10 mm) nodules with the halo sign were more often
associated with a negative outcome in IANA patients.
Funding sources
This work was supported by the Fungal Infection Network of Switzerland (FUNGINOS).
