Professional Documents
Culture Documents
DOI: 10.1111/myc.13012
ORIGINAL ARTICLE
1
Division of Infectious Diseases, University
of Michigan, Ann Arbor, MI, USA Abstract
2
Infectious Diseases Section, Veterans Background: The diagnosis of invasive pulmonary aspergillosis (IPA) remains challeng‐
Affairs Ann Arbor Healthcare System, Ann
ing. We evaluated the performance characteristics of a newly formatted Aspergillus
Arbor, MI, USA
lateral flow test, AspLFD, in bronchoalveolar lavage (BAL) fluid from patients with
Correspondence
classic risk factors for IPA.
Marisa H. Miceli, Division of Infectious
Diseases, University of Michigan Health Methods: Prospectively banked BAL samples from 14 patients with proven or prob‐
System, F4005 University Hospital South,
able IPA defined by EORTC/MSG criteria and 28 BAL samples from age‐matched
1500 E. Medical Center Dr. Ann Arbor, Ann
Arbor, MI 48109, USA. high‐risk patients without IPA were tested with AspLFD according to manufacturer's
Email: mmiceli@med.umich.edu
directions. Results were read by two independent observers, and test performance
Funding information was calculated.
The Veterans Education and Research
Results: Age, gender and underlying risk factors, except for neutropenia and haema‐
Association of Michigan
tological malignancy, were similar between IPA cases and controls. Seven patients
(50%) in the IPA group received a mould‐active agent within 5 days prior to bron‐
choscopy compared with only three patients (11%) in the control group, P = .004. Of
14 patients with proven/probable IPA, AspLFD was positive in 3 and negative in 9;
two tests yielded invalid results. All 28 control patients had a negative AspLFD test.
AspLFD showed low sensitivity (25%, 95% CI: 5.5% to 57.2%), but high specificity
(100%. (95% CI: 87.7% to 100%).
Conclusions: A positive AspLFD test in BAL fluid of patients with classic risk factors
for IPA could be useful to support the diagnosis of proven/probable IPA because of
its high specificity. However, as a stand‐alone test for IPA, the use of AspLFD is lim‐
ited by low sensitivity.
KEYWORDS
bronchoalveolar lavage, diagnostics, invasive aspergillosis, lateral flow device
1 | I NTRO D U C TI O N shown increased sensitivity and specificity for IPA in patients who
have haematological malignancies and in those who have received a
Diagnosis of invasive pulmonary aspergillosis (IPA) is challenging, hematopoietic cell transplant (HCT).1-3
and delays in diagnosis contribute to increased mortality. Non‐in‐ The Aspergillus lateral flow device (AspLFD) is a newer non‐cul‐
vasive fungal biomarkers, such as the serum Aspergillus galactoman‐ ture‐based diagnostic test that utilises mouse monoclonal antibody
nan (GM) assay, have proved useful in the diagnosis of IPA.1,2 When JF5 to detect an extracellular glycoprotein secreted during active
performed on bronchoalveolar lavage (BAL) fluid, the GM assay has growth of Aspergillus species.4 This assay has been developed as
a point of care test that can be performed on serum or BAL fluid. matching; but in some cases, matching was accomplished by using
Previous studies from our laboratory using a prototype lateral flow other classic risk factors.
device (LFD) demonstrated high specificity (94%) but low sensitivity The newly formatted AspLFD was used in accordance with the
(38%) on BAL samples from a population at high risk for IPA.5 AspLFD, manufacturer's protocol (OLM Diagnostics). BAL samples were
the newly formatted assay that has been approved for use in Europe, thawed, vortexed and centrifuged for 1 minute at 14 000 rpm, fol‐
6
was developed to enhance the performance of the LFD test. A lowing which 70 μL of the supernatant was applied to the device.
multicentre study of haematology patients using the AspLFD found After fifteen minutes, results were read independently by two in‐
a sensitivity of 71% and a specificity of 85% when comparing cases vestigators blinded to the patient's disease status and were scored
of proven or probable IPA with control patients without IPA.7 We as negative or positive; positive results were graded as +, ++ and
sought to determine the test performance of the AspLFD in BAL fluid +++ (Figure 1). If results were discordant, a third blinded investigator
from a broad population of patients with classic risk factors for IPA. was consulted as a tiebreaker. Results were considered invalid if the
control line was not positive after two separate runs. Of note, no
specimen required pretreatment per package insert instructions.
2 | M ATE R I A L S A N D M E TH O DS We calculated the sensitivity and specificity of the AspLFD for
the diagnosis of proven/probable IPA. Differences in categorical
This study was performed at the University of Michigan Health variables or continuous numbers were analysed by Fischer's exact
System, and approval for the study was obtained from the University test or unpaired t test. All statistical analyses were completed using
of Michigan Institutional Review Board. Leftover BAL specimens SPSS software, version 25.0 (SPSS, Inc).
from all bronchoscopies performed on adult patients between
11/2015 and 2/2018 were collected prospectively. BAL fluid was
stored at −70°C in an established BAL repository in the Infectious 3 | R E S U LT S
Diseases Laboratory at the VA Ann Arbor Healthcare System.
The medical records of all patients whose BAL fluid was collected A total of 795 patients had samples stored in the BAL repository and
were reviewed to identify adult patients ≥ 18 years of age with classic clinical data recorded in the REDCap system. Sixteen of these pa‐
risk factors for IPA. Classic risk factors were defined as haematolog‐ tients had IPA: two proven, 12 probable and two possible. The latter
ical malignancy, neutropenia (absolute neutrophil count ≤ 500 cells/ two patients were excluded from further analysis.
μL for ≥ 10 days at time of BAL), solid organ or HCT recipient, inher‐
ited severe immunodeficiency, HIV with CD4 count ≤ 200 cells/μL,
use of T‐cell immunosuppressive agents within 90 days of BAL or use TA B L E 1 Demographics of patients with invasive pulmonary
aspergillosis (IPA) and control patients who had bronchoalveolar
of corticosteroids at a dose equivalent to ≥ 0.3 mg/kg of prednisone
lavage samples tested with AspLFD
daily for ≥ 3 weeks at time of BAL. Receipt of a mould‐active antifun‐
gal agent within 5 days prior to bronchoscopy was noted. Previously IPA cases Controls
(n = 14) (n = 28)
obtained results from serum and BAL Aspergillus GM (PlateliaTM
Aspergillus EIA, Viracor‐IBT Laboratories) and respiratory fungal cul‐ Number (%) Number (%) P‐value
tures were recorded. GM results with optical density index (ODI) Mean age (years + SD) 52.7 ± 17.5 53.6 ± 15.6 .86
values ≥ 0.5 were considered positive. All data were entered into a
Female 6 (43) 8 (29) .35
REDCap research database.
Classic risk factors for IPA
The diagnosis of IPA was determined using EORTC/MSG cri‐
T‐cell depleting agent 9 (64) 20 (71) .64
teria; cases were categorised as proven, probable, possible or no
Solid organ transplant 4 (29) 14 (50) .19
IPA.8 Patients with proven or probable IPA were matched 1:2 by
High dose corticosteroidsa 4 (29) 11 (39) .49
age within 5 years and by risk factors with controls who did not
have IPA. Whenever possible, the same risk factors were used for Haematological malignancy 6 (43) 3 (11) .02
Neutropeniab 3 (21) 0 (0) .01
Allogeneic HCT 0 (0) 2 (7) .31
Additional features
Anti‐mould exposurec 7 (50) 3 (11) .004
Age/Sex Host Factors Mould‐active agenta IPA BAL/sputum culture BAL GMb Serum GMb BAL AspLFD
60M T‐cell immunosuppression Voriconazole Probable A fumigatus 1.36 Not done Positive (+++)
69F Neutropenia, corticosteroids, T‐cell None Probable Negative 2.80 Negative Positive (+)
immunosuppression
79F Haematological malignancy Voriconazole Probable Aspergillus spp. Not done Negative Positive (+)
c
42M Corticosteroids (Cushing's syndrome) Voriconazole, Liposomal ampho‐ Proven Negative 0.08 Negative Negative
tericin B
32F Corticosteroids, T‐cell None Provenc A fumigatus 0.68 Negative Negative
immunosuppression
60M Haematological malignancy, T‐cell None Probable A fumigatus 0.21 Not done Negative
immunosuppression
22M Neutropenia, Haematological malignancy Voriconazole, Micafungin Probable Negative 6.21 Negative Negative
42M Lung Tx, T‐cell immunosuppression None Probable A fumigatus 0.15 Not done Negative
62M Lung Tx, T‐cell immunosuppression None Probable A fumigatus 0.12 Not done Negative
23F Lung Tx, T‐cell immunosuppression None Probable A fumigatus 0.21 Negative Negative
66M Haematological malignancy, T‐cell Isavuconazole Probable Negative 0.11 1.75 Negative
immunosuppression
63F Lung Tx, T‐cell immunosuppression Voriconazole Probable A fumigatus (sputum) 0.14 Negative Negative
58F Corticosteroids None Probable Negative 7.6 Negative Invalidd
60M Neutropenia, Haematological malig‐ Voriconazole, Liposomal ampho‐ Probable Negative 2.3 Negative Invalidd
nancy, T‐cell immunosuppression tericin B
Abbreviation: AspLFD, Aspergillus lateral flow device; GM, galactomannan; Tx, transplant.
a
Patients received a mould‐active antifungal agent within 5 days prior to bronchoscopy.
b
Expressed as optical density index (ODI); values ≥ 0.5 ODI defined as positive.
c
Autopsy‐proven aspergillosis.
d
Results were considered invalid if the control line was not positive on two separate tests.
|
3
|
4 LINDER et al.
The mean age of the 14 patients with proven/probable IPA was LFD assay. 5,17,18 Additionally, the prototype LFD assay performed
52.7 ± 17.5 years, and 6 (43%) were female. Of the 28 control pa‐ better in studies evaluating its use in patients who had received a
tients without IPA, the mean age was 53.6 ± 15.6 years, and 8 (29%) solid organ transplant than in patients with haematological malig‐
were female. Comorbid conditions and risk factors for IPA were sim‐ nancies, which has been postulated to be due to the use of prophy‐
ilar for both groups except for haematological malignancy and neu‐ laxis with mould‐active antifungal agents in the latter group.18 One
tropenia, which were more common in patients with IPA (Table 1). study has addressed this issue using the newly formatted AspLFD
Seven patients (50%) in the IPA group were on a mould‐active anti‐ assay.7 In the subgroup of patients who had received empirical
fungal agent within the 5 days prior to bronchoscopy compared with mould‐active agents prior to AspLFD testing, the sensitivity of the
only 3 patients (11%) in the control group, P = .004. AspLFD assay decreased from 71% to 47%. In our study, 50% of pa‐
Among the 14 patients with proven/probable IPA, BAL fluid in 7 tients with proven/probable IPA had received a mould‐active anti‐
patients and sputum in 1 patient yielded Aspergillus species; six pa‐ fungal agent within the 5 days prior to the collection of BAL fluid;
tients had a BAL GM ≥ 0.5 ODI, 5 of which were ≥ 1.0 ODI (Table 2). this may have led to decreased sensitivity of the AspLFD assay.
computarised tomography scans were performed in 13 patients; A similar phenomenon has been described previously for the GM
nodules were present in 10 patients, tree‐in‐bud changes in 3 and a assay and for PCR.19-21
halo sign, a cavitary lesion and a mycetoma in one patient each. One Similar to other qualitative tests, interpretation of AspLFD results
patient, who expired and had autopsy‐proven IPA, did not have a is operator dependent. In our study, overall concordance among in‐
scan performed before death. vestigators was high, but Inter‐grader variability could contribute to
For the 14 patients with proven/probable IPA, AspLFD was pos‐ differences in performance that have been noted among different
itive in 3 and negative in 9 (Table 2). Four of the 7 patients on a studies. A recent large multicentre study used a digital reader in
mould‐active agent had negative AspLFD test results, and 2 had a addition to visual assessment of test results and noted use of the
positive AspLFD test result. Two patients with probable IPA had in‐ reader enhanced the ability to detect lines.7 The use of such a digital
valid AspLFD results because the control line did not appear, and reader should be explored further.
these results were excluded from further analysis. All 28 control pa‐ Some authors have suggested that a dual testing approach com‐
tients had a negative AspLFD result. Inter‐grader concordance was bining an LFD assay and a GM assay could improve sensitivity for the
high; an adjudicator was needed for only three samples. diagnosis of IPA.5,22,23 Studies using this combined testing approach
Sensitivity of the AspLFD in BAL fluid was 25% (95% CI: 5.5% to with the prototype LFD and GM are contradictory, with some show‐
57.2%), and specificity was 100% (95% CI: 87.7% to 100%). ing increased sensitivity and others not. One group used this ap‐
proach with the newly formatted AspLFD and reported enhanced
sensitivity.15
4 | D I S CU S S I O N Limitations of this study include the single‐centre retrospective
nature and the small number of patients with proven or probable
The AspLFD is an attractive option for testing for IPA, as it can be IPA. Our samples had been frozen at −70°C before use, and al‐
done quickly as a point of care test and can be used with either though, unlikely, this possibly could have decreased the sensitivity
serum or BAL fluid. However, test performance has varied among of the AspLFD test.
different laboratories. Initial studies showed promising results using Benefits of the AspLFD include the ease of use and high inter‐
a prototype LFD.5,9-13 These studies each involved different patient observer agreement regarding endpoints, which are important for
populations, but in general, the prototype LFD assay was noted to point of care tests. A positive test is highly specific for IPA, but sen‐
have specificity > 90%. However, sensitivity varied from 38% to sitivity is low, which would not allow its use as a stand‐alone test for
94%. Reasons for this wide discrepancy could, in part, be related to IPA. Further studies are warranted to determine the ultimate role of
different populations that were studied and to variations in study this assay in the clinical setting.
design. For instance, in several studies, a modified definition of IPA
was used to enable comparison of the LFD with other biomarkers,
AC K N OW L E D G E M E N T S
such as GM.5,11
The newly formatted AspLFD was developed to improve the per‐ This study was supported by the Veterans Education and Research
formance of the prototype LFD assay. Five prior reports using the Association of Michigan.
AspLFD noted sensitivity that varied from 46% to 78% and speci‐
ficity that varied from 66% to 100%.6,7,14-16 When using the AspLFD
C O N FL I C T S O F I N T E R E S T
assay in a population with classic risk factors for IPA, we found im‐
provement in sensitivity and specificity over the prototype LFD, as Carol A. Kauffman is a member of the Data Safety Monitoring Board
6
previously had been noted by Hoenigl et al. for CIDARA. Marisa H. Miceli is a consultant for Astellas and a mem‐
Several prior studies have noted that exposure to mould‐ac‐ ber of the Data Review Committee for Scynexis. Shiwei Zhou and
tive antifungal agents decreased the sensitivity of the prototype Kathleen Linder have no conflicts to declare.
LINDER et al. |
5