Detection of non-infectious conditions mimicking pneumonia in the intensive care setting: usefulness of bronchoalveolar fluid cytology.

Abstract

The present study investigated the usefulness of bronchoalveolar (BAL) fluid cytology in the identification of non-infectious pulmonary conditions in patients hospitalized in the intensive care unit (ICU) and suspected of pneumonia. A total of 182 BAL fluid samples obtained during a 27-month period from 130 ICU patients with suspected pneumonia were quantitatively cultured and investigated for opportunistic pathogens. Cytocentrifuged preparations stained with the May-Grünwald Giemsa and Perls's methods were reviewed. A non-infectious aetiology was considered when cultures yielded micro-organisms in quantities < 10(3) colony-forming units (CFU) per ml, in the absence of any other pathogen and in conjunction with one or more of the following cytological findings: > 20% haemosiderin macrophages, > 10% lymphocytes, the presence of activated lymphocytes, plasma cells, > 5% eosinophils, a preponderance of foamy macrophages, reactive type II pneumocytes or malignant cells. Patients' clinical records were reviewed to identify a clinical diagnosis for these episodes. In thirty-five (19.2%) BAL fluid samples from 26 patients, the cytological findings pointed to a non-infectious origin. An alternative diagnosis was ascertained in 20 of 26 patients. Diagnoses included: drug-induced pneumonitis (n = 7), aspiration of gastric contents (n = 2), pulmonary emboli (n = 3), adult respiratory distress syndrome (n = 4), lung contusion (n = 1), cardiogenic pulmonary oedema (n = 1), and carcinomatous lymphangitis (n = 2). The BAL fluid cytological findings were readily discernable and proved to be useful in the diagnostic work-up of samples obtained from ICU patients with suspected pneumonia.

Full text

Page 1

RESPIRATORY MEDICINE (1999) 93,571-578
Detection of non-infectious
pneumonia in the intensive care setting:
usefulness of bronchoalveolar
conditions mimicking
fluid cytology
J. A. JACOBS*, E. I. G. B. DE BRAUWER*,
SJ. SC. WAGENAAR~, A. J. A. M. VAN DER VEN*, C. A. BRUGGEMAN*
G. RAMSAY+, N. A. M. COBBEN~,
AND
M. DRENT$
Departments of *Medical Microbiology, tIntensive Care Unit and SPulmonology,
University Hospital Maastricht and BDepartment of Pathology, Onze Lieve Vrouwen Gasthuis, Amsterdam,
The Netherlands
The present study investigated the usefulness of bronchoalveolar (BAL) fluid cytology in the identification of non-
infectious pulmonary conditions in patients hospitalized in the intensive care unit (ICU) and suspected of
pneumonia. A total of 182 BAL fluid samples obtained during a 27-month period from 130 ICU patients with
suspected pneumonia were quantitatively cultured and investigated for opportunistic pathogens. Cytocentrifuged
preparations stained with the May-Grtinwald Giemsa and Perls’s methods were reviewed. A non-infectious
aetiology was considered when cultures yielded micro-organisms in quantities < lo3 colony-forming units (CFU)
per ml, in the absence of any other pathogen and in conjunction with one or more of the following cytological
findings: >20% haemosiderin macrophages, > 10% lymphocytes, the presence of activated lymphocytes, plasma
cells, > 5% eosinophils, a preponderance of foamy macrophages, reactive type II pneumocytes or malignant cells.
Patients’ clinical records were reviewed to identify a clinical diagnosis for these episodes.
In thirty-five (19.2%) BAL fluid samples from 26 patients, the cytological findings pointed to a non-infectious
origin. An alternative diagnosis was ascertained in 20 of 26 patients. Diagnoses included: drug-induced pneumonitis
(n = 7), aspiration of gastric contents (n = 2), pulmonary emboli (n = 3), adult respiratory distress syndrome (n = 4),
lung contusion (n = l), cardiogenic pulmonary oedema (n = l), and carcinomatous lymphangitis (n = 2). The BAL
fluid cytological findings were readily discernable and proved to be useful in the diagnostic work-up of samples
obtained from ICU patients with suspected pneumonia.
RESPIR. MED. (1999) 93,571-578
0 1999 HARCOURT
bLISHER!? h
Introduction
Pneumonia is common in patients in intensive care
units (ICUs), especially in mechanically ventilated
patients. Establishing the diagnosis of pneumonia in these
patients is notoriously difficult as clinical and radiological
parameters offer a high sensitivity at the cost of an
unacceptably low specificity (1). Adding the results of
quantitative cultures of bronchoalveolar lavage (BAL) fluid
increases the diagnostic specificity (2) but preliminary
and definitive culture results take 24 h and up to 72 h,
respectively.
Apart from infectious pneumonia, a number of non-
infectious pulmonary conditions may explain the clinical
Received 23 February 1999 and accepted in revised form 3 May
1999.
Correspondence should be addressed to: Jan A. Jacobs, Depart-
ment of Medical Microbiology, University Hospital of Maastricht,
P.O. Box 5800, 6202 AZ Maastricht, The Netherlands. Fax: 31-43-
387 66 43; E-mail jja@ms-azm-3.azm.nl
0954-6111/99/080571+08 $12,00/O
symptoms of the ICU patient with suspected pneumonia.
Such conditions include pulmonary haemorrhage, malig-
nancy, drug-induced toxicity, the adult respiratory distress
syndrome (ARDS) and cardiogenic pulmonary oedema. It
is important to distinguish these conditions from pneumo-
nia as the management and prognosis of these entities is
quite different (3,4). Distinctive findings in BAL fluid
cytology, such as lymphocytosis, the presence of activated
lymphocytes, plasma cells and eosinophils or the prepon-
derance of foamy macrophages point to drug-induced
pulmonary disease (5,6), and the presence of reactive type
II pneumocytes has been described in association with
ARDS (7). BAL fluid cytological findings in ARDS are
characterized by a marked predominance of neutrophils in
the early phase and a recruitment of macrophages,
lymphocytes and eosinophils in the late phase (8). A
number of >20% haemosiderin macrophages has been
demonstrated to be indicative for alveolar haemorrhage (9).
Until now, however, the usefulness of BAL fluid cytology
in the diagnosis of these non-infectious pulmonary condi-
tions in ICU patients has received little attention in the
0 1999 HARCOURT PUBLEHERS LTD

Page 2

572 J. A. JACOBS ETAL.
literature. In daily practice, the use of BAL fluid cytology is
limited to the enumeration of infected cells and to the
identification of squamous epithelial cells which are
indicative of oropharyngeal contamination (10,ll).
In our hospital, bronchoscopy with BAL is routinely
used in the diagnosis of pneumonia in the ICU setting (2).
Recently, we introduced a standardized protocol for the
cytocentrifugation process and included the differential cell
count as part of the routine microbiological work-up of
BAL fluid specimens. As we collected a series of
consecutive BAL fluid samples performed in ICU patients
with suspected pneumonia, we decided to evaluate the
usefulness of BAL fluid cytology in the prediction of non-
infectious pulmonary conditions. We therefore retrospec-
tively looked for the diagnosis in those BAL fluid samples
that showed cytological findings consistent with a non-
infectious condition.
Materials and Methods
STUDY POPULATION
Over a 27-month period (January 1996 to April 1998), BAL
fluid samples from ICU patients in the University Hospital
Maastricht were obtained. All patients were suspected of
having pneumonia as defined by clinical and radiological
criteria (2).
SAMPLING TECHNIQUE
A fiber-optic bronchoscope (Pentax FB-lSH/FB- 15X,
Pentad Medicals, Tokyo, Japan) was introduced through
a special adaptor (Swivel Connector, Gibeck Respiration,
Upplands V&by, Sweden) and ‘wedged’ into
affected segmental or subsegmental bronchus. The
fluid was instilled into the subsegment through the biopsy
channel of the bronchoscope in four aliquots of 50 ml
sterile saline (0.9% NaCl, room
immediately aspirated and recovered. The BAL fluid
samples were transported to the laboratory within 1.5 min
of collection and analysed within 1 h of arrival in the
laboratory.
the
temperature) and
PROCESSING OF BAL FLUID SPECIMENS
The volume of the recovered BAL fluid was recorded. The
first fraction, representing the bronchial fraction, was
separated for mycobacterial culture and the remaining
fractions were pooled for further analysis. The total cell
count was performed in a FuchsRosenthal
ometer chamber. BAL fluid samples were quantitatively
cultured for bacteria and yeasts by means of a calibrated
loop technique (12). They were also cultured for filamen-
tous fungi and mycobacteria and, if clinically indicated, for
viruses and Legionella
spp. In addition, stains for detection
of Pneumocystis carinii and Legionella
munofIuorescent monoclonal antibody stainings), filamen-
tous fungi (Methenamine-silver stain) and acid-fast bacteria
haemocyt-
pneumophiia
(im-
(auramine-rhodamine stain) were performed on cytocen-
trifuged preparations. When a community-acquired pneu-
monia was suspected, serology for detection of viral
pathogens was done and polymerase chain reactions for
detection of Chlamydia pneumoniae
pneumoniae
were performed.
Cytocentrifugation was done
3 apparatus (Shandon Scientific Ltd, Astmoor, U.K.)
as previously described (13). Differential
were made on May-Grtinwald
preparations, by the first or second author examining
500 nucleated cells. Both macrophages and neutrophils
were screened for the presence of intracellular organisms
and the number of infected cells was expressed as
a percentage of the 500-cell aliquot
preparations were sealed (Xylene substitute mountant,
Shandon) and stored at room temperature. When baso-
philic granules were seen in macrophages or polymorpho-
nuclear cells, Perls’ stain for haemosiderin visualization was
performed (14).
and Mycoplasma
with the Cytospin
cell counts
Giemsa (MGG) stained
counted. The
REJECTION CRITERIA
BAL fluid samples were excluded if the retrieved volume
was less than 20 ml, if the cytocentrifuged preparations
showed excessive amounts of red blood cells, intercellular
debris or damaged nucleated cells, or if the differential cell
count yielded 1% squamous epithelial cells or 2 5%
ciliated cells.
DEFINITION
BAL fluid samples were categorized into four groups based
on the following diagnostic criteria.
Group I, pneumonia microbiologically
this group consisted of samples for which quantitative
cultures yielded micro-organisms in quantities >104 colony
forming units per millilitre (cfu ml-‘), or for which an
obligatory respiratory pathogen such as P. carinii
demonstrated.
Group II, pneumonia not conclusive: BAL fluid samples
were categorized as “not conclusive” when they had
borderline quantitative culture results, i.e. counts 2 lo3
and < IO4 cfu ml-‘.
Group III, pneumonia excluded, no cytological abnorm-
alities: infectious pneumonia was considered as microbio-
logically excluded at a culture threshold of < lo3 cfu ml-’
and < 2% infected cells, and with no obligatory respiratory
pathogen demonstrated. On cytological examination of the
BAL fluid, none of the findings cited for Group IV was
observed.
Group IV pneumonia excluded, non-infectious aetiology:
BAL fluid samples were considered as of ‘non-infectious
etiology’ when pneumonia was microbiologically excluded
(criteria as for Group III) in the absence of previous (within
72 h) antimicrobial therapy and when one or more of the
following cytological findings were observed: the presence
of malignant cells, a count of > 10% lymphocytes, a count
of > 5% eosinophils, the presence of activated lymphocytes
confirmed:
was

Page 3

or plasma cells and the preponderance of foamy macro-
phages. Moreover, a number of > 20% haemosiderin laden
macrophages was considered as diagnostic for alveolar
haemorrhage (9).
REVIEW OF CLINICAL RECORDS
The patients’ clinical records were reviewed for clinical,
radiological and laboratory evidence to identify diagnoses
of non-infectious aetiology.
Results
During the study period, 1721 patients were admitted to the
ICU and 228 bronchoscopies with BAL were performed
(Fig. 1). Forty-six BAL fluid samples were excluded from
analysis, resulting in 182 BAL fluid samples obtained from
130 patients.
Seventy-seven BAL fluid samples were considered as
microbiologically proved pneumonia and allocated to
Group I. L.
was identified as the pathogen
in one BAL fluid sample, P.
samples (three patients) and a viral pneumonia was
detected in three BAL fluid samples obtained in two
patients (RSV and para-influenza virus, respectively). At a
threshold of lo4 cfu ml-‘,
further confirmed in 69 BAL fluid samples. Group II
(pneumonia not conclusive) consisted of 23 BAL fluid
samples and 47 BAL fluid samples were included in Group
III (pneumonia microbiologically excluded, no cytological
abnormalities).
Based on the exclusion of a microbiological pathogen
and on the presence of one or more of the cytological
pnewnophila
carinii
in four BAL fluid
infectious pneumonia was
BAL FLUIDCYTOLOGY~\~PNF~UMONL~M~MI~S
573
findings, 35 BAL fluid samples were assigned to Group IV
(‘non-infectious etiology’). These 35 BAL fluid samples
accounted for 19.2% of the 182 included BAL fluid samples
and were obtained from 26 (20%) of 130 patients (nine
patients underwent repeat bronchoscopy with BAL).
Table 1 lists the non-infectious conditions identified by
review of the clinical records in the 26 patients and the
differential cell counts together with the cytological findings
in the corresponding BAL fluid samples.
The patients’ mean age was 62.8 Z!Z 14.2 years, and the
male to female ratio 1.36,: 1. Half (13) of the patients were
admitted from the community, one patient was admitted
from a nursing home and 12 were transferred from another
hospital ward.
Bronchoscopy was performed after 11.2 f 13.9 days of
ICU admission (range 1-57 days), with seven BAL fluid
samples obtained on the day of the patient’s admission to
the ICU. Twenty-six of 35 BAL fluid samples were
obtained from ventilated patients; in 21 samples the patient
was ventilated for more than 72 h prior to suspicion of
pneumonia. For two BAL fluid samples, no data on
ventilation were retrieved upon chart review.
Three patients demonstrated amiodarone-induced pneu-
monitis, which was confirmed histologically in one case.
Nortriptyline pulmonary toxicity was observed after an
accidental overdose in a psychiatric patient. Carbamaze-
pine-induced pneumonitis occurred due to self-poisoning.
Methotrexate-induced pneumonitis occurred in a patient
treated because of rheumatoid arthritis. Melloquine-in-
duced pneumonitis was seen in a patient with hemizygote
glucose-6-phosphate-dehydrogenase deficiency. Aspiration
of gastric contents was documented in a patient during
surgical treatment for achalasia and in another patient
who aspirated enteral feeding during a prolonged ICU
1721 ICU admissions in a 27-month period
1
228 BALs performed in 161 patients
1 46 BALs
in 31 patients excluded
182 BALs in 130 patients
/------
included
Pneumonia microbiologically
77 BALs
confirmed
in 68 patients*
Group
Diagnosis
II
of pneumonia
23 BALs
< IO4 cfu ml-’ and 2 lo3 cfu ml-
not conclusive
in 23 natienta*
Group I
Pneumocystis
Legionella
Viral
Bacteria
carinii
pneumophila
pneumonia
2 lo4 CPU/ml-’
Bacteria
Group
Pneumonia
No cytological
III
microbiologically
abnormahties
47 BALs
excluded **
in 43 patients*
Group
Pneumonia
Cytological
IV
microbiologically
evidence
non-infectious
35 BALs
excluded
to a
origin
**
pointing
in 26 patients*
Diagnosis of non-infectious
30 BALs
origin ascertained
in 20 patients
FIG. 1. Flow chart of BAL fluid selection and definitions. *Numbers exceed the total number of patients included ‘as several
patients underwent repeat bronchoscopy with BAL. **Bacteria < 103/ml-’ and ~2% cells containing intracellular
organisms.
’

Page 4

TABLE
1. Patients with microbiologically excluded pneumonia and with cytological evidence of a non-infectious pulmonary condition. The clinical diagnoses, the day of
ICU stay at which bronchoscopy with BAL was performed, the BAL fluid differential cell counts and striking cytological findings are listed
2
Y
Patient
BAL fluid differential cell count?
?
Gender, age
7
No.
(years)
Clinical diagnosis
Day TCC*
Pmn Lym
Am
Eos MC P
BAL fluid cytological findings
8
I
1
F, 74
2
M, 77
3
M, 74
4
M, 70
5
F, 35
6
F, 70
I
M, 64
8
M, 53
9
M, 65
10
M, 72
11
M, 66
12
F, 73
13
F, 52
14
F, 66
Arniodarone pulmonary toxicity Amiodarone pulmonary toxicity Amiodarone pulmonary toxicity
Nortryptiline-intoxication
Overdose with carbamazepine
Methotrexate-induced toxicity Mefloquine-induced pulmonary toxicity
Aspiration of gastric contents
Aspiration of gastric contents
Pulmonary embolisms
Pulmonary embolisms
Pulmonary embolisms
ARDS ARDS
15
F, 67
ARDS
16
M, 47
ARDS
17
M, 34
Lung contusion
2
16
1
11
6
57
2 3 6
1
5
21
1
10
1
3
1 5
10 17
32 43 45
5
1
8
213
9-6 18.0 69.7
0.2 2.5 O-0 Activated lymphocytes, foamy macrophages
157
10.6 14-O 75.0
0.0 O-4 0.0 Activated lymphocytes, foamy macrophages
148
27.0
12.4 57.8
2.2 0.6 O-0 Foamy macrophages
159
74-7
7.9
16.2
O-8 O-4 O-0 Foamy macrophages
320
26.6
17.2 53.2
0.6 O-4 2.0 Foamy macrophages, activated lymphocytes,
plasma cells
76
30.2
38.0
15.8 10.4 5-4 O-2 Activated lymphocytes, plasma cells,
elevated eosinophil count
159
89.4
4.6
6.0
O-0 0.0 O-0 Activated lymphocytes, foamy macrophages,
reactive type II pneumocytes
80
9.2
67.6 21.2
1.8 0.0 0.2 Marked lymphocytosis, plasma cells
103
O-2 75.2 22.6
O-0 I.0 1-O Marked lymphocytosis, plasma cells
127
26.2
8.8 63.0
1.6 O-4 0.0 Abundance of lipid-laden, foamy macrophages
400
77.6
5.4
15.2
1.8 O-0 0.0 Activated lymphocytes, foamy macrophages
215
25.4
38-O 35-O
1-O 0.6 0.0 Activated lymphocytes, foamy macrophages
207
73.8
10.4 14.6
1.0 0.2 0.0 Reactive type II pneumocytes
107
41.4
8.4 42.8
5.8 1.6 O-0 Reactive type II pneumocytes
124
18.9 49.2
30.5
0.6 O-8 0.0 Activated lymphocytes
254
73.4
9.8
16.8
0.0 O-0 0.0 Foamy macrophages, alveolar macrophages with
large and distinct vacuoles
110 44.0
17.7 30.9
5.8 1.2 0.4 Activated lymphocytes, plasma cells, reactive type II
pneumocytes
120
27.0
10.8 62.2
O-0 0.0 0.0 Activated lymphocytes, foamy macrophages, reactive
type II pneumocytes
403
68.4
8-8 22.2
O-6 O-0 0.0 Activated lymphocytes
130
58.0
19.4 21.6
1.0 O-0 O-0 Activated lymphocytes,
of foamy
presence
macrophages
658
8.8
58.8 29.8
0.0 O-2 2.4 Activated lymphocytes, plasma cells,
of
presence
foamy macrophages
224
58.4
18.8 22.8
O-0 0.0 O-0 Activated lymphocytes,
of foamy macrophages
presence
421
51.0
8-O 40.4
O-6 O-0 0.0 Activated lymphocytes
460
84.9
8-7
2.0
4.4 0.0 0.0 Activated lymphocytes
320
13.4 28.6
58.0
0.0 O-0 0.0 Elevated lymphocyte count
210
48.2
5.2 45.6
0.8 0.2 O-0 Activated lymphocytes, foamy macrophages,
20% haemosiderin macrophages

Page 5

18
F, 77
19
M, 78
20
M, 65
21
F, 64
22
F, 49
23
F, 27
24
M, 72
25
M, 65
26
M, 76
Pulmonary oedema with
alveolar haemorrhage
Carcinomatous lymphangitis Carcinomatous lymphangitis No diagnosis ascertained No diagnosis ascertained No diagnosis ascertained No diagnosis ascertained No diagnosis ascertained No diagnosis ascertained
5
184
9
1280
6
300
4
432
1
237
6
346
19
84
25
90
3
73
1.4
3.5 94-9
O-2 0.0 0.0 Alveolar macrophages with haemosiderin granules,
60% haemosiderin macrophages
93.6
2-2
4-2
O-0 O-0 O-0 Malignant cells
1 l-8
21.2 65.6
0.0 1.4 0.0 Malignant cells
2.2
2.0 95-4
0.4 0.0 0.0 Abundance of foamy macrophages
52-O 16.8 31-2
O-0 0.0 O-O Abundance of foamy macrophages
27-7
15.7 53-2
3-2 O-2 O-O Activated lymphocytes, abundance of foamy macrophages
26.2
35.6 34.0
4.2 0.0 O-O Lymphocytosis
31.2
31.6 35.4
l-8 0.0 0.0 Lymphocytosis, activated lymphocytes
9-O 16-O 73,8
O-6 0.6 O-0 Activated lymphocytes, foamy macrophages
with distinct vacuoles
* TCC = total cell count, values expressed in lo3 ml-‘.
t Counts performed on 500 nucleated cells. Pmn = polymorphonuclear neutrophils, Lym = lymphocytes, Am = alveolar macrophages, Eos = eosinophils, MC = mast
cells, P = plasma cells.

Page 6

576 J. A. JACOBS ETAL.
stay. Pulmonary emboli were found at autopsy in two
patients and a third patient developed multiple pulmonary
emboli after surgical revision of a hip prosthesis. Based
on respiratory parameters and a clinical picture, the
diagnosis of ARDS was assessed in four patients; in one
of them the diagnosis was confirmed on histological
examination at post-mortem. Numerous iron-laden macro-
phages were seen in one patient with lung contusion and in
another patient with severe cardiogenic pulmonary oede-
ma. Malignant cells originating from a disseminated lung
adenocarcinoma were seen in two patients; in one of them,
the BAL fluid cytology findings provided the first suspicion
of malignancy.
Overall, in seven patients the aetiological diagnosis was
established by histology obtained by biopsy or autopsy and
in 13 patients the diagnosis was supported by clinical,
radiological and laboratory findings. In the six remaining
patients, data regarding the clinical status were not
conclusive.
Discussion
Based upon the absence of a microbiological pathogen and
upon cytological evidence present in MGG-stained cyto-
centrifuged preparations, in this study we classified 35/182
(19.2%) BAL fluid samples as of non-infectious origin.
These BAL fluid samples were obtained in 26 ICU patients
with suspected pneumonia. Retrospectively, we identified a
diagnosis explaining the non-infectious condition in 20 of
the 26 patients.
We realize that one of the major limitations of this study
was its retrospective design. This design did not allow
tracing of all cases of non-infectious lung conditions. As the
use of antimicrobial agents prior to bronchoscopy was not
systematically documented, reliable exploration of Group
II (pneumonia not conclusive, BAL fluids with borderline
quantitative culture results) and Group III (pneumonia
excluded, no cytological abnormalities) was not possible.
Neither were we able to identify a diagnosis in all non-
infectious BAL fluid samples. In particular, difficulties were
encountered in the retrospective diagnosis of ARDS and
drug-induced pneumonitis. In the case of a suspected drug-
induced pneumonitis, a definite role for a particular drug
can not be proven in the absence of rechallenge. In the
present series, rechallenge supported the diagnosis of
carbamazepine-induced toxicity as the patient was re-
admitted with a second auto-intoxication.
mefloquine was confirmed by restarting of the drug prior to
the drug being suspected. Both cases have been described
elsewhere in detail (15,16). Among others, amiodarone,
methotrexate and anti-depressants are known to be capable
of causing drug-induced pneumonitis (17-19). In the cases
presently described, these drugs were considered as
causative because withdrawal of the drug resulted in
clinical and radiological improvement.
Reviewing a large number of studies, Timsit et al.
concluded that about two-thirds of the episodes of
suspected pneumonia in ventilated patients do not meet
quantitative culture criteria but are related to other
The role of
conditions. The authors however admitted that no data
definitely supported their conclusion, as many of these
episodes were believed to be false-negative pneumonias in
which bacterial growth was suppressed by the administra-
tion of antimicrobial agents prior to bronschoscopy (20). In
view of their supposition, it is striking that few studies list
an alternative diagnosis for those episodes with no
microbiological evidence of pneumonia (3,4,21,22). The
reported incidences of non-infectious conditions in these
studies were higher than in the present study, but non-
infectious BAL fluid samples were defined at a culture
threshold of < lo4 CFU ml-’ , whereas in the present
study a threshold of < lo3 CFU ml-’ was applied to
exclude BAL fluid samples with borderline quantitative
culture results. The low culture threshold in the present
study was combined to a threshold for infected cells of 2%,
which is the lowest cited cutoff value (23). By using these
stringent criteria, we aimed to minimise the diagnosis of
false-negative infectious pneumonia but probably may have
underestimated the number of non-infectious conditions.
Second, the authors of previous studies looked for
alternative diagnoses in all non-infectious BAL fluid
samples, irrespective of the BAL fluid cytological findings.
In this way, their alternative diagnoses included more cases
of atelectasis, pleural involvement and pulmonary oedema,
which are conditions for which, to our knowledge, no
specific cytological BAL fluid findings have been described.
Although the present study did not intend to search an
alternative diagnosis in Groups II and III, we assume that
conditions such as atelectasis and pleural involvement may
have been present in at least part of these episodes. The
exact incidences and nature of the non-infectious condi-
tions in this setting should however be addressed by a
prospective study.
In contrast to previous studies, we added the findings of
the cytological examination to direct the investigation to a
non-infectious condition. Retrospectively, we were able to
demonstrate an alternative diagnosis for infectious pneu-
monia in 20 (77%) out of the 26 patients that were selected
based on well-defined cytological criteria. Most microbio-
logical laboratories are reluctant to perform BAL differ-
ential cell counts as these counts are not always
unequivocal (24) and cannot differentiate bronchial infec-
tion from lung infection (25). Moreover, performing BAL
fluid differential cell counts requires skilled microscopists
and is looked upon as labourious and time-consuming in
times of limited sources and tendencies to automation. The
cytological findings looked for in the present study are,
however, easily discernable on MGG-stained cytocentri-
fuged preparations. The processes of cytocentrifugation,
MGG staining and cell differentiation as described here can
be achieved within a 2-h period, making the results of the
BAL fluid cytology available within a very short delay. The
MGG stain (or an equivalent one) with the differential cell
count can be conveniently incorporated into the routine
microbiological work-up of BAL fluid samples. Other
advantages of the MGG stain are its ability to detect P.
carinii (12) and its reliability for enumeration of intracel-
lular organisms (13). The sensitivity of the cytological
findings for the detection of a non-infectious origin,

Page 7

together with the feasibility and the low turn-over time of
the procedure, lead us to recommend BAL fluid cytology as
part of the routine work-up in cases of ICU patients with
suspected pneumonia.
One should note that the cytological findings under
consideration are not pathognomonic. They point to the
possibility of a non-infectious condition which has to be
confirmed and identified by clinical, radiological or
laboratory findings. In our experience, the above-listed
cytological findings were not specific for the non-infectious
BAL fluid samples. In this way, we also observed alveolar
haemorrhage in seven of the 69 bacterial pneumonias, in
agreement with the association between alveolar haemor-
rhage and infection reported by others (9). Moreover, seven
of the patients in whom a non-infectious aetiology was
diagnosed developed a pneumonia during their further ICU
stay. In BAL fluid samples from patients with P. carinii
pneumonia, we occasionally observed foamy macrophages
and plasma cells similar to cases of drug-induced toxicity
(unpublished observations) and we have noted elevated
numbers of lymphocytes in BAL fluids obtained from
patients with tuberculosis, in line with the findings of others
(26,27). These observations emphasize the importance of a
thorough microbiological investigation, even if BAL fluid
cytology at first glance points to a non-infectious condition.
Further, it should be noted that the presence of foamy
macrophages in the case of amiodarone intake may
represent amiodarone-impregnation and do not necessarily
imply toxicity (6).
From Table 1, it is clear that, apart from malignancy,
none of the cytological findings was confined to a particular
non-infectious aetiology. This may partly be explained by
an overlap between the distinct conditions, i.e., both lung
contusion and aspiration of gastric contents are we&known
risk factors for the development of ARDS (28) and non-
infectious processes may co-exist with pneumonia (29). It is
clear that the discriminative power of the cytological
findings will be augmented if the number of BAL fluid
samples included for comparison increases. For this reason,
we are presently conducting a prospective study that will
investigate the predictive value of the different cytological
parameters and their combinations.
In conclusion, the present study demonstrated the value
of the BAL fluid cytological findings for the diagnosis of
non-infectious conditions jn ICU patients with suspected
pneumonia. We recommend incorporating
cytology as part of the routine work-up of BAL fluid
samples obtained in the ICU setting. A prospective study
should address the exact incidence and aetiologies of non-
infectious pulmonary conditions in the ICU setting and
allow refinement of the cytological description. In this way,
cytology may further improve the diagnostic accuracy of
BAL fluid analysis in ICU patients with suspected
pneumonia.
BAL fluid
Acknowledgement
The authors thank Dr J. Tjhie for performing the PCR
analysis.
References
1. Johanson WG Jr, Seidenfeld JJ, Gomez P, Los Santos
de R, Coalson JJ. Bacteriologic diagnosis of
nosocomial pneumonia
mechanical ventilation. Am Rev Respir Dis 1988; 137:
259-264.
2. Bonten MJM, Bergmans DCJ, Stobberingh EE, et al.
Implementation of bronchoscopic techniques in the
diagnosis of ventilator-associated pneumonia to reduce
antibiotic use. Am J Respir Crit Care Med 1997; 156:
1920-1924.
3. Guerra LF, Baugham RP. Use of bronchoalveolar
lavage to diagnose bacterial pneumonia in mechani-
cally ventilated patients. Crit Cure Med 1990; 1S: 16%
172.
4. Timsit JF, Misset B, Goldstein FW, Vaury P, Carlet J.
Reappraisal of distal diagnostic testing in the diagnosis
of ICU-acquired pneumonia. Chest 1995; 108: 1632-
1639.
5.
Drent M, van Velzen-Blad H, Diamant M, Wagenaar
Sj SC, Donkerwolcke-Bogaert
JMM. Differential diagnostic value of plasma cells
in bronchoalveolar lavage fluid. Chest 1993; 103:
1720-1724.
Rosenow EC III, Martin WJ II. Drug-induced inter-
stitial lung disease. In: Schwarz MI, King TE (eds).
Interstitial Lung Disease. 2nd edn, 1993: 255-270.
Grotte D, Stanley MW, Swanson PE, Henry-Stanley J,
Scott D. Reactive Type II pneumocytes in bronchoal-
veolar lavage fluid from adult respiratory distress
syndrome can be mistaken for cells of adenocarcinoma.
Diagn. Cytopathol. 1990; 6: 317-322.
Nakos G, Kitsiouli EL, Tsangaris L, Lekka MD.
Bronchoalveolar lavage fluid characteristics of early
intermediate and late phases of ARDS. Intensive Care
Med 1988; 24: 296-303.
Lassence de A, Fleury-Feith J, Escudier E, Beaune J,
Bernaudin JF, Cordonnier C. Alveolar hemorrhage.
Diagnostic criteria and results in 194 immunocompro-
mised hosts. Am J Respir
157-163.
Kahn FW, Jones JM. Diagnosing bacterial respiratory
infection by bronchoalveolar lavage. J Infect Dis 1987;
155: 8624369.
Chastre J, Fagon JY, Soler P, et al. Quantification of
BAL cells containing intracellular bacteria rapidly
identifies ventilated patients with nosocomial pneumo-
ma. Chest 1989; 95: 19OS-192s.
Baselski VS, El-Torky M, Coalson JJ, Griffin JP. The
standardization of criteria for processing and inter-
preting laboratory specimens in patients with suspected
ventilator-associated pneumonia.
s571-s579.
De Brauwer EIGB, Jacobs JA, Nieman FH, Brugge-
man CA, Drent M. Test characteristics of Acridine
Orange, Gram and May-Griinwald
for the detection of intracellular organisms in bronch-
oalveolar lavage fluid. J Clin
427429.
following prolonged
6.
7.
8.
9.
10.
11.
12.
13.
M, van den Bosch
Crit Care Med 1995; 151:
Chest
1992; 102:
Giemsa stains
Microbial 1999: 37:

Page 8

578
J. A. JACOBS ETAL.
14. Dacie JV, Lewis SM (eds). Practical
Edition. Edinburgh: Churchill Livingstone 1984.
15. Wilschut FA, Cobben NAM,
Lamers RJS, Wouters EFM, Drent M. Acute inter-
stitial pneumonitis due to single overdose with carba-
mazepine. Eur Respir J 1997; 10: 2163-2165.
16. Drent M. Drug-induced pneumonia associated with
hemizygote glucosed-phosphate dehydrogenase defi-
ciency. Eur J Haematol
1998; 61: 218-220.
17. Cooper AD, White DA, Matthay RA. Drug-induced
pulmonary disease. Part 1: cytotoxic drugs. Am Rev
Respir Dis 1986; 133: 321-340.
18. Cooper AD, White DA, Matthay RA. Drug-induced
pulmonary disease. Part 2: noncytotoxic drugs. Am Rev
Respir Dis 1986; 133: 4888505.
19. Hubbard R, Venn A, Smith C, Cooper M, Johnston I,
Britton J. Exposure to commonly prescribed drugs and
the etiology of cryptogenic fibrosing alveolitis. Am J
Respir Crit Care Med 1998; 157: 743-747.
20. Timsit JF, Chevret S, Valcke J, Misset B, Renaud B,
Goldstein FW, Vaury P, Carlet J. Mortality
nosocomial pneumonia in ventilated patients: influence
of diagnostic tools. Am J Respir Crit Care Med 1996;
154: 116-123.
21. Aubas S, Aubas P, Capdevila X, Darbas H, Roustan J-
P, du Cailar J. Bronchoalveolar lavage for diagnosing
bacterial pneumonia in mechanically ventilated pa-
tients. Am J Respir Crit Care Med 1994; 149: 860-866
22. Meduri GU, Mauldin GJ, Wunderink RJ, et al. Causes
of fever and pulmonary densities in patients with
Haematology.
6th
Thunissen FBJM,
of
clinical manifestations of ventilator-associated pneu-
monia. Chest 1994; 106: 221-233.
23. Allaouchiche B, Jaumain H, Dumontet C, Motin J.
Early diagnosis of ventilator-associated pneumonia. Is
it possible to define a cutoff value of infected cells in
BAL fluid? Chest 1996; 110: 1158-l 165.
24. Baughman RP, Strohofer S, Kim CK. Variation of
differential cell counts of bronchoalveolar lavage fluid.
Arch Path01 Lab Med 1986; 110: 341-343.
25. Thorpe JE, Baughman RP, Frame PT, Wesselet TA,
Staneck JL. Bronchoalveolar lavage for diagnosing
acute bacterial pneumonia. J Infect Dis 1987; 155: 855-
861.
26. Baughman RP, Dohn MN, London RG, Frame PT.
Bronchoscopy with bronchoalveolar lavage in tubercu-
losis and fungal infections. Chest 1991; 99: 92-97.
27. Drent M, Wagenaar Sj SC, Mulder PHC, Velzen-Blad
van H. Bronchoalveolar lavage fluid profiles in
sarcoidosis, tuberculosis, non-Hodgkin’s and Hodg-
kin’s disease: an evalutaion of differences. Chest 1994;
105: 514-520.
28. Steinberg KP, Milberg JA, Martin TR, Mauncer RJ,
Cockill BA, Hudson LD. Evolution of bronchoalveolar
cell population in the adult respiratory distress syn-
drome. Am J Respir Crit Care Med 1994; 150: 113-122.
29. Marquette CH, Copin M-C, Wallet F, et al. Diagnostic
tests for pneumonia in ventilated patients: prospective
evaluation of diagnostic accuracy using histology as a
diagnostic gold standard. Am J Respir Crit Care Med
1995; 151: 1878-1888.