807
Can CT Distinguish Hypersensitivity Pneumonitis from Idiopathic Pulmonary Fibrosis?
David A. Lynch 2 John D. Newell’ P. Mark Logan3 Talmadge E. King, Jr.2 Nestor L. MUller3
Received
January 25, 1995; accepted after revi-
sion May 9, 1995.
Supported in part by Specialized Center of Research grant HL 27353 from the National Heart, Lung, and Blood Institute to T. E. King, Jr. I Department of Radiology, National Jewish Center for Immunology and Respiratory Medicine and University of Colorado Health Sciences Center, Denver, CO. Address correspondence to D. A. Lynch. Department of Radiology, Box A030, University of Colorado ver, CO 80262. 2Division
Health
of Pulmonary
Sciences
Center,
Den-
Care
Medi-
and Critical
cine, National Jewish Center for Immunology and Respiratory Medicine and University of Colorado Health Sciences Center, Denver, CO. 3Department Columbia,
of Radiology, University B.C. , Canada.
Vancouver,
0361-803)(/95/1654---807 © American Roentgen Ray Society
of British
OBJECTIVE. The clinical management of patients with idiopathic pulmonary fibrosis differs markedly from that of patients with hypersensitivity pneumonitis. However, the two diseases often cannot be differentiated on clinical grounds. The purpose of this study was to establish whether CT can be used to make the distinction. MATERIALS AND METHODS. Thirty-six patients with idiopathic pulmonary fibrosis and 27 patients with hypersensitivity pneumonitis were studied. All diagnoses were confirmed or supported by open lung biopsy. Three of the patients with idiopathic pulmonary fibrosis had desquamative interstitial pneumonia, and the remainder had usual interstitial pneumonia. In 19 of the 27 patients with hypersensitivity pneumonitis, the disease was chronic (symptoms lasting more than I year), while eight had acute or subacute symptoms. Two radiologists, who had not previously seen any of the cases and were blinded to the diagnosis, reviewed the CT images by consensus. The extent and distribution of CT features (including ground-glass attenuation, honeycombing, and micronodules) were recorded. In each case, a CT diagnosis was made, and the level of diagnostic confidence was recorded. RESULTS. A CT diagnosis was made with a high level of confidence in 39 (62%) of 63 patients. In these patients, the CT diagnosis was correct in 35 cases (90%): 23 of 26 patients with a CT diagnosis of idiopathic pulmonary fibrosis and 1 2 of 1 3 patients with a CT diagnosis of hypersensitivity pneumonitis. In the three patients with desquamative interstitial pneumonia, the CT diagnoses were probable hypersensitivity pneumonitis in two cases and definite hypersensitivity pneumonitis in one case. Of the 19 patients with chronic hypersensitivity pneumonitis, only seven had a definite diagnosis of hypersensitivity pneumonitis based on CT findings; three had a defmite diagnosis of idiopathic pulmonary fibrosis. The patients with Idiopathic pulmonary fibrosis and usual interstitial pneumonia were more likely to have honeycombing and peripheral or lower lung zone predominance of disease, and less likely to have micronodules, than were patients with chronic hypersensitivity pneumonitis. The patients with idiopathic pulmonary fibrosis and desquamative interstitial pneumonia had widespread ground-glass opacity indistinguishable from some cases of acute or subacute hypersensitivity pneumonitis. CONCLUSION. Our results show that CT can be used to distinguish idiopathic pulmonary fibrosis from hypersensitivity pneumonitis in most but not all cases. Desquamative interstitial pneumonia cannot reliably be distinguished from acute or subacute hypersensitivity pneumonitis. Chronic hypersensitivity pneumonitis may have findings identical to those of usual interstitial pneumonia. Lung biopsy should still be considered the gold standard for diagnosis of interstitial lung disease. AJR 1995;165:807-811
The distinction of idiopathic pulmonary fibrosis (IPF) from chronic hypersensitivity pneumonitis (HP) is important because the clinical features of these disorders are often similar, but their management is different. The diagnosis of HP is an indication for thorough evaluation of the environment and removal of any potential causative agents [1 2]. Corticostenoid treatment may be used as an adjunct to these measunes. The treatment of IPF, on the other hand, usually includes corticostenoids and often other immunosuppressive agents, without environmental intervention. ,
LYNCH
808
High-resolution CT of the chest improves the accuracy of diagnosis of diffuse parenchymal lung disease when compared with the accuracy of diagnosis using chest radiographs [3-5]. However, the accuracy of CT diagnosis of HP has not been systematically studied. The purpose of this study was to establish whether CT can distinguish IPF from HP. We hypothesized that HP could be identified by the presence of poorly defined micronodules, widespread ground-glass attenuation, and predominance of disease in the upper or middle lung zone [6-8], while IPF would be characterized by predominant reticular opacities or honeycombing, by a peripheral distribution, and by lower lobe predominance [9].
ET AL.
(p
AJR:165,
.085).
=
pared
three
at selected
Population
This
was
Retargeting but
not at the
obtained
when
there
dependent
study
done
were 27 patients who had a diagnosis
at two
institutions.
Included
of HP and 36 patients who had
a diagnosis of IPF. All patients had high-resolution CT images formed. The diagnosis of IPF was confirmed on open lung biopsy cases. For this study, the diagnosis of IPF included the histologic ties
of desquamative
monia.
Because
history
of exposure,
interstitial
patients the
pneumonia
and
usual
with HP do not always diagnosis
of HP was
interstitial
perin all entipneu-
from
scored
have a recognizable
made
under
two
circum-
stances: (1) when the patient had a clinical history strongly suggesting HP and findings compatible with HP on transbronchial or open lung biopsy;
or
biopsy
(2)
strongly
lymphocyte
patients biopsy
when
the
patient
suggesting accumulation
with
clinical
findings
results,
were
excluded,
had
a transbronchial
HP
(usually
consisting
and
poorly
granulomas).
HP, but a study
open
lung
with
population
nonspecific of 63.
and
The
was
CT scans to the
Features
micronodules,
A
patients
with
IPF
per-
at one
prone
were
density
pied
by each
percentage ground-glass
disease of
Diagnoses
in
the
occupied
probable, of the
and
air-
emphysema, bronchiectasis. zones and
occu-
an overall
then estimated. Extensive be present when isolated 25% with were of poorly
of the
three HP
lung.
levels was
fulfilled:
of disease
presence
were
used. attenuation,
lung
Definite
criteria
at
opacities,
visually,
than
possible.
following and
traction lower
CT scans
predominance
lungs,
and
to
29
review
who
was
reticular
and
more
the
sheet
honeycombing,
was
from
a joint
radiologists
estimated
IPF)
while
by radiologists during
ground-glass
said
attenuation
or more
zones
was
were
made
scored
with
middle,
Jewish with
Hospital,
done score
distortion,
feature
National
patients
(eight from patients
thoracic
lines,
lung involvement attenuation was
definite,
were
isolated
upper,
the
General
associated
of the
linear
the
in
15 from
was
of two
parenchymal
percentage
years,
for
patient
Hospital
IPF)
irregular
of the
years
performed
the
from
and
Scoring
included
consolidation,
two
with
attenuation
The
HP
A standardized
scored
middle
(±13.3)
Each
a high-resolution
increase
General
by consensus
attenuation,
58.3
corn-
(1 or 1 .5 mm)
routinely with
at Vancouver
those
ground-glass
with
was
34 CT studies
Center.
cases of HP were classified as acute, subacute, or chronic HP based on the duration of symptoms. Patients with acute HP (n = 2) had symptoms that had lasted less than 1 month, those with subacute HP (n = 6) had symptoms that had lasted between 1 month and 1 year, and patients with chronic HP (n = 19) had symptoms that had lasted longer than 1 year. The mean (±SD) age of patients with HP was 52.5 (±12.5) compared
female,
of techniques. using
Images
with
diagnosis.
ground-glass
when
were
0.124).
=
images
significant
bias,
21 from
blinded
dence:
lung
other.
patients
Jewish
of the
Two
HP
(p
reconstructed
from Vancouver
ground-glass
of penibronchiolar
formed
suggesting leaving
or
and
by radiologists
National
space
IPF
collimation
to one
observer
(19
HP
the
with
with
lung.
To reduce
with
a retrospective
levels
institution
CT studies Study
patients
using a variety
to 10 thin
algorithm.
were
and Methods
were obtained
included
formed
of the
of the patients
1995
Evaluation
CT scans study
(56%)
13 (36%)
Radiologic
Center
Materials
Fifteen
with
October
widespread
in the defined
of conf idiagnosed upper
or
micronod-
on CT scans (Fig. 1 ). Probable HP was diagnosed when only one of these criteria was fulfilled. Definite IPF was diagnosed when two or more of the following criteria were fulfilled: lower zone predominance of disease, definite peripheral predominance of disease, and predominant ules
opacities
or honeycombing
(Fig.
2). Probable
IPF was
diagnosed
B Fig. 1 -chronic
hypersensitivity
pneumonitis.
50-year-old
woman with biopsy-proven
chron-
Ic hypersensitivity pneumonitis (cause not identified). A and B, CT scans at two levels show scattered poorly defined nodules (arrows) in left upper lobe, associated with linear areas of increased attenuation and ground-glass opacity. Larger nodule visible posteriorly is calcified granuloma. Some focal thickening of extrapleural fat and fissural thickening is presumed related to upper lobe fibrosis.
Fig. 2.-Idiopathic
pulmonary
fibrosis.
50-year old
woman with biopsy-proven idiopathic pulmonary brosis. High-resolution cT scan shows typical tern of honeycombing, traction bronchiectasis, considerable architectural distortion. Abnormality most marked peripherally and at lung bases.
f ipatand is
AJR:165,
October
1995
IDIOPATHIC
PULMONARY
FIBROSIS
when only one of these criteria was fulfilled. When a combination of criof possible IPF or possible HP was made depending on the dominant CT appearance. After reviewing the initial results of this study, we tested our sub-
tenia for IPF and HP was fulfilled, a diagnosis
jective
impression
that sparing
of the extreme
lung
bases
seen
on
CT scans could discriminate between HP and IPF. Two radiologists again reviewed the CT scans by consensus and recorded whether the extent of disease in the lower half of the lower lung zone (below the inferior pulmonary vein) was greater than, less than, or the same as that in the upper half of the lower lung zone.
StatisticalAnalysis Data were analyzed on a Macintosh SE-30 (Apple, Cupertino, CA) using the JMP-SAS statistical package (SAS Institute, Cary, NC). The ages of patients with HP and IPF, and the CT extent of ground-glass attenuation, were compared using Student’s ttest. The CT features of chronic HP and usual interstitial pneumonia and the sex distribution of patients with HP and IPF were compared using the Pearson’s chi-square test. Results Accuracy
of CT Diagnosis
The accuracy of CT diagnosis of IPF and HP is summanized in Table 1 A diagnosis was made with a high level of confidence (definite) in 39 (62%) of 63 patients. In these patients, the diagnosis was correct in 35 cases (90%): 23 of 26 patients with a CT diagnosis of IPF and 12 of 13 .
TABLE 1 : Accuracy of CT Diagnosis of Idiopathic Pulmonary Fibrosis (IPF) and Hypersensitivity Pneumonitis (HP) CT Diagnosis
Pathologic
Diagnosis
Correct
Level of Confidence
No. of Cases
IPF
Definite
26
23
3
88
IPF IPF
Probable Possible
6 2
4 1
2 1
66 50
HP HP HP
Definite Probable Possible
13 12 4
1 5 2
12 7 2
92 58 50
IPF
Fig. 3.-Desquamative interstitial pneumonia. 28-year-old man with biopsy-proven desquamative interstitial pneumonia. cT scan shows diffuse ground-glass attenuation.
Fig. 4.-Acute hypersensitivity pneumonitis, identical to desquamative interstitial pneumonia. 49-year-old woman with biopsy-proven acute hypersensitivity pneumonitis (bird-fancier’s lung). CT scan with patient in prone position shows widespread ground-glass attenuation indistinguishable from that seen in desquamative interstitial pneumonia. No micronodules are seen.
Diagnoses
(%)
*
& HYPERSENSITIVITY
PNEUMONITIS
809
patients with a CT diagnosis of HP. The degree of certainty of diagnosis of IPF was considerably higher than that of HP: 26 of 34 cases with a CT diagnosis of IPF were “definite” compared with 1 3 of 29 cases with a CT diagnosis of HP(p= .036). Of the 36 patients with IPF, eight had a misdiagnosis of HP, including all three patients with desquamative interstitial pneumonia and five of the 33 patients with usual interstitial pneumonia; the three patients with desquamative interstitial pneumonia had CT diagnoses of probable (n = 2) or definite HP (n = 1). Of the 27 patients with HP, 21 had the disease conrectly diagnosed, including all eight with acute or subacute HP and 13 ofthe 19 with chronic HP; six patients with chronic HP had the disease misdiagnosed, with CT diagnoses of definite (n = 3), probable (n = 2), or possible (n = 1 ) IPF.
CT Features Because the patients with acute and subacute HP had similar CT features and only two patients had acute HP, the patients with acute and subacute HP were grouped together for the purposes of analysis. The three patients with desquamative interstitial pneumonia and the eight acute on subacute HP all had widespread ground-glass attenuation (Figs. 3 and 4). Five
(63%)
patients
with
acute
on subacute
HP had
micro-
nodules, compared with one patient (33%) with desquamative interstitial pneumonia. No significant differences were seen between the CT features of those with acute on subacute HP and those with desquamative interstitial pneumonia. The CT features of patients with chronic HP and usual interstitial pneumonia are compared in Table 2. Patients with usual interstitial pneumonia were more likely to have honeycombing, traction bronchiectasis, and peripheral on lower zone predominance of disease and were less likely to have micronodules than were patients with chronic HP. Although lower zone predominance and peripheral predominance were significantly more common in patients with IPF than in patients with HP, these findings were also present in most patients with chronic HP (Fig. 5). However, the combination of lower zone predominance and peripheral predominance was present in only five patients with chronic HP. When the extent of disease in the lower half of the
LYNCH
810
ET AL.
AJR:165,
October
1995
Fig. 5.-Chronic hypersensitivity pneumonitis, identical to usual interstitial pneumonia. 48-year-old woman with biopsy-proven chronic hypersensitivity pneumonitis (tobacco-stripper’s lung). A and B, CT scans at two levels show patchy basal lung fibrosis with ground-glass attenuation, honeycombing, and traction bronchiectasis, indistinguishable from idiopathic pulmonary fibrosis.
I ..
TABLE 2: CT Features of Patients with Chronic Hypersensitivity Pneumonitis (HP) and Usual Interstitial Pneumonia (UIP)