[ Diffuse Lung Disease Original Research ]

Integration and Application of Radiologic

Patterns From Clinical Practice Guidelines
on Idiopathic Pulmonary Fibrosis and
Fibrotic Hypersensitivity Pneumonitis
Daniel-Costin Marinescu, MD; Cameron J. Hague, MD; Nestor L. Muller, MD; Darra Murphy, MD; Andrew Churg, MD;
Joanne L. Wright, MD; Amna Al-Arnawoot, MD; Ana-Maria Bilawich, MD; Patrick Bourgouin, MD; Gerard Cox, MD;
Celine Durand, MA; Tracy Elliot, MD; Jennifer Ellis, MD; Jolene H. Fisher, MD; Derek Fladeland, MD;
Amanda Grant-Orser, MD; Gillian C. Goobie, MD; Zachary Guenther, MD; Ehsan Haider, MD; Nathan Hambly, MD;
James Huynh, MD; Kerri A. Johannson, MD; Geoffrey Karjala, MD; Nasreen Khalil, MD; Martin Kolb, MD;
Jonathon Leipsic, MD; Stacey Lok, MD; Sarah MacIsaac, MD; Micheal McInnis, MD; Helene Manganas, MD;
Veronica Marcoux, MD; John Mayo, MD; Julie Morisset, MD; Ciaran Scallan, MD; Tony Sedlic, MD; Shane Shapera, MD;
Kelly Sun, MD; Victoria Tan, MD; Alyson W. Wong, MD; Boyang Zheng, MD; and Christopher J. Ryerson, MD

BACKGROUND: Clinical practice guidelines separately describe radiologic patterns of usual

interstitial pneumonia (UIP) and fibrotic hypersensitivity pneumonitis (fHP), without di-
rection on whether or how to apply these approaches concurrently within a single patient.
RESEARCH QUESTION: How can we integrate guideline-defined radiologic patterns to diagnose
interstitial lung disease (ILD) and what are the pitfalls associated with described patterns that
require reassessment in future guidelines?
STUDY DESIGN AND METHODS: Patients from the Canadian Registry for Pulmonary Fibrosis
underwent detailed reevaluation in standardized multidisciplinary discussion. CT scan fea-
tures were quantified by chest radiologists masked to clinical data, and guideline-defined
patterns were assigned. Clinical data then were provided to the radiologist and an ILD
clinician, who jointly determined the leading diagnosis.
RESULTS: Clinical-radiologic diagnosis in 1,593 patients was idiopathic pulmonary fibrosis
(IPF) in 26%, fHP in 12%, connective tissue disease-associated ILD (CTD-ILD) in 34%,
idiopathic pneumonia with autoimmune features in 12%, and unclassifiable ILD in 10%.
Typical and probable UIP patterns corresponded to a diagnosis of IPF in 66% and 57% of
patients, respectively. Typical fHP pattern corresponded to an fHP clinical diagnosis in
65% of patients, whereas compatible fHP was nonspecific and associated with CTD-ILD or
IPAF in 48% of patients. No pattern ruled out CTD-ILD. Gas trapping affecting > 5% of lung
parenchyma on expiratory imaging was an important feature broadly separating compatible
and typical fHP from other patterns (sensitivity, 0.77; specificity, 0.91).
INTERPRETATION: An integrated approach to guideline-defined UIP and fHP patterns is
feasible and supports > 5% gas trapping as an important branch point. Typical or probable
UIP and typical fHP patterns have moderate predictive values for a corresponding diagnosis
of IPF and fHP, although occasionally confounded by CTD-ILD; compatible fHP is
nonspecific. CHEST 2023; 164(6):1466-1475

KEY WORDS: hypersensitivity pneumonitis; idiopathic pulmonary fibrosis; interstitial lung

disease; multidisciplinary discussion; usual interstitial pneumonia

1466 Original Research [ 164#6 CHEST DECEMBER 2023 ]

 Idiopathic pulmonary fibrosis (IPF) and fibrotic
Take-home Points hypersensitivity pneumonitis (fHP) are two common
Study Question: How can we integrate guideline- subtypes of ILD, but have overlapping characteristics
defined radiologic patterns to diagnose interstitial that frequently make it difficult to establish a confident
lung disease (ILD) and what are the corresponding diagnosis, even in the setting of multidisciplinary
diagnoses and potential pitfalls associated with discussion. Clinical practice guidelines better
described patterns? standardize diagnosis by highlighting key clinical,
Results: Typical usual interstitial pneumonia (UIP), radiologic, and pathologic criteria suggestive of IPF and
probable UIP, and typical fibrotic hypersensitivity fHP.3-5 However, given the risks of lung biopsy and its
pneumonitis (fHP) are relatively specific in guiding declining use, accurate identification of specific imaging
clinicians toward a particular ILD diagnosis, whereas features and patterns fulfills an increasingly central role
no radiologic pattern confidently rules out connec- in ILD diagnosis.
tive tissue disease-related ILD. Gas trapping affecting
> 5% of lung parenchyma on expiratory imaging High-resolution CT (HRCT) imaging patterns of usual
helps to distinguish compatible and typical fHP interstitial pneumonia (UIP) and fHP each are
broadly from other patterns. described in separate guidelines, but without direction
Interpretation: Integrated application of idiopathic on whether or how to apply these approaches
pulmonary fibrosis and fHP guidelines to high- concurrently within a single patient. In addition,
resolution CT imaging is feasible, with > 5% lung criteria for these patterns are driven largely by expert
affected by gas trapping acting as a key branch point. consensus, often without rigorous data describing how
Typical UIP and fHP patterns are diagnostically these patterns perform or apply to real-world
helpful, but compatible fHP is nonspecific and may populations of patients with a variety of ILD subtypes.
require reevaluation in future guidelines. Therefore, we sought to integrate guideline-defined
UIP and fHP HRCT imaging patterns and to describe
the clinical diagnoses corresponding to each integrated
Fibrotic interstitial lung disease (ILD) is a heterogeneous pattern in a real-world multicenter cohort, while also
group of diseases whose classification relies on the identifying a practical, data-driven means of broadly
assessment of features from multiple disciplines.1,2 separating UIP and fHP patterns.

Study Design and Methods were eligible if they were aged $ 18 years and demonstrated fibrosis
Study Overview on HRCT imaging (# 2-mm sections) performed within 1 year before
Consecutive patients with fibrotic ILD of any subtype enrolled in seven of or up to 3 months after initial evaluation in an ILD clinic. Patients
eight centers in the prospective Canadian Registry for Pulmonary Fibrosis who experienced an acute exacerbation at the time of imaging or who
underwent detailed reevaluation in standardized multidisciplinary underwent surgical lung biopsy were excluded. The current work was
discussion between January 2021 and March 2022 (Fig 1).6,7 Patients approved by the research ethics board at each site.

ABBREVIATIONS: CTD-ILD = connective tissue disease-related inter- Montreal, the Département de Médecine (C. D., H. M., and J. M.),
stitial lung disease; fHP = fibrotic hypersensitivity pneumonitis; Centre de recherche du Centre hospitalier de l’Université de Montréal,
HRCT = high-resolution CT; ILD = interstitial lung disease; IPAF = Montréal, QC, the Department of Radiology (T. E., and Z. G.), the
interstitial pneumonia with autoimmune features; IPF = idiopathic Department of Medicine (A. G.-O. and K. A. J.), University of Calgary,
pulmonary fibrosis; NSIP = nonspecific interstitial pneumonia; UIP = Calgary, AB, the Department of Medical Imaging (D. F. and G. K.), the
usual interstitial pneumonia Department of Medicine (S. L. and V. M.), University of Saskatchewan,
AFFILIATIONS: From the Department of Medicine (D.-C. M., G. C. G., Saskatoon, SK, Canada; the Department of Radiology (D. M.), St
N. K., A. W. W., B. Z., and C. J. R.), the Department of Radiology (C. J. James’ Hospital, Dublin, Ireland; and the Division of Pulmonary, Al-
H., N. L. M., A.-M. B., J. E., J. L., J. M., and T. S.), the Department of lergy and Critical Care Medicine (G. C. G.), Department of Medicine,
Pathology (A. C., and J. L. W.), University of British Columbia, the University of Pittsburgh, Pittsburgh, PA.
Centre for Heart Lung Innovation (D.-C. M., G. C. G., A. W. W., B. Z., CORRESPONDENCE TO: Daniel-Costin Marinescu, MD; email: daniel.
and C. J. R.), St. Paul’s Hospital, Vancouver, BC, the Department of marinescu@vch.ca
Radiology (A. A.-A., E. H., J. H., and V. T.), the Department of Crown Copyright Ó 2023 Published by Elsevier Inc. under license from
Medicine (G. C., N. H., M. K., S. M., and C. S.), McMaster University, the American College of Chest Physicians.
Hamilton, the Department of Medicine (J. H. F., S. S., and K. S.), the
Department of Medical Imaging (M. M.), University of Toronto, DOI: https://doi.org/10.1016/j.chest.2023.07.068
Toronto, ON, the Department of Radiology (P. B.), University of

chestjournal.org 1467
 CARE-PF

1,593 consecutive patients

Recording of clinical details HRCT imaging available 1 y before or up

by independent reviewer to 3 mo after initial evaluation

Modified MDD with radiologist & respirologist

• Quantified radiographic features

Masked HRCT imaging analysis • Guideline-defined UIP & fHP pattern
• Radiologist-assigned pattern & confidence

Introduction of clinical information

Discussion • Clinical diagnosis & confidence

Figure 1 – Diagram showing study overview. CARE-PF ¼ Canadian Registry for Pulmonary Fibrosis; fHP ¼ fibrotic hypersensitivity pneumonitis;
HRCT ¼ high-resolution CT; MDD ¼ multidisciplinary discussion; UIP ¼ usual interstitial pneumonia.

Multidisciplinary Discussion 50% confidence were used in all subsequent analyses; instances
Key clinical and radiologic features of IPF and fHP were identified from where no single pattern or diagnosis had > 50% confidence were
clinical practice guidelines published at the time of study initiation labeled as having “no confident pattern” or unclassifiable ILD,
(e-Appendix 1, e-Table 1).3,5 A thorough and standardized chart respectively.8 A total of 14 ILD respirologists and 18 chest
review was undertaken to record key information available at the time radiologists participated in the review of cases, each of whom
of assessment in a specialty ILD clinic. This included clinical details initially were trained by the same core group of investigators leading
(demographics, family history of ILD, symptoms at presentation, the study. Detailed case examples illustrative of our methodology are
tobacco use history, comorbidities, exposures, previous ILD therapy, provided (e-Appendix 2, e-Figs 1-3).
disease course), laboratory findings (autoimmune serologic findings,
HRCT Imaging Assessment of Features
BAL cell count and differential), and pulmonary function testing (FVC
and diffusion capacity of lungs for carbon monoxide), wherever available. During masked HRCT imaging assessment, the radiologist visually
quantified to the nearest integer the percentage of lung
A modified multidisciplinary discussion then was performed for each parenchyma affected by honeycombing, isolated reticulation,
patient by an ILD respirologist and experienced chest radiologist reticulation with ground glass, isolated ground glass,
who did not participate in the chart review and initially were hypoattenuating lung, consolidation, and emphysema. The total
masked to all clinical data except age and sex. HRCT imaging amount of these features, when added to remaining normal
features were quantified by the chest radiologist as detailed herein. parenchyma, was mandated to sum to 100%. Percent of lung with
Each HRCT image then was categorized according to combinations gas trapping was estimated when expiratory imaging was available.
of patterns described in IPF and fHP guidelines, strictly adhering to Traction bronchiectasis was quantified as the percentage of lung
guideline definitions of each pattern (guideline-defined pattern, parenchyma fed by an airway exhibiting this feature. Categorical
described herein).3,5 The radiologist then provided a differential variables included the presence of asymmetry (defined as $
diagnosis of between one and five possible single radiologic patterns 50% more disease involvement in one lung compared with the
(eg, UIP, fHP, nonspecific interstitial pneumonia [NSIP], and so other lung),9 subpleural sparing, sparing of the extreme
forth) according to their own interpretation (radiologist-assigned costophrenic angle, three-density pattern, nodules, cysts, cavitation,
pattern), which could differ from the integrated guideline-defined pleural thickening, pleural plaques, pleural effusion, pericardial
pattern and in so doing represented any differences between expert effusion, humeral head erosion, axillary or mediastinal
radiologist gestalt and guideline criteria. Each radiologist-assigned lymphadenopathy (one or more nodes of $ 1 cm), and dilated
pattern was ascribed a confidence, all summing to 100%. Clinical esophagus (defined as the largest linear measurement of the
information then was introduced, and the respirologist and esophageal air column of > 1.5 cm on transverse section below the
radiologist discussed the case to generate up to five possible ventricles and above the diaphragm).10,11 Craniocaudal (basal,
consensus clinical diagnoses and diagnostic confidence for each, all upper, or diffuse or random) and axial (peripheral, central, both
summing to 100%.8 The radiologist-assigned pattern and peripheral and central, or diffuse or random) distributions were
multidisciplinary discussion-assigned clinical diagnosis with > recorded.

1468 Original Research [ 164#6 CHEST DECEMBER 2023 ]

 UIP Patterns

Typical Probable Indeterminate Alternative

Typical fHP
Typical (alternative UIP /
typical fHP)
fHP Patterns

Mutually exclusive
Compatible fHP
Compatible (alternative UIP /
compatible fHP)

Typical UIP Probable UIP Indeterminate Alternative UIP

Indeterminate (typical UIP / (probable UIP / (Indeterminate UIP / (alternative UIP /
indeterminate fHP) indeterminate fHP) indeterminate fHP) indeterminate fHP)

Typical UIP Probable UIP Indeterminate Compatible fHP Typical fHP Alternative UIP
(typical UIP / (probable UIP / (indeterminate UIP / (alternative UIP / (alternative UIP / (alternative UIP /
indeterminate fHP) indeterminate fHP) indeterminate fHP) compatible fHP) typical fHP) indeterminate fHP)

Integrated UIP/fHP Pattern

Figure 2 – Diagram showing an integrated approach to UIP and fHP radiologic patterns as defined in clinical practice guidelines.2,3,5 fHP ¼ fibrotic
hypersensitivity pneumonitis; UIP ¼ usual interstitial pneumonia.

Integration of Guideline-Defined Radiologic Patterns typical fHP), and (6) alternative (alternative UIP or indeterminate
When assigning radiologic patterns, concurrent application of the fHP).2 Analyses described herein are stratified by these categories.
guidelines generated mutually exclusive UIP and fHP pattern
combinations. Specifically, a typical or compatible fHP pattern Statistical Analysis
coexisted only with a pattern alternative to UIP because of significant Data are presented as mean  SD or number (percentage). Consensus
airways disease excluding all other UIP categories. Similarly, a typical or diagnoses are presented as the relative percentage of cases within each
probable UIP coexisted only with a pattern indeterminate for fHP in guideline-defined HRCT imaging pattern and are subdivided further
which excessive signs of airways disease were absent.3,5 Six integrated according to high or low diagnostic confidence ($ 70% and 51%-
guideline-defined UIP and fHP radiologic patterns therefore were 69% confidence, respectively).8 Cohen’s k coefficient was used to
possible (Fig 2): (1) typical UIP (typical UIP or indeterminate fHP), (2) describe the extent of agreement. A receiver operating characteristic
probable UIP (probable UIP or indeterminate fHP), (3) indeterminate analysis was performed to identify the amount of gas trapping that
(indeterminate UIP or indeterminate fHP), (4) compatible fHP best identifies an fHP pattern. Analyses were performed in R version
(alternative UIP or compatible fHP), (5) typical fHP (alternative UIP or 3.6.3 software (R Foundation for Statistical Computing).

Results Radiologist-Assigned Pattern and

Guideline-Defined Radiologic Pattern
Cohort Characteristics
We compared HRCT imaging patterns assigned by strict
A total of 1,593 consecutive patients were included adherence to guideline criteria with favored patterns
(Table 1). Stratifying by guideline-defined radiologic assigned by experienced chest radiologists (Fig 3A).
pattern, individuals with HRCT imaging patterns Most guideline-defined typical and probable UIP
favoring UIP were more often male and formerly used patterns were assigned a pattern of UIP based on
tobacco, whereas those with imaging that was radiologist gestalt (95% and 76%, respectively), with
indeterminate, suggestive of fHP or an alternative, more similar high consistency for radiologist-assigned fHP in
commonly were female. Mean FVC and diffusion patients with typical fHP based on guideline definition
capacity of lungs for carbon monoxide in the cohort (88%). In contrast, HRCT imaging compatible for fHP
were 76  20% and 55  19%, respectively, and constituted a highly variable group of patterns as
individuals with probable UIP or an indeterminate assessed by radiologists, favored to represent fHP (35%),
radiologic pattern showed less severe physiologic NSIP (26%), UIP (9%), or no confident pattern (29%).
characteristics. The indeterminate category also was highly variable,

chestjournal.org 1469
TABLE 1 ] Clinical Characteristics by Integrated Guideline-Defined UIP and fHP Radiologic Pattern
Guideline-Defined UIP and fHP Radiologic Pattern
Variable Total Typical UIP Probable UIP Indeterminate Compatible fHP Typical fHP Alternative
Total no. 1,593 326 (21) 212 (13) 100 (6) 171 (11) 164 (10) 620 (39)
Age, y 66  12 71  9 70  10 65  10 67  11 69  10 61  13
Male sex 800 (50) 233 (71) 128 (60) 43 (43) 66 (39) 65 (40) 265 (43)
Ethnicity
White 1,241 (78) 273 (84) 169 (80) 82 (82) 136 (80) 134 (82) 447 (72)
Asian 199 (12) 29 (9) 27 (13) 11 (11) 11 (6) 16 (10) 105 (17)
Black 19 (1) 4 (1) 1 (0) 1 (1) 1 (1) 2 (1) 10 (2)
American Indian 60 (4) 13 (4) 7 (3) 0 13 (8) 5 (3) 22 (4)
Native Hawaiian/ 14 (1) 2 (1) 1 (0) 3 (3) 0 1 (1) 7 (1)
Pacific
Islander
Missing 60 (4) 5 (2) 7 (3) 3 (3) 10 (6) 6 (4) 29 (5)
Family history of ILD 144 (9) 33 (10) 29 (14) 14 (14) 15 (9) 14 (9) 39 (6)
Tobacco use status
Current 77 (5) 21 (6) 5 (2) 4 (4) 8 (5) 5 (3) 34 (5)
Former 917 (58) 241 (74) 132 (62) 61 (61) 94 (55) 95 (58) 294 (47)
Never 598 (38) 64 (20) 75 (35) 35 (35) 69 (40) 64 (39) 291 (47)
FVC, % predicted 76  20 78  18 83  20 83  19 77  19 65  18 74  20
DLCO, % predicted 55  19 50  17 61  20 63  19 54  17 48  17 56  19

Data are presented as No. (%) or mean  SD. DLCO ¼ diffusion capacity for carbon monoxide; fHP ¼ fibrotic hypersensitivity pneumonitis; ILD ¼ interstitial
lung disease; UIP ¼ usual interstitial pneumonia.

with radiologists most often assigning favored patterns least common in typical and probable UIP (5% and 8%,
of NSIP (46%) or UIP (15%), with inability to generate a respectively). A subgroup of 122 patients (e-Appendix 3)
confident pattern in 37% of cases. Imaging believed to be underwent evaluations by two distinct multidisciplinary
alternative by guideline criteria commonly was identified discussion teams, with an unweighted k for diagnostic
as NSIP (60%). agreement of 0.61 with inclusion of all assigned
diagnoses, and a k of 0.80 with exclusion of patients
Consensus Clinical Diagnoses and
considered to have unclassifiable ILD by at least one of
Guideline-Defined Radiologic Pattern
the multidisciplinary discussion teams.
The most common diagnoses were connective tissue
disease-associated ILD (CTD-ILD; 34%), IPF (26%), fHP HRCT Imaging Features by Guideline-Defined
(13%), interstitial pneumonia with autoimmune features Radiologic Pattern
(IPAF; 12%), and unclassifiable ILD (10%) (Fig 3B). The extent of individual fibrotic features varied across
Typical and probable UIP patterns led to a diagnosis of guideline-defined HRCT imaging pattern (Fig 4). More
IPF in 66% and 57% of patients, respectively, with little honeycombing, reticulation, and traction bronchiectasis
difference in the composition of diagnoses between these was observed in typical UIP and typical fHP, whereas
two patterns. A typical fHP pattern corresponded to a the quantity of fibrosis was less in indeterminate and
clinical diagnosis of fHP in 65% of patients and virtually alternative patterns. Ground-glass and hypoattenuating
excluded IPF (1% of patients with typical fHP). lung were more extensive in typical and compatible fHP.
CTD-ILD or IPAF existed across all guideline-defined Expiratory scans were available for 855 patients (54%).
radiologic patterns (ranging from 22% in typical fHP to Gas trapping was most frequent in imaging compatible
63% in an alternative pattern) and constituted the most with and typical of fHP, with > 5% of lung affected by
frequent diagnosis when imaging was compatible with gas trapping being a highly specific threshold for
fHP (48%). Unclassifiable disease was present among all identifying either of these patterns (sensitivity, 77%;
imaging categories (ranging from 5% to 15%), but was specificity, 91%) (Fig 5). Some patients with an

1470 Original Research [ 164#6 CHEST DECEMBER 2023 ]

A Radiologist-Assigned Pattern by Integrated Guideline-Defined Radiologic Pattern
100

75
Percentage of Cases

50

25

0
Typical UIP Probable UIP Indeterminate Compatible fHP Typical fHP Alternative
Integrated Guideline-Defined Pattern
n= 326 212 100 171 164 620

Radiologist-Assigned Pattern
UIP - High Confidence UIP - Low Confidence fHP - High Confidence fHP - Low Confidence
NSIP - High Confidence NSIP - Low Confidence Other Pattern - High Confidence Other Pattern - Low Confidence
No Confident Pattern

B Final MDD Diagnosis by Integrated Guideline-Defined Radiologic Pattern

100

75
Percentage of Cases

50

25

0
Typical UIP Probable UIP Indeterminate Compatible fHP Typical fHP Alternative
Integrated Guideline-Defined Pattern
n= 326 212 100 171 164 620

Clinical Diagnosis
IPF - High Confidence IPF - Low Confidence fHP - High Confidence fHP - Low Confidence
CTD - ILD - High Confidence CTD - ILD - Low Confidence IPAF - High Confidence IPAF - Low Confidence
iNSIP - High Confidence iNSIP - Low Confidence Sarcoidosis - High Confidence Sarcoidosis - Low Confidence
Other ILD - High Confidence Other ILD - High Confidence Unclassifiable ILD

Figure 3 – A, Bar graph showing radiologist-assigned pattern stratified as low (51%-69%) or high ($ 70%) confidence compared with integrated
guideline-defined radiologic pattern. B, Bar graph showing MDD consensus diagnoses stratified as low (51%-69%) or high ($ 70%) confidence and
displayed as relative percentage of cases within each integrated guideline-defined radiologic pattern. CTD ¼ connective tissue disease; fHP ¼ fibrotic
hypersensitivity pneumonitis; ILD ¼ interstitial lung disease; IPAF ¼ idiopathic pneumonia with autoimmune features; IPF ¼ idiopathic pulmonary
fibrosis; iNSIP ¼ idiopathic nonspecific interstitial pneumonia; MDD ¼ multidisciplinary discussion; NSIP ¼ nonspecific interstitial pneumonia;
UIP ¼ usual interstitial pneumonia.

chestjournal.org 1471
Figure 4 – A-C, Graphs showing features present on A Features on Inspiratory HRCT imaging by Integrated Guideline-Defined Radiologic Pattern
inspiratory imaging (A), gas trapping assessed on
expiratory imaging (B), and bronchiectatic airways 100
(C) as percentage of lung involved, stratified by
integrated radiologic guideline-defined pattern.

Percent Lung Involved

80
fHP ¼ fibrotic hypersensitivity pneumonitis;
GGO ¼ ground-glass opacity; HRCT ¼ high- 60
resolution CT; UIP ¼ usual interstitial pneumonia.
40

20

0
Typical UIP Probable Indeterminate Compatible Typical fHP Alternative
UIP fHP
n= 326 212 100 171 164 620
Radiographic Features
Normal Lung Emphysema Consolidation Hypoattenuating Lung
GGO, pure Reticulation with GGO Reticulation, pure Honeycombing

B Air Trapping on Expiratory HRCT imaging by Integrated Guideline-Defined Radiologic Pattern

50
Percent Lung Involved

40

30

20

10

0
Typical UIP Probable Indeterminate Compatible Typical fHP Alternative
UIP fHP
n= 161 105 48 141 112 288
Gas Tapping

C Traction Bronchiectasis on Inspiratory HRCT imaging by Integrated Guideline-Defined

Radiologic Pattern

60
Percent Lung Involved

40

20

0
Typical UIP Probable Indeterminate Compatible Typical fHP Alternative
UIP fHP
n= 326 212 100 171 164 620
Integrated Guideline-Defined Pattern
Traction Bronchiectasis

alternative pattern for both UIP and fHP showed a small Discussion
degree of gas trapping, often representing concomitant Representing the largest ILD cohort to date with detailed
airways involvement overlying an NSIP pattern. Patients characterization of imaging features and patterns, our
with rare cases of typical and probable UIP showed study is the first to our knowledge to implement and
substantial gas trapping believed to be unrelated to the evaluate an integrated approach to proposed IPF and
underlying ILD, often in the setting of significant fHP HRCT imaging patterns.3-5 Our findings
emphysema. demonstrate the clinical usefulness of an integrated

1472 Original Research [ 164#6 CHEST DECEMBER 2023 ]

 Figure 5 – Receiver operating characteristic
1.00
curve showing gas trapping on expiratory CT
imaging as a predictor of guideline-defined
Gas trapping = 3% (Sensitivity 0.93, Specificity 0.75) typical or compatible fibrotic hypersensitivity
Gas trapping = 5% (Sensitivity 0.91, Specificity 0.78) pneumonitis.

Gas trapping = 6% (Sensitivity 0.77, Specificity 0.91)

0.75
Gas trapping = 10% (Sensitivity 0.74, Specificity 0.91)
Sensitivity

0.50

0.25

0
1.00 0.75 0.50 0.25 0
Specificity

approach to diagnosing ILD, to identifying areas of disease severity in IPF.17 However, probable UIP on
requiring refinement in updated clinical practice imaging in cases where the clinical likelihood of IPF is
guidelines, and to providing a template for future studies lower (eg, presence of an inciting antigen) may not
examining other major ILD patterns. correlate as well with pathologic UIP on biopsy18 and
may require careful consideration in a
We showed that a UIP pattern was relatively specific
multidisciplinary conference.
for IPF, with approximately 70% and 60% of cases
with typical and probable UIP imaging corresponding A typical fHP pattern similarly was specific,
to a final diagnosis of IPF. The remainder commonly corresponding to a confident clinical diagnosis of
represented CTD-ILD or IPAF, which generally are fHP in approximately 70% of patients, with the
straightforward to separate from IPF based on clinical remainder again largely representing CTD-ILD or
and laboratory features.12 This is consistent with IPAF. Prior studies have struggled to discern a
previous work demonstrating that UIP and probable consistent threshold of airway abnormality that
UIP HRCT imaging patterns predict UIP on histologic identifies fHP, partly reflecting the difficulty of
analysis and a diagnosis of IPF in > 95% and > quantifying this feature. Although some smaller
80% of patients when a clear alternative cause like studies describe a similar extent of airways disease in
CTD-ILD is absent.13-16 Although the similar IPF and fHP,19,20 larger cohorts suggest this remains
diagnostic composition of typical and probable UIP in an important distinguishing feature,15 with one study
our study offers rationale for merging these two finding that an increasing number of lobules
categories in guideline algorithms, we emphasize that exhibiting mosaic attenuation in $ 3 lobes increases
this may be appropriate only in cases where the specificity for fHP.21 In our cohort, we examined the
clinical likelihood for IPF is substantial (eg, older, usefulness of gas trapping in discerning patterns,
male, those who formerly used tobacco without rather than diagnoses, and identified > 5% lung
underlying CTD or an inciting antigen). Indeed, parenchyma affected as an optimal threshold for
evidence exists to indicate that honeycombing in IPF separating typical or compatible fHP from other
may be the result of end-stage traction bronchiectasis patterns, including UIP (sensitivity, 77%; specificity,
already present in less advanced disease, suggesting 91%). Although requiring validation, this threshold is
that probable and typical UIP represent a continuum based on the collective gestalt of 18 expert

chestjournal.org 1473
radiologists in the largest cohort to date and may be arises from having non-fHP diagnoses presenting
less cumbersome to apply than counting lobules with what is currently described as an fHP
within lobes, particularly when expiratory HRCT radiologic pattern.
imaging protocols lack contiguous images. Although
Evaluation of how imaging features described in
it is challenging to estimate the amount of lung
guidelines influence guideline-defined radiologic
parenchyma exhibiting gas trapping accurately to
patterns represents a form of incorporation bias;
single percentages, we suggest using the 5% number
however, our study provides an understanding of
as a guide, generally approximating < 5%, 5%-10%,
how experienced radiologists interpret and apply the
and > 10% gas trapping on expiratory CT scan
approaches described in recent clinical practice
imaging as being unlikely, possible, and likely
guidelines. Only 54% of individuals had expiratory
suggestive of a fHP pattern, respectively.
images available, which may impact our diagnosis of
In contrast to typical fHP, typical UIP, and probable fHP and analyses on gas trapping thresholds. If those
UIP, the pattern of compatible fHP showed a varied patients who are suspected to have fHP are
diagnostic composition that most often represented preferentially the ones undergoing expiratory
CTD-ILD or IPAF and less commonly led to fHP itself, imaging, the > 5% threshold may be an
IPF, and unclassifiable ILD. Although both typical and overestimation; however, this ensures a higher
compatible fHP require fibrosis with signs of small specificity of this threshold, more meaningfully
airways disease, the critical difference between these two separates UIP and fHP patterns based on extent of
patterns lies in disease distribution, with typical fHP signs of airways disease. Another limitation of our
being mid-lung predominant or diffuse craniocaudally study lies in each case having been read by a single
and diffuse axially (ie, both peripheral and central radiologist, although with a moderate k value (k ¼
components).5 Other distributions (eg, basal 0.61) for inter-multidisciplinary discussion agreement
predominant) lead to compatible fHP, which we show to in the final diagnosis for a subgroup of patients who
be less helpful diagnostically. This emphasis on underwent two separate evaluations and a high k
distribution represents a key difference between the 2020 value (k ¼ 0.80) when acknowledging major
American Thoracic Society/Japanese Respiratory uncertainty expressed by one of the multidisciplinary
Society/Asociación Latinoamericana del Tórax discussion teams. Despite attempts to standardize
hypersensitivity pneumonitis guideline, on which the measurement of features across radiologists, visual
current study is based,5 and the more recent 2021 quantification of imaging is imperfect and we cannot
CHEST guideline and expert panel report that places less eliminate interobserver variability.23-25 Furthermore,
emphasis on distribution.22 Our work suggests that we did not test the correlation of imaging features
consideration of distribution significantly alters the and diagnoses with pathologic findings for several
makeup of final clinical diagnoses and is justified in reasons, including the infrequent performance of
future guidelines. biopsies in contemporary ILD evaluation and the
challenges in central examination of biopsies from
Using fHP as a diagnosis and also a radiologic and
multiple centers. Our findings therefore should be
pathologic pattern creates terminological confusion,
generalized to the large majority of patients with ILD
which is worsened by the imperfect and poor
in whom a biopsy is not performed. Finally, we
specificities of typical and compatible fHP patterns,
evaluated patients based on data available at the time
respectively, for a diagnosis of fHP. This issue
of first assessment in an ILD clinic to reflect better
suggests the need for an alternative term that clearly
the initial diagnostic process, although imaging
distinguishes an fHP pattern from a clinical
features may change as disease progresses.
diagnosis of fHP. Because fHP patterns are defined
by signs of airways disease whose pathologic features Interpretation
and resultant imaging characteristics originate in This work represents the largest cohort in which recent
bronchioles, these may be labeled better as fibrotic clinical practice guidelines have been evaluated
bronchiolocentric interstitial pneumonia patterns. methodically, providing novel insights into how
This terminology would mirror the current guideline criteria are applied in a real-world population
framework used in UIP, eliminating confusion that and what changes are needed in future clinical practice

1474 Original Research [ 164#6 CHEST DECEMBER 2023 ]

guidelines. Future work should focus on validating Funding/Support
specific features that distinguish each pattern, The Canadian Registry for Pulmonary Fibrosis is
refinement of future guidelines to be more evidence sponsored by Boehringer Ingelheim. D.-C. M. was
based, and extension of our integrated model to other supported by a fellowship from the Canadian
radiologic patterns such as NSIP. Collectively, this work Pulmonary Fibrosis Foundation.
will generate an increasingly comprehensive,
reproducible, and clinically applicable framework for the Financial/Nonfinancial Disclosures
diagnosis of ILD. None declared.

Acknowledgments Fibrosis: design and rationale of a national 16. Chung JH, Chawla A, Peljto AL, et al. CT
pulmonary fibrosis registry. Can Respir J. scan findings of probable usual interstitial
Author contributions: D.-C. M. and C. J. R. 2016;2016:3562923. pneumonitis have a high predictive value
conceived the study. All authors contributed for histologic usual interstitial
to data acquisition. D.-C. M. conducted 7. Fisher JH, Kolb M, Algamdi M, et al.
Baseline characteristics and comorbidities pneumonitis. Chest. 2015;147(2):450-459.
statistical analysis. D.-C. M. and C. J. R.
in the CAnadian REgistry for Pulmonary 17. Ikezoe K, Hackett TL, Peterson S, et al.
interpreted the data and drafted the Fibrosis. BMC Pulm Med. 2019;19(1):223. Small airway reduction and fibrosis is an
manuscript. All authors reviewed and edited early pathologic feature of idiopathic
the final draft of the manuscript. 8. Ryerson CJ, Corte TJ, Lee JS, et al.
pulmonary fibrosis. Am J Respir Crit Care
A standardized diagnostic ontology for
Role of sponsors: Neither sponsor had input fibrotic interstitial lung disease. An Med. 2021;204(9):1048-1059.
on any aspect of this study. international working group perspective. 18. Tibana RCC, Soares MR, Storrer KM,
Am J Respir Crit Care Med. 2017;196(10): et al. Clinical diagnosis of patients
Additional information: The e-Appendixes, 1249-1254. subjected to surgical lung biopsy with a
e-Figures, and e-Tables are available online probable usual interstitial pneumonia
under “Supplementary Data.” 9. Tcherakian C, Cottin V, Brillet PY, et al.
pattern on high-resolution computed
Progression of idiopathic pulmonary
tomography. BMC Pulm Med. 2020;20(1):
fibrosis: lessons from asymmetrical
References disease. Thorax. 2011;66(3):226-231. 299.
1. Travis WD, Costabel U, Hansell DM, et al. 10. Schraufnagel DE, Michel JC, Sheppard TJ, 19. Hochhegger B, Sanches FD,
An official American Thoracic Society/ Saffold PC, Kondos GT. CT of the normal Altmayer SPL, et al. Air trapping in usual
European Respiratory Society statement: interstitial pneumonia pattern at CT:
esophagus to define the normal air
update of the international column and its extent and distribution. prevalence and prognosis. Sci Rep.
multidisciplinary classification of the AJR Am J Roentgenol. 2008;191(3): 2018;8(1):17267.
idiopathic interstitial pneumonias. Am J 748-752. 20. Okabayashi H, Fukuda T, Iwasawa T, et al.
Respir Crit Care Med. 2013;188(6): The new useful high-resolution computed
11. Takekoshi D, Arami S, Sheppard TJ, et al.
733-748. tomography finding for diagnosing
Computed tomography of the esophagus
2. Marinescu DC, Raghu G, Remy-Jardin M, in scleroderma and lung disease. Tohoku J fibrotic hypersensitivity pneumonitis:
et al. Integration and application of “hexagonal pattern”: a single-center
Exp Med. 2015;237(4):345-352.
clinical practice guidelines for the retrospective study. BMC Pulm Med.
diagnosis of idiopathic pulmonary fibrosis 12. Chung JH, Cox CW, Montner SM, et al. 2022;22(1):76.
and fibrotic hypersensitivity pneumonitis. CT features of the usual interstitial
pneumonia pattern: differentiating 21. Barnett J, Molyneaux PL, Rawal B, et al.
Chest. 2022;162(3):614-629. Variable utility of mosaic attenuation to
connective tissue disease-associated
3. Raghu G, Remy-Jardin M, Myers JL, et al. distinguish fibrotic hypersensitivity
interstitial lung disease from idiopathic
Diagnosis of idiopathic pulmonary pulmonary fibrosis. AJR Am J Roentgenol. pneumonitis from idiopathic pulmonary
fibrosis. An official ATS/ERS/JRS/ALAT 2018;210(2):307-313. fibrosis. Eur Respir J. 2019;54(1):1900531.
clinical practice guideline. Am J Respir 22. Fernandez Perez ER, Travis WD,
13. Raghu G, Mageto YN, Lockhart D,
Crit Care Med. 2018;198(5):e44-e68. Lynch DA, et al. Diagnosis and evaluation
Schmidt RA, Wood DE, Godwin JD. The
4. Raghu G, Remy-Jardin M, Richeldi L, accuracy of the clinical diagnosis of new- of hypersensitivity pneumonitis: CHEST
et al. Idiopathic pulmonary fibrosis (an onset idiopathic pulmonary fibrosis and guideline and expert panel report. Chest.
update) and progressive pulmonary 2021;160(2):e97-e156.
other interstitial lung disease: a
fibrosis in adults: an official ATS/ERS/ prospective study. Chest. 1999;116(5): 23. Watadani T, Sakai F, Johkoh T, et al.
JRS/ALAT clinical practice guideline. Am 1168-1174. Interobserver variability in the CT
J Respir Crit Care Med. 2022;205(9): assessment of honeycombing in the lungs.
14. Hunninghake GW, Zimmerman MB,
e18-e47. Radiology. 2013;266(3):936-944.
Schwartz DA, et al. Utility of a lung biopsy
5. Raghu G, Remy-Jardin M, Ryerson CJ, for the diagnosis of idiopathic pulmonary 24. Walsh SL, Calandriello L, Sverzellati N,
et al. Diagnosis of hypersensitivity fibrosis. Am J Respir Crit Care Med. Wells AU, Hansell DM, Consort UIPO.
pneumonitis in adults. An official ATS/ 2001;164(2):193-196. Interobserver agreement for the ATS/ERS/
JRS/ALAT clinical practice guideline. Am JRS/ALAT criteria for a UIP pattern on
15. Chung JH, Oldham JM, Montner SM,
J Respir Crit Care Med. 2020;202(3): CT. Thorax. 2016;71(1):45-51.
et al. CT-pathologic correlation of major
e36-e69.
types of pulmonary fibrosis: insights for 25. Widell J, Liden M. Interobserver
6. Ryerson CJ, Tan B, Fell CD, et al. The revisions to current guidelines. AJR Am J variability in high-resolution CT of the
Canadian Registry for Pulmonary Roentgenol. 2018;210(5):1034-1041. lungs. Eur J Radiol Open. 2020;7:100228.

chestjournal.org 1475