www.thelancet.com/respiratory Vol 2 April 2014 277

Articles

Diagnosis of idiopathic pulmonary fi brosis with high-resolution CT in patients with little or no radiological evidence of honeycombing: secondary analysis of a randomised, controlled trialGanesh Raghu, David Lynch, J David Godwin, Richard Webb, Thomas V Colby, Kevin O Leslie, Juergen Behr, Kevin K Brown, James J Egan, Kevin R Flaherty, Fernando J Martinez, Athol U Wells, Lixin Shao, Huafeng Zhou, Patricia S Pedersen, Rohit Sood, A Bruce Montgomery, Thomas G O’Riordan

SummaryBackground Present guidelines for the diagnosis of idiopathic pulmonary fi brosis require histological confi rmation of surgical lung biopsy samples when high-resolution CT images are not defi nitive for usual interstitial pneumonia. We aimed to assess the predictive value of high-resolution CT in a cohort of patients with suspected idiopathic pulmonary fi brosis from a previous randomised trial.

Methods ARTEMIS-IPF was a randomised, double-blind, placebo-controlled, multicentre, phase 3 trial of ambrisentan for adults aged 40–80 years with well-defi ned idiopathic pulmonary fi brosis and 5% or less honeycombing on high-resolution CT. In December, 2010, an interim analysis showed lack of effi cacy and the trial was stopped. In the present follow-on analysis, we assessed patients who were screened for ARTEMIS-IPF who underwent high-resolution CT as part of screening and surgical lung biopsy as part of standard clinical care. A radiologist and a pathologist from a central panel independently assessed anonymised CT scans and biopsy samples. We calculated the positive and negative predictive value of high-resolution CT (classifi ed as usual interstitial pneumonia, possible usual interstitial pneumonia, and inconsistent with usual interstitial pneumonia) for confi rmation of histological patterns of usual interstitial pneumonia. This study is registered with ClinicalTrials.gov, number NCT00768300.

Findings 315 (29%) of 1087 consecutively screened patients in ARTEMIS-IPF had both high-resolution CT and surgical lung biopsy samples. 108 of 111 patients who met high-resolution CT criteria for usual interstitial pneumonia had histologically confi rmed usual interstitial pneumonia (positive predictive value 97·3%, 95% CI 92·3–99·4), as did 79 of 84 patients who met high-resolution CT criteria for possible usual interstitial pneumonia (94·0%, 86·7–98·0). 22 of 120 patients had an inconsistent high-resolution CT pattern for usual interstitial pneumonia that was histologically confi rmed as not or possible usual interstitial pneumonia (negative predictive value 18·3%, 95% CI 11·9–26·4).

Interpretation In the appropriate clinical setting, for patients with possible usual interstitial pneumonia pattern on high resolution CT, surgical lung biopsy sampling might not be necessary to reach a diagnosis of idiopathic pulmonary fi brosis if high-resolution CT scans are assessed by experts at regional sites familiar with patterns of usual interstitial pneumonia and management of idiopathic interstitial pneumonia.

Funding Gilead Sciences.

IntroductionIdiopathic pulmonary fi brosis is characterised by worsening respiratory symptoms and pulmonary function and usually leads to death from respiratory failure within 5 years of diagnosis.1 Accurate diagnosis of the disease is essential to provide clinical guidance for treatment decisions, stratify enrolment into clinical trials, and establish prognosis and priority for lung allograft allocation. Diagnosis is made with the highest degree of confi dence when clinical, radiological, and pathological data are reviewed by a multidisciplinary discussion team, consisting of a clinician, a radiologist, and a pathologist who all have expertise in the diagnosis of interstitial lung diseases.1,2 Assessment of surgical lung biopsy samples in patients with idiopathic pulmonary fi brosis will show a

pattern classifi ed as usual interstitial pneumonia with increased diagnostic accuracy if specimens are obtained from more than one lobe.3 With the advent of high-resolution CT, characteristic patterns of abnormalities have been identifi ed in patients with idiopathic pulmonary fi brosis.4–6 Evidence-based guidelines provide specifi c criteria for the radiological pattern of usual interstitial pneumonia and, in the appropriate clinical setting, histological confi rmation is not needed when a confi dent diagnosis is made with high-resolution CT.1,7 The usual interstitial pneumonia pattern on high-resolution CT has a high positive predictive value for the presence of a histological pattern of usual interstitial pneumonia. The appropriate clinical setting is defi ned as, typically, a male patient older than 60 years with unexplained dyspnoea on

Lancet Respir Med 2014; 2: 277–84

Published OnlineFebruary 18, 2014http://dx.doi.org/10.1016/S2213-2600(14)70011-6

See Comment page 249

University of Washington, Seattle, WA, USA (Prof G Raghu MD, Prof J D Godwin MD); University of Colorado, Denver, CO, USA (Prof D Lynch MD, Prof K K Brown MD); University of California, San Francisco, CA, USA (Prof R Webb MD); Mayo Clinic, Scottsdale, AZ, USA (Prof T V Colby MD, Prof K O Leslie MD); Klinikum der Ludwig-Maximilians-Universität Munich, Munich, Germany (Prof J Behr MD); Mater Misericordiae Hospital, Dublin, Ireland (Prof J J Egan MD); University of Michigan, Ann Arbor, MI, USA (Prof K R Flaherty MD, Prof F J Martinez MD); Royal Brompton Hospital, Chelsea, London, UK (Prof A U Wells MD); Gilead Sciences, Seattle, WA, USA (L Shao MD, H Zhou PhD, P S Pedersen BSN, T G O’Riordan MD); Perceptive Informatics, Boston, MA, USA (R Sood MD); and Cardeas Pharma, Seattle, WA, USA (A B Montgomery MD)

Correspondence to: Prof Ganesh Raghu, Center for Interstitial Lung Diseases, University of Washington Medical Center, 1959 North East Pacifi c, Campus Box 356175, Seattle, WA 98195, USAgraghu@u.washington.edu

Articles

278 www.thelancet.com/respiratory Vol 2 April 2014

exertion and pulmonary fi brosis of unknown cause, including absence of collagen vascular disease and absence of identifi able cause or environmental factor attributable to the manifested pulmonary fi brosis.1 To classify a high-resolution CT as showing usual interstitial pneumonia, honey combing and reticular abnormalities in a predomi nantly basilar and subpleural distribution need to be present. In addition, the high-resolution CT should not show any atypical features for usual interstitial pneumonia, including prominent ground glass abnormalities, nodules, air trapping, and peribroncho-vascular or upper lung predominant patterns of distribution. If such atypical features are prominent, the high-resolution CT pattern is designated as inconsistent with usual interstitial pneumonia1 and procurement of a surgical lung biopsy should be considered. The guidelines recognise a third potential radiological diagnosis: possible usual interstitial pneumonia pattern, which is characterised by the presence of bilateral lower lobe subpleural reticular abnormalities without honeycombing and without atypical features. In patients with this pattern, the guidelines1 require surgical lung biopsy evidence of usual interstitial pneumonia to confi rm the diagnosis of idiopathic pulmonary fi brosis.

We aimed to assess the value of high-resolution CT showing a possible usual interstitial pneumonia pattern for prediction of a histological pattern consistent with usual interstitial pneumonia in a subgroup of participants who were screened for inclusion in a multicentre trial8 of ambrisentan.

MethodsStudy design and participantsARTEMIS-IPF was a randomised, double-blind, placebo-controlled, multicentre, event-driven phase 3 trial8 to assess whether the selective endothelin receptor antagonist ambrisentan was eff ective for reduction of the rate of progression of idiopathic pulmonary fi brosis. All participants provided written, informed consent, and the study was approved by local ethics committees. After 1087 patients had been screened between Jan 21, 2009, and Dec 21, 2010, and 494 patients randomly allocated to study drugs, the sponsor terminated the study prematurely after an interim analysis showed lack of effi cacy.8

ARTEMIS-IPF was started before publication of the 2011 ATS-ERS-JRS-ALAT evidence-based guidelines for the diagnosis and management on idiopathic pulmonary fi brosis.1 The criteria used for diagnostic classifi cation by the radiology charter for the study were equivalent to those of current practice guidelines, however the terminology used was slightly diff erent. To minimise potential confusion and to ensure consistency of this report with the conventions likely to be adopted by future publications on this topic, we amended our charter terms to match those in recently published guidelines. The ARTEMIS-IPF charter term “defi nite idiopathic pulmonary fi brosis” was therefore changed to “usual

interstitial pneumonia”, the term “consistent with idiopathic pulmonary fi brosis” was changed to “possible usual interstitial pneumonia”, and the term “inconsistent with idiopathic pulmonary fi brosis” was changed to “inconsistent with usual interstitial pneumonia”. The histopathological nomenclature was the same in both the ARTEMIS-IPF charter and in the 2011 published guidelines, although the ARTEMIS-IPF charter did not include a category of unclassifi able fi brosis.

Eligible participants were aged 40–80 years, and had clinically well-defi ned idiopathic pulmonary fi brosis for at least 3 months. Participants had undergone assessment at community and academic clinical centres for their clinical manifestation of pulmonary fi brosis and had been diagnosed as idiopathic pulmonary fi brosis, and were thus referred for consideration of participation in the clinical trial.1 The study was done at 136 sites in North America, South America, western Europe, and Australia. High-resolution CT was a required study procedure but surgical lung biopsy samples were obtained as part of standard care. As a part of screening to assess eligibility for participation in the clinical trial, all of the high-resolution CT images were reviewed by one expert chest radiologist (DL) from a centralised panel of radiologists (DL, JDG, RW) and the surgical lung biopsy slides were reviewed by one expert pathologist (TVC) from a central panel of pathologists (TVC, KOL). The expert chest radiologist and the expert pathologist independently interpreted the radiological and histological features from all 315 patients with both high resolution CT and surgical lung biopsy data who were screened for enrolment. We used precise prespecifi ed criteria in case report forms to interpret the patterns of usual interstitial pneumonia and each investigator was masked to the results of the corollary interpretation. The decision of one expert from the panel was regarded as fi nal in all cases.

On the basis of a previous study,9 patients with 5% or less honeycombing on high-resolution CT images of the lung were postulated to respond to endothelin-receptor blockade and thus patients with more than 5% honeycombing on the high-resolution CT were excluded. Other key exclusion criteria were congestive heart failure, history of collagen vascular disease, evidence of occupational lung disease, evidence of coexisting obstructive airfl ow defect or prominent emphysema on high-resolution CT (defi ned as emphysema more prominent than reticular abnormality), recent hospital admission or respiratory infection within 60 days, chronic treatment for pulmonary hypertension, and chronic treatment with immunosuppressive therapy. Of 1087 participants screened for ARTEMIS-IPF, 315 had both high-resolution CT images and surgical lung biopsies submitted for assessment by central reviewers, and data from that cohort were analysed in this study (appendix). Therefore the cohort for the primary analysis in our present study included both a subset of parti-cipants eligible for ARTEMIS-IPF and some individuals

See Online for appendix

Articles

www.thelancet.com/respiratory Vol 2 April 2014 279

who were screened but were ineligible for ARTEMIS-IPF. Analysed participants classifi ed as possible usual interstitial pneumonia by high-resolution CT had, by defi nition, no honeycombing on high-resolution CT. Although participants in the usual interstitial pneumonia and inconsistent high-resolution CT categories, who were eligible for ARTEMIS-IPF had less than 5% honey-combing, these categories also included ineligible participants with more than 5% honeycombing.

ProceduresCentralised interpretation and obtaining of radiological images was supported by a clinical trial service provider (Perceptive Informatics, Billerica, MA, USA). We acquired three sets of axial spaced high-resolution CT images for all patients (supine inspiration, supine expiration, and prone inspiration), anatomically extending from lung apices to costophrenic recess (appendix).

We used prespecifi ed criteria for categorisation of high-resolution CT images into usual interstitial pneumonia, possible usual interstitial pneumonia, and inconsistent with usual interstitial pneumonia (panel 1). Radiologists were also asked to determine if two other exclusion criteria were met: an extent of honeycombing involving more than 5% of the lung area and presence of prominent emphysema, defi ned as emphysema aff ecting an area of more than the area aff ected by the reticular pattern.

High-resolution CT images were anonymised by a third-party vendor (Perceptive Informatics) and fi lms were shown to one of the three radiologists at the time of screening for study enrolment. The radiology reader interpreted the images with a standardised scoring sheet.

Radiologists were provided high-resolution CT scans for assessment of patterns and distribution to gauge eligibility status for participation in ARTEMIS-IPF study, but did not have access to other clinical or histological data. The radiologists had a face-to-face training session before study start to agree on the scoring criteria and jointly reviewed non-study standard cases. However, once the study was started, no discussion was undertaken of the eligibility of individual study participants between radiologists or with the study site or sponsor.

For histopathological analysis, four categories were used to interpret the microscopic features in the surgical lung biopsy (usual interstitial pneumonia, probable usual interstitial pneumonia, possible usual interstitial pneumonia, and not usual interstitial pneumonia).1 In the fi rst draft of the protocol, patients classifi ed with possible usual interstitial pneumonia were eligible for inclusion, but after consultation with regulatory authorities, participants with possible fi ndings were regarded as ineligible to ensure recruitment of a well-defi ned study population. For analysis in the present study, we consolidated surgical lung biopsy results into two categories: defi nite and probable fi ndings were grouped as usual interstitial pneumonia whereas possible and not usual fi ndings were grouped as not usual

interstitial pneumonia. Slides were anonymised before review by the pathologists (TVC or KOL). Pathologists were provided slides from the surgical lung biopsy for interpretation of the histopathological patterns to gauge eligibility status for participation in the ARTEMIS-IPF study, but did not have access to other clinical or radiological data.

Statistical analysis We calculated positive predictive values for patterns consistent with usual interstitial pneumonia and possible usual interstitial pneumonia on high-resolution CT for a

Panel 1: High-resolution CT diagnostic criteria1

Usual interstitial pneumonia pattern (all of the following four features)• Subpleural, basal predominance• Reticular abnormality• Honeycombing (<5% of lung area)• No features listed as inconsistent with usual interstitial

pneumonia pattern

Possible usual interstitial pneumonia pattern (all of the following four features)• Subpleural, basal predominance• Reticular abnormality• No honeycombing• No features listed as inconsistent with usual interstitial

pneumonia pattern

Inconsistent with usual interstitial pneumonia (any of the following seven features)• Upper of mid-zone predominance• Peribronchovascular predominance• Extensive ground glass abnormality (ie, greater in extent

than reticular abnormalities)• Prominent micronodules• Discrete cysts (multiple, bilateral away from area of

honeycombing)• Diff use mosaic attenuation or air trapping• Consolidation of bronchopulmonary segments

Figure 1: Study profi le

1087 screened for ARTEMIS–IPF

111 no high-resolution CT

661 no biopsy sampling

976 had high-resolution CTsubmitted for central review

315 had a surgical lung biopsy sampling111 usual interstitial pneumonia on high-resolution CT

84 possible usual interstitial pneumonia on high-resolution CT120 high-resolution CT inconsistent with usual interstitial pneumonia

Articles

280 www.thelancet.com/respiratory Vol 2 April 2014

histological diagnosis of usual interstitial pneumonia and probable usual interstitial pneumonia. We also calculated the negative predictive value of an inconsis tent with usual interstitial pneumonia fi nding on high-resolution CT for a histological diagnosis of possible usual interstitial pneumonia or not usual interstitial pneumonia. We used the Clopper-Pearson method to calculate 95% CIs. We did not calculate sensitivity and specifi city of high-resolution CT because the study population was enriched for patients with possible usual interstitial pneumonia and thus the true prevalence of disease in patients with high-resolution CT patterns not of this type was not known. We used SAS version 9.2 for analyses.

This study is registered with ClinicalTrials.gov, number NCT00768300.

Role of the funding sourceGilead Sciences funded both the original ARTEMIS-IPF trial and this study. TGO’R is an employee of Gilead Sciences and cowrote the fi rst draft of the manuscript with GR. LS and HZ are employees of the funding source and were responsible for statistical analysis. LS, HZ, TGO’R, and PSP had full access to raw data. TGO’R and GR had the fi nal responsibility for the decision to submit to publication.

ResultsOf 1087 consecutive participants screened for enrolment in ARTEMIS-IPF study (fi gure 1), 315 patients had both a high-resolution chest CT (a required study procedure) and a surgical lung biopsy sampling (not a required

study procedure but obtained as part of standard clinical care before enrolment) that was interpreted by the local radiologists and pathologists. Figure 2 shows representative examples of high-resolution CT images. Six screened participants had surgical lung biopsy data but not high-resolution CT data; one of these biopsy samples showed defi nite usual interstitial pneumonia, one showed possible usual interstitial pneumonia with features of hypersensitivity pneumonia, three had other diagnoses (one showed hypersensitivity pneumonia and two non-specifi c intestinal pneumonia), and one was a non-diagnostic transbronchial lung biopsy submitted in error.

30 (10%) of 315 surgical lung biopsy samples were interpreted as not usual interstitial pneumonia or possible usual interstitial pneumonia (table 1). Of the 30 surgical lung biopsy samples that did not confi rm usual interstitial pneumonia, suggested alternative diagnoses included hypersensitivity pneumonia (11 patients), non-specifi c intestinal pneumonia (seven patients), desquamative interstitial pneumonia (one patient), cryptogenic organising pneumonia (two patients), non-specifi c airway disease (three patients), adenocarcinoma (one patient), or inadequate sampling to suggest a diagnosis (fi ve patients).

Demographics and baseline lung function did not diff er between groups based on high-resolution CT diagnosis or availability of surgical lung biopsy samples (table 2). Despite the inclusion requirement of less than 5% honeycombing on high-resolution CT, we noted a wide range of impairment of lung function in all groups. Overall mean forced vital capacity was 69·8% (SD 15·6)

A CB

D FE

Figure 2: High-resolution axial CT images of upper and lower lung(A–B) Usual interstitial pneumonia; images show peripheral predominant, basal predominant reticular abnormality with mild subpleural honeycombing (B, arrows); extent of honeycombing is <5%. (C–D) Possible usual interstitial pneumonia; images show peripheral predominant, basal predominant reticular abnormality without honeycombing. (E–F) Inconsistent pattern of usual interstitial pneumonia; images show upper lung predominant abnormality characterised by ground glass abnormality with mild reticular abnormality.

Articles

www.thelancet.com/respiratory Vol 2 April 2014 281

and diff usion capacity was 42·6% (17·3) of predicted, suggesting that many of the screened patients had strikingly impaired lung function.

Table 1 shows the positive predictive value of a radiological diagnosis of usual interstitial pneumonia or possible usual interstitial pneumonia and the negative predictive value of radiological diagnosis of inconsistent with usual interstitial pneumonia. Table 1 also shows data for the four subcategories of histological diagnosis. The histological pattern of usual interstitial pneumonia was 3–4 times more common than was the pattern of probable usual interstitial pneumonia in this cohort of patients.

In a post-hoc analysis of the 120 participants with a high-resolution CT diagnosis of fi ndings inconsistent with usual interstitial pneumonia, we assessed diff erences in baseline characteristics between 98 patients with histologically confi rmed or probable usual interstitial pneumonia and 22 patients with histological fi ndings of possible or not usual interstitial

pneumonia. Groups did not diff er at baseline in terms of mean percentage predicted forced vital capacity, haemoglobin-adjusted diff usion capacity, or sex, although patients who were subsequently histologically diagnosed with possible or not usual interstitial pneumonia had a mean age of 58·2 years (SD 10·6) compared with 63·2 years (7·6) for patients histologically diagnosed with confi rmed or probable usual interstitial pneumonia (p=0·0525; appendix).

DiscussionFor assessment of patients with interstitial lung disease, presence of unambiguous honeycombing on a high-resolution CT has a valuable positive predictive value for usual interstitial pneumonia.1,3–7,10–12 When honeycombing is not present or minimal, diagnosis of idiopathic pulmonary fi brosis is challenging. Present guidelines recommend that patients suspected to have idiopathic pulmonary fi brosis with possible usual interstitial pneumonia pattern on high resolution CT also have a

Patients with a concordant histological diagnosis

Patients without a concordant histological diagnosis

Positive predictive value Negative predictive value

UIP on high-resolution CT (n=111) 108 (84 UIP, 24 probable UIP) 3 (1 possible UIP, 2 not UIP) 108/111 (97·3%, 95% CI 92·3–99·4)* ··

Possible UIP on high-resolution CT (n=84) 79 (65 UIP, 14 probable UIP) 5 (4 possible UIP, 1 not UIP) 79/84 (94·0%, 95% CI 86·7–98·0)* ··

Inconsistent with UIP on high-resolution CT (n=120) 22 (10 possible UIP, 12 not UIP) 98 (76 UIP, 22 probable UIP) ·· 22/120 (18·3%, 95% CI 11·9–26·4)*

UIP=usual interstitial pneumonia. *Clopper-Pearson 95% CI.

Table 1: Diagnostic value of high-resolution CT

Usual interstitial pneumonia on high-resolution CT

Possible usual interstitial pneumonia on high-resolution CT

Inconsistent with usual interstitial pneumonia on high-resolution CT

With SLB (n=111) Without SLB (n=388) With SLB (n=84) Without SLB (n=102) With SLB (n=120) Without SLB (n=171)

Age, years

Mean (SD) 64·2 (8·0) 68·0 (6·9) 64·4 (7·4) 67·1 (7·9) 62·3 (8·4) 64·7 (9·1)

Median (IQR, range) 66 (60–70, 35–79) 69 (63–73, 48–87) 65 (59–70, 47–78) 68 (61–73, 49–80) 63 (56–69, 39–81) 65 (58·5–73, 40–81)

Sex

Male 86/111 (77%) 278/387 (72%) 64/84 (76%) 73/102 (72%) 73/120 (61%) 85/169 (50%)

Female 25/111 (23%) 109/387 (28%) 20/84 (24%) 29/102 (28%) 47/120 (39%) 84/169 (50%)

Smoking status

Current or ex-smoker 60/93 (65%) 185/270 (69%) 48/68 (71%) 30/51 (59%) 60/87 (69%) 19/38 (50%)

Never smoker 33/93 (35%) 85/270 (31%) 20/68 (29%) 21/51 (41%) 27/87 (31%) 19/38 (50%)

Predicted forced vital capacity (%)

Patients assessed 105 331 78 74 106 101

Mean (SD) 69·6% (14·7) 70·2% (15·2) 69·3% (12·3) 71·4% (15·8) 68·2% (13·9) 66·9% (13·8)

Median (IQR, range) 68·6% (58·5–77·9, 40·0–105·4)

68·9% (59·9–79·2, 34·0–131·1)

70·3% (59·8–77·8, 40·8–100·5)

70·2% (57·0–81·4, 43·6–115·8)

66·2% (58·9–75·1, 42·4–113·2)

66·5% (55·9–76·0, 39·5–98·4)

Predicted haemoglobin-adjusted DLCO

Patients assessed 103 309 77 68 103 81

Mean (SD) 43·7% (14·9) 42·4% (18·8) 44·4% (15·4) 44·7% (18·0) 44·3% (14·4) 39·2% (18·6)

Median (IQR, range) 42·0% (33·9–51, 13·1–101·7)

41·5% (31·5–51·4, 0·0–242·4)

41·9% (35·1–54·2, 17·9–103·9)

44·6% (34·7–55·6, 0·0–101·1)

42·9% (34·4–51·3, 15·7–106·1)

37·7% (28·5–48·5, 0·0–86·8)

Data are n/n with available data. Data for 976 patients who had high-resolution CT data submitted during ARTEMIS-IPF screening. DLCO=diff usion capacity.

Table 2: Baseline characteristics stratifi ed by availability of surgical lung biopsy (SLB)

Articles

282 www.thelancet.com/respiratory Vol 2 April 2014

surgical lung biopsy taken to confi rm diagnosis from histopathological features.1 Surgical lung biopsy is an invasive procedure associated with morbidity and mortality, especially in elderly patients with interstitial lung disease.13,14 Patients with the disease are usually older than 60 years15 and often have comorbidities, and thus an accurate diagnosis of idiopathic pulmonary fi brosis without the need for surgical lung biopsy is desirable. Our study suggests that, if the following two key conditions are met, radiological diagnosis of possible usual interstitial pneumonia is suffi cient for diagnosis of idiopathic pulmonary fi brosis to be made without surgical lung biopsy. The fi rst condition is that the clinical presentation and demographics are typical of idiopathic pulmonary fi brosis and were ascertained by an expert in interstitial lung disease. Patients should have a thorough history taken for possible exposure to overt and occult factors in the patient’s environment and lack clinical and serological features of collagen vascular diseases. The second condition is that high-resolution CT images, assessed by a radiologist with expertise in interstitial lung disease, show a subpleural basal predominant reticular abnormality with the absence of atypical features for usual interstitial pneumonia pattern (panel 1). With these conditions, absence of honeycombing in the high-resolution CT images of the chest should not always, in isolation, mandate a lung biopsy to establish a diagnosis of idiopathic pulmonary fi brosis.

The term possible usual interstitial pneumonia has a precise radiological criteria and pattern.1 In an earlier classifi cation, the term possible usual interstitial pneumonia specifi cally applied to patients with high-resolution CT that showed both an absence of honeycombing and atypical fi ndings.6 Studies assessing the predictive value of high-resolution CT images in patients with idiopathic interstitial lung disease have tended to group patients whose CT images are otherwise consistent with usual interstitial pneumonia (apart from absent honeycombing) with those who have atypical features.3–6,7,10,11 To our knowledge, no previous studies have assessed large numbers of patients with suspected idiopathic pulmonary fi brosis after expert clinical assessment by an interstitial lung disease clinician and multidisciplinary discussion, high-resolution CT images that are consistent with a possible usual interstitial pneumonia pattern, no atypical fi ndings on high-resolution CT, and alternative diagnoses to usual interstitial pneumonia made on histological analyses of surgical lung biopsy samples. Nevertheless, the results of our retrospective analyses of data gathered prospectively from patients screened for the ARTEMIS-IPF study complement those of a recent retrospective single-centre study16 of 135 patients who had surgical lung biopsy samples obtained for diagnosis of idiopathic interstitial lung disease (panel 2). Use of a semiquantitative scoring system for fi brosis accurately diagnosed usual interstitial pneumonia in the absence of honeycombing in patients

aged 55 years and older, without prespecifi cation of atypical radiological fi ndings as an exclusion criteria.16 The investigators suggested that older age was a key factor predictive of usual interstitial pneumonia and the mean age in our analysis supports these fi ndings. In another recent study,17 retrospective diagnosis of idiopathic pulmonary fi brosis could be made in 21 patients on the basis of high-resolution CT criteria in the absence of honeycombing in 44 cases of biopsy-proven usual interstitial pneumonia in the appropriate clinical setting.17

The focus of our study was on patients with possible usual interstitial pneumonia on high-resolution CT rather than on patients with CT fi ndings that were inconsistent with usual interstitial pneumonia (who, by defi nition, have atypical fi ndings on high-resolution CT. Any such patients who had an alternative diagnosis established by surgical lung biopsy would not have been referred to this study). Our fi ndings in patients with possible usual interstitial pneumonia on high resolution CT are pertinent and clinically relevant because the diagnostic criteria for possible usual interstitial pneumonia are much narrower than they are for the inconsistent pattern and patients in this group are more likely than those with inconsistent fi ndings to undergo surgical lung biopsy in clinical practice to clarify or confi rm the diagnosis of idiopathic pulmonary fi brosis because they had little or no honeycombing on high-resolution CT. The 2011 guidelines1 for diagnosis of idiopathic pulmonary fi brosis emphasise the exclusion of known causes or conditions associated with pulmonary fi brosis, the need for multidisciplinary discussions, and recommend referrals to regional centres with expertise in interstitial lung disease for patients similar to the cohorts enrolled in our study.

Although the focus of our study was on the predictive value of a radiological diagnosis of possible usual interstitial pneumonia, 98 (82%) of 120 patients with an inconsistent with usual interstitial pneumonia radiological pattern had either usual interstitial pneumonia or probable usual interstitial pneumonia on surgical lung biopsy. The study design restricts extrapolation to a wider population and warrants further investigations in future studies. Our fi ndings complement results from a previous single centre prospective study,18 which noted a high specifi city of a radiological diagnosis of idiopathic pulmonary fi brosis (90%) but a fairly low sensitivity (78·5%).

If an inconsistent with usual interstitial pneumonia radiological pattern is noted in patients after thorough assessment of other causes of interstitial lung disease, the possibility of the diagnosis of idiopathic pulmonary fi brosis still exists and under these circumstances surgical lung biopsy evidence of usual interstitial pneumonia is needed to ascertain the diagnosis. Such a procedure is in keeping with the 2011 published guidelines because the combination of the histopathological evidence of usual interstitial pneumonia in the patient showing radiological diagnosis of inconsistent pattern for usual interstitial

Articles

www.thelancet.com/respiratory Vol 2 April 2014 283

pneumonia allows the diagnosis of idiopathic pulmonary fi brosis in the appropriate clinical setting. Assessment is warranted of whether any of the seven components characterising the inconsistent pattern (panel 1) have a high or unique negative predictive value for histo-pathological features of usual interstitial pneumonia in this population. For example, some criteria such as ground glass attenuation have been well characterised in previous studies16 but others such as mosaic attenuation in exhalation images might be worthy of further exploration.18 Similarly, whereas the distinction between usual interstitial pneumonia, idiopathic pulmonary fi brosis, and non-specifi c interstitial pneumonia has been well studied,16 diff erentiation between a diagnosis of chronic hypersensitivity pneumonia and idiopathic pulmonary fi brosis continues to pose a challenge to expert clinicians.19 In a recent prospective study, diagnosis of chronic hypersensitivity pneumonia was ascertained when patients were re-evaluated in more than 40% of patients presenting with features of idiopathic pulmonary fi brosis.20 This fi nding is in keeping with the 2011 guidelines that emphasise the need to exclude environmental and other causes associated with pulmonary fi brosis and for multidisciplinary discussions among expert clinicians, radiologists and pathologists familiar with interstitial lung disease to confi rm a diagnosis of idiopathic pulmonary fi brosis. The need for such discussions is further emphasised when there are discordances between the radiological and histopatho logical criteria.

Our study had limitations. Patients were referred to have a surgical lung biopsy procedure only on the basis of the clinical judgment of the primary physician and the decision was made before screening and independent from the study. Patients with defi nite usual interstitial pneumonia on high-resolution CT were eligible to enter ARTEMIS-IPF without undergoing a surgical lung biopsy procedure and were therefore not included in our analysis. ARTEMIS-IPF’s inclusion criteria of less than 5% honeycombing also enriched the population with a radiological diagnosis of possible usual interstitial pneumonia, the true sensitivity and specifi city of high-resolution CT images for a histological diagnosis of usual interstitial pneumonia in the general population with features of idiopathic pulmonary fi brosis and idiopathic interstitial pneumonias cannot be calculated from our dataset.

If patients with a high-resolution CT pattern of possible usual interstitial pneumonia had an alternative histological diagnosis, they would not have been referred for participation in this study. 102 patients who were referred with possible usual interstitial pneumonia on high-resolution CT were not eligible to participate in the trial because they had not had a biopsy sample taken. Whether alternative diagnoses would have been made for some of these patients is not known. However, only fi ve of 84 patients with possible usual interstitial pneumonia on high-resolution CT did not have a

histological diagnosis of usual interstitial pneumonia (one had features suggestive of non-specifi c interstitial pneumonia, one had features of hyper sensitivity pneumonitis, one had non-specifi c airway disease, and two had inadequate sampling).

Notably, ARTEMIS-IPF was started before the publication of the 2011 guidelines.1 These guidelines state that if a patient in the appropriate clinical setting has a high-resolution CT pattern consistent with a pattern of possible usual interstitial pneumonia, a diagnosis of idiopathic pulmonary fi brosis can be confi dently established if the histopathology features in the surgical lung biopsy show patterns of usual or probable usual interstitial pneumonia; a diagnosis of idiopathic pulmonary fi brosis is probable if the histopathology features in the surgical lung biopsy shows pattern of possible usual interstitial pneumonia following a multidisciplinary discussion; and a diagnosis of idiopathic pulmonary fi brosis is not made if the histopathology features in the surgical lung biopsy show a pattern of not usual interstitial pneumonia. By grouping patients with histological features of possible or not usual interstitial pneumonia as not usual interstitial pneumonia (for the purposes of the study), we were conservative in the approach used for this analysis and the positive predictive value of possible fi ndings on high-resolution CT would otherwise have been higher than the 94% reported in the results if we had applied the 2011 criteria. Even with expert histological assessment, distinguishing usual interstitial pneumonia from non-specifi c interstitial pneumonia and other interstitial disease is challenging because patients with usual interstitial pneumonia have substantial geographical heterogeneity in the histological fi ndings.3,12,16 The probable explanation for the low prevalence of alternative histological diagnoses to usual interstitial pneumonia in our series is that the patients

Panel 2: Research in context

Systematic reviewWe searched PubMed for reports published before Sept 30, 2013, with the search terms “idiopathic pulmonary fi brosis” and “usual interstitial pneumonia patterns”. Apart from the 2011 joint evidence-based guidelines on diagnosis for idiopathic pulmonary fi brosis,1 we identifi ed two small single-centre retrospective studies of prognostic value of high-resolution CT for diagnosis of usual interstitial pneumonia and idiopathic pulmonary fi brosis.16,17

InterpretationOur multicentre study supports that in patients suspected to have idiopathic pulmonary fi brosis on the basis of clinical features (including absence of collagen vascular diseases and elimination of exposures in domestic or work environment attributable to the manifested pulmonary fi brosis), the presence of possible usual interstitial pneumonia pattern on high-resolution CT might be suffi cient to ascertain the diagnosis of idiopathic pulmonary fi brosis. Our fi ndings might lessen the need for surgical lung biopsy in patients with possible usual interstitial pneumonia on high resolution CT. The 2011 guidelines1 emphasise the need for assessment of patients suspected to have idiopathic pulmonary fi brosis at regional centres with experienced assessors with multidisciplinary discussions and our results are in keeping with this recommendation.

Articles

284 www.thelancet.com/respiratory Vol 2 April 2014

considered for enrolment were selected to meet the inclusion criteria for the study purposes. Alternatively, in the appropriate clinical presentation, patients with concomitant possible usual interstitial pneumonia on high-resolution CT and not usual interstitial pneumonia on surgical lung biopsy might be uncommon.

On review of free-text comments of the expert pathology reviewers, we also noted that for possible usual interstitial pneumonia and not usual interstitial pneumonia categories, reviewers frequently inserted comments that suggested either suspicion of other disorders or concerns about adequacy of sampling. By contrast, we noted few comments about alternative diagnoses for slides classifi ed as usual interstitial pneumonia and probable usual interstitial pneumonia. Therefore, we therefore felt that grouping possible usual interstitial pneumonia and not usual interstitial pneumonia was appropriate for the purpose of this analysis.

Radiological diagnoses of possible usual interstitial pneumonia in our study were made on the basis of analysis of clinical data by an expert clinician familiar with interstitial lung diseases. Although the reliance on a single expert radiologist to assess each individual case could be regarded as a limitation, the positive predictive value of the radiological diagnosis of possible usual interstitial pneumonia was suffi ciently high (94%) that addition of a second reader could not have signifi cantly improved the positive predictive value. Finally, the observations from our study should not be extrapolated to the setting of inconsistent patterns on high-resolution CT.

Absence of defi nitive honeycombing on high-resolution CT should prompt increased scrutiny for detection of atypical clinical presentation (including consideration of underlying environmental and collagen vascular disease) and radiological fi ndings. However, our study suggests that high-resolution CT is appropriate for diagnosis of idiopathic pulmonary fi brosis in patients with a possible pattern of usual interstitial pneumonia in the appropriate clinical setting. If such patients are assessed by experienced experts, histological confi rmation might not be essential to obtain this diagnosis.

Contributors GR was ARTEMIS Steering Committee Chair, overlooked the study

design, interpreted data, reviewed the results with TGO’R (ARTEMIS

Medical Monitor), wrote the fi rst draft with TGO’R, and incorporated

feedback from co-authors with TGO’R. DL, JDG, RW, and RS interpreted

the high-resolution CT images and revised the report. TVC and KOL

interpreted the histopathological features of the surgical lung biopsy and

revised the report. JB, KKB, JJE, KRF, FJM, and AUW were ARTEMIS

Steering Committee Members and revised the report. LS and HZ did the

data analysis. PSP designed the clinical trial. ABM designed the clinical

trial and revised the report.

Declaration of interestsGR, DL, JDG, RW, TVC, KOL, JB, KKB, JJE, KRF, FJM, and AUW are

paid consultants for the sponsor, Gilead Sciences. LS, HZ, PSP, and

TGO’R are employees of the sponsor and own Gilead stock. RS is an

employee of Perceptive Informatics, which is a vendor under contract

with Gilead Sciences. ABM owns Gilead stock.

AcknowledgmentsWe thank the patients and site coordinators for their contributions and

the members of the Steering Committee of the ARTEMIS IPF trial who

are the authors of this paper, and Steven Kawut for reviewing this report.

References1 Raghu G, Collard HR, Egan JJ, et al. ATS/ERS/JRS/ALAT

Committee on Idiopathic Pulmonary Fibrosis. An offi cial ATS/ERS/JRS/ALAT statement: idiopathic pulmonary fi brosis: evidence-based guidelines for diagnosis and management. Am J Respir Crit Care Med 2011; 183: 788–824.

2 Flaherty KR, King TE Jr, Raghu G, et al. Idiopathic interstitial pneumonia: what is the eff ect of a multidisciplinary approach to diagnosis? Am J Respir Crit Care Med 2004; 170: 904–10.

3 Flaherty K, Travis W, Colby T, et al. Histopathologic variability in usual and nonspecifi c interstitial pneumonias. Am J Respir Crit Care Med 2001; 164: 1722–27.

4 Hunninghake GW, Lynch DA, Galvin JR, et al. Radiologic fi ndings are strongly associated with a pathologic diagnosis of usual interstitial pneumonia. Chest 2003; 124: 1215–23.

5 Hunninghake GW, Zimmerman MB, Schwartz DA, et al. Utility of a lung biopsy for the diagnosis of idiopathic pulmonary fi brosis. Am J Respir Crit Care Med 2001; 164: 193–96.

6 Flaherty K, Thwaite E, Kazerooni E, et al. Radiological versus histological diagnosis in UIP and NSIP: survival implications. Thorax 2003; 58: 143–48.

7 American Thoracic Society. Idiopathic pulmonary fi brosis: diagnosis and treatment. International Consensus Statement. American Thoracic Society (ATS) and the European Respiratory Society (ERS). Am J Respir Crit Care Med 2000; 161: 646–64.

8 Raghu G, Behr J, Brown KK, et al. Treatment of idiopathic pulmonary fi brosis with ambrisentan, a selective antagonist of the endothelin A receptor: a randomized trial. Ann Intern Med 2013; 158: 641–49.

9 King TE Jr, Behr J, Brown KK, et al. BUILD-1: a randomized placebo-controlled trial of bosentan in idiopathic pulmonary fi brosis. Am J Respir Crit Care Med 2008; 177: 75–81.

10 Sverzellati N, Wells AU, Tomassetti S, et al. Biopsy-proved idiopathic pulmonary fi brosis: spectrum of nondiagnostic thin-section CT diagnoses. Radiology 2010; 254: 957–64.

11 Elliot TL, Lynch DA, Newell JD Jr, et al. High-resolution computed tomography features of nonspecifi c interstitial pneumonia and usual interstitial pneumonia. J Comput Assist Tomogr 2005; 29: 339–45.

12 American Thoracic Society/European Respiratory Society. International multidisciplinary consensus classifi cation of the idiopathic interstitial pneumonias. Am J Respir Crit Care Med 2002; 165: 277–304.

13 Utz JP, Ryu JH, Douglas WW, et al. High short-term mortality following lung biopsy for usual interstitial pneumonia. Eur Respir J 2001; 17: 175–79.

14 Kreider ME, Hansen-Flaschen J, Ahmad NN, et al. Complications of video-assisted thoracoscopic lung biopsy in patients with interstitial lung disease. Ann Thorac Surg 2007; 83: 1140–44.

15 Raghu G, Weycker D, Edelsberg J, Bradford WZ, Oster G. Incidence and prevalence of idiopathic pulmonary fi brosis. Am J Respir Crit Care Med 2006; 174: 810–16.

16 Fell CD, Martinez FJ, Liu LX, et al. Clinical predictors of a diagnosis of idiopathic pulmonary fi brosis. Am J Respir Crit Care Med 2010; 181: 832–37.

17 Gruden JF, Panse PM, Leslie KO, Tazelaar HD, Colby TV. UIP diagnosed at surgical lung biopsy, 2000–2009: HRCT patterns and proposed classifi cation system. AJR Am J Roentgenol 2013; 200: W458–67.

18 Raghu G, Mageto YN, Lockhart D, Schmidt RA, Wood DE, Godwin JD. The accuracy of the clinical diagnosis of new-onset idiopathic pulmonary fi brosis and other interstitial lung disease: a prospective study. Chest 1999; 116: 1168–74.

19 Glazer CS, Rose CS, Lynch DA. Clinical and radiologic manifestations of hypersensitivity pneumonitis. J Thorac Imaging 2002; 17: 261–72.

20 Morell F, Villar A, Montero M-A, et al. Chronic hypersensitivity pneumonitis in patients diagnosed with idiopathic pulmonary fi brosis: a prospective case-cohort study. Lancet Respir Med 2013; 1: 685–94.

Comment

www.thelancet.com/respiratory Vol 2 April 2014 249

No gold standard test exists for diagnosis of idiopathic pulmonary fi brosis, but the presence on a thoracic high resolution CT scan of honeycomb change, defi ned as dilatations of small airways giving the appearance of clumps of small cysts 3–10 mm in diameter (ie, mature and irreversible fi brosis), has a high predictive value for a pathological diagnosis of usual interstitial pneumonia.1 Together with a typical history of pro gressive breath-lessness in an older person without signi fi cant environ-mental exposures, drug toxicities, or evidence of a connective tissue disease, presence of usual interstitial pneumonia signifi es a diagnosis of idiopathic pulmonary fi brosis. A confi dent diagnosis of idiopathic pulmonary fi brosis is needed because its prognosis is worse than that of other interstitial lung diseases2 and specifi c treatment recommendations apply, including avoidance of interventions that cause harm.3

Many patients with suspected idiopathic pulmonary fi brosis will not have honeycombing on high-resolution CT, but will otherwise fulfi l the radiological criteria for a usual interstitial pneumonia pattern with reticulation (lines) in a peripheral and basal distribution, without atypical features (eg, extensive ground glass change, bronchovascular distribution, nodules, and consolidation). In the 2011 international guidelines3 for diagnosis

and management of idiopathic pulmonary fi brosis, honeycombing is specifi ed as a requirement for usual interstitial pneumonia pattern on high-resolution CT, with lung biopsy sampling required when honeycombing is absent. However, lung biopsy sampling in this population of older people has substantial risks, and in practice many patients with possible usual interstitial pneumonia pattern on high-resolution CT will not undergo this invasive procedure. These patients are therefore neglected because, according to the guidelines, a confi dent diagnosis of idiopathic pulmonary fi brosis cannot be made without a biopsy and treatment recommendations do not apply.4

In The Lancet Respiratory Medicine, Ganesh Raghu and colleagues5 go some way towards addressing this problem. In a retrospective study of 315 patients previously screened for a randomised trial in idiopathic pulmonary fi brosis, 79 (94%) of 84 patients (mean age 64·4 years, range 47–78 years) who had a high resolution CT pattern of possible usual interstitial pneumonia (without honeycombing) had histopathological usual interstitial pneumonia after analysis of lung biopsy samples. Consideration should be given to potential selection bias, because the pathology of 102 patients who did not undergo biopsy sampling was not known, although their demographics and lung function were much the same

continuity of care is needed, with an eye to the prevailing epidemiology, to assess signs and symptoms over time. Access to spirometry is crucial, but inevitably, this assessment will often be marred with clinical uncertainty. Better to acknowledge this uncertainty upfront; for both doctor and patients.

Chris van WeelAustralian Primary Health Care Research Institute, Australian National University, Canberra, Australia; Department of Primary and Community Care, Raboud University Medical Centre, Nijmegen, Netherlands chris.vanweel@radboudumc.nl

I declare that I have no competing interests.

1 Tirimanna PR, van Schayck CP, den Otter JJ et al. Prevalence of asthma and COPD in general practice in 1992: has it changed since 1977? Br J Gen Pract 1996; 46: 277–281.

2 Jones RCM, Price D, Ryan D, et al, on behalf of The Respiratory Eff ectiveness Group. Opportunities to diagnose chronic obstructive pulmonary disease in routine care in the UK: a retrospective study of a clinical cohort. Lancet Respir Med 2014; published online Feb 13. http://dx.doi.org/10.1016/S2213-2600(14)70008-6.

3 Global Initiative for Chronic Obstructive Lung Disease. Global Strategy for the Diagnosis, Management, and Prevention of Chronic Obstructive Pulmonary Disease. http://www.goldcopd.org/uploads/users/fi les/GOLD_Report_2014_Jan23.pdf (accessed Jan 26, 2014).

4 Albers M, Schermer T, Heijdra Y, Molema J, Akkermans R, van Weel C. Predictive value of lung function below the normal range and respiratory symptoms for progression of chronic obstructive pulmonary disease. Thorax 2008; 63: 201–07.

5 Boom G, van den Schayck CP, van Rutten-van Mölken MPMH, et al. Active detection of COPD and asthma in the general population: results and economic consequences of the DIMCA programme. Am J Respir Crit Care Med 1998; 158: 1730–38.

6 Albers M, Schermer T, Molema J, et al. Do family physicians’ records fi t guideline diagnosed COPD? Family Practice 2009; 26: 81–87.

7 Knottnerus JA. Challenges in dia-prognostic research. J Epidemiol Community Health 2002; 56: 340–41.

8 Gillies JCM, Mercer SW, Lyon A, Scott M, Watt GCM. Distilling the essence of general practice: a learning journey in progress. Br J Gen Pract 2009; 59: e167–76.

9 Schers H, Bor H, Hoogen HJM, van Weel C. What went and what came? Morbidity trends in general practice from the Netherlands. Eur J Gen Pract 2008; 14 (suppl 1): 13–24.

10 Starfi eld B, Shi L, Macinko J. Contribution of primary care to health systems and health. Milbank Q 2005; 83: 457–502.

11 Woolf SH. The meaning of translational research and why it matters. JAMA 2008; 299: 211–13.

Understanding CT patterns in idiopathic pulmonary fi brosis

Published OnlineFebruary 18, 2014http://dx.doi.org/10.1016/S2213-2600(14)70030-X

See Articles page 277

Mau

ro F

erm

arie

llo/S

cienc

e Ph

oto

Libr

ary

Comment

250 www.thelancet.com/respiratory Vol 2 April 2014

as people who had the procedure. Furthermore, some participants might have had biopsy fi ndings inconsistent with idiopathic pulmonary fi brosis and not been referred for consideration in the trial. Previously, Fell and colleagues showed, in a retrospective study6 of 135 participants with interstitial lung disease, that in the absence of honeycombing, age and the presence of reticulation on high-resolution CT were able to confi dently predict idiopathic pulmonary fi brosis. Age 70 years or older had a positive predictive value of at least 95%, and the presence of more widespread reticulation on high-resolution CT allowed prediction of idiopathic pulmonary fi brosis in younger patients. Although further prospective studies could provide confi rmation, patients who have suspected idiopathic pulmonary fi brosis without honeycombing on high-resolution CT could in future be spared the risks of lung biopsy sampling and be given a confi dent diagnosis of idiopathic pulmonary fi brosis on the basis of clinical and radiological fi ndings alone.

Raghu and colleagues’ fi ndings5 will probably persuade clinicians to make more inclusive and confi dent diagnoses of idiopathic pulmonary fi brosis than are advocated by the 2011 guidelines. The results will also help to provide clarity about treatment options for a greater number of patients and increase the pool of patients suitable for inclusion in clinical trials of new therapies. This new inclusivity might be regarded as a throwback to a time when the umbrella term cryptogenic fi brosing alveolitis was preferred in the UK. Arguments for abandoning this terminology included concerns that it encompassed fi brotic interstitial lung diseases other than idiopathic pulmonary fi brosis with diff ering outcomes.7 Cryptogenic fi brosing alveolitis also encompassed the entity of fi brotic non-specifi c interstitial pneumonia with a better prognosis compared with typical idiopathic pulmonary fi brosis, although its pathological similarity to usual interstitial pneumonia8 and coexistence of both pathologies in individual patients9 suggest a common pathogenesis. Raghu and colleagues also present tantalising data that patients with suspected idiopathic pulmonary fi brosis and a high-resolution CT scan deemed inconsistent with usual interstitial pneumonia had histopathologically confi rmed usual interstitial pneumonia in 81·7% of biopsied cases. Which criteria were met for scans deemed inconsistent (eg, ground glass change, air trapping, atypical distribution) would be interesting to

know. A strong argument will be made for a change of terminology if such an appearance is not inconsistent with usual interstitial pneumonia at all.

Interpretation of high-resolution CT patterns should continue to be made together with a typical clinical history, and careful clinical assessment is mandatory to identify those patients who might have immunologically driven disease in which infl ammation precedes fi brosis, supported by adjunctive tests such as bronchoalveolar lavage.10 In the modern era with optimal high-resolution CT imaging and thoracic radiologists familiar with interpretation of interstitial lung disease patterns through participation in multidisciplinary meetings,11 the role of lung biopsy assessment might diminish further. In future, combination of high-resolution CT with new non-invasive biomarkers and functional imaging could be used to better defi ne phenotypes of fi brotic interstitial lung disease.

Simon P HartHull York Medical School, Centre for Cardiovascular and Metabolic Research, Academic Respiratory Medicine, Castle Hill Hospital, Cottingham HU16 5JQ, UKs.hart@hull.ac.uk

I declare that I have no confl icts of interest.

1 Sundaram B, Gross BH, Martinez FJ, et al. Accuracy of high-resolution CT in the diagnosis of diff use lung disease: eff ect of predominance and distribution of fi ndings. AJR Am J Roentgenol 2008; 191: 1032–39.

2 Bjoraker JA, Ryu JH, Edwin MK, et al. Prognostic signifi cance of histopathologic subsets in idiopathic pulmonary fi brosis. Am J Respir Crit Care Med 1998; 157: 199–203.

3 Raghu G, Collard HR, Egan JJ, et al. An offi cial ATS/ERS/JRS/ALAT statement: idiopathic pulmonary fi brosis: evidence-based guidelines for diagnosis and management. Am J Respir Crit Care Med 2011; 183: 788–824.

4 Wells AU. The revised ATS/ERS/JRS/ALAT diagnostic criteria for idiopathic pulmonary fi brosis (IPF)—practical implications. Respir Res 2013; 14 (suppl 1): S2.

5 Raghu G, Lynch D, Godwin JD, et al. Diagnosis of idiopathic pulmonary fi brosis with high-resolution CT in patients with little or no radiological evidence of honeycombing: secondary analysis of a randomised, controlled trial. Lancet Respir Med 2014; published online Feb 18. http://dx.doi.org/10.1016/S2213-2600(14)70011-6.

6 Fell CD, Martinez FJ, Liu LX, et al. Clinical predictors of a diagnosis of idiopathic pulmonary fi brosis. Am J Respir Crit Care Med 2010; 181: 832–37.

7 Bradley B, Branley HM, Egan JJ, et al. Interstitial lung disease guideline: the British Thoracic Society in collaboration with the Thoracic Society of Australia and New Zealand and the Irish Thoracic Society. Thorax 2008; 63 (suppl 5): 1–58.

8 Nicholson AG, Colby TV, du Bois RM, Hansell DM, Wells AU. The prognostic signifi cance of the histologic pattern of interstitial pneumonia in patients presenting with the clinical entity of cryptogenic fi brosing alveolitis. Am J Respir Crit Care Med 2000; 162: 2213–17.

9 Flaherty KR, Travis WD, Colby TV, et al. Histopathologic variability in usual and nonspecifi c interstitial pneumonias. Am J Respir Crit Care Med 2001; 164: 1722–27.

10 Morell F, Villar A, Montero M-A, et al. Chronic hypersensitivity pneumonitis in patients diagnosed with idiopathic pulmonary fi brosis: a prospective case-cohort study. Lancet Respir Med 2013; 1: 685–94.

11 Flaherty KR, King TE Jr, Raghu G, et al. Idiopathic interstitial pneumonia: what is the eff ect of a multidisciplinary approach to diagnosis? Am J Respir Crit Care Med 2004; 170: 904–10.