The chest roentograph is abnormal in most patients with IPF (Figure ). Nevertheless, approximately ten percent of patients with histologically proven IPF have a normal roentograph. In these cases, high-resolution computed tomography (HRCT) will reveal evidence of the disease that has been missed by a plain roentograph [1
Figure 1 PA chest radiograph of a 67-year old man with progressive dyspnea revealing bilateral reticular infiltrates with lower lobe predominance. * Reprinted from Fishman's Pulmonary Diseases and Disorders, 4th edition 2007. Meltzer, EB and Noble, PW: Chapter (more ...)
Roentographic images of IPF can demonstrate reticular markings (net-like curvilinear opacities). These markings are distributed in a bilateral, asymmetric pattern with predilection toward the lower lobes. A particular pattern of course reticular lines juxtaposed between areas of focal round translucency is known as honeycombing. Roentographic honeycombing emerges late in the course of disease and portends poor prognosis [64
Use of standard roentographs lacks diagnostic accuracy. A correct diagnosis of IPF (true positive) is made by a roentograph in less than 50% of cases. Furthermore, the radiographic interpretation of interstitial patterns has the characteristic of poor interobserver agreement. Studies examining this particular test characteristic have reported meager 70% concordance amongst radiologists [65
The development of HRCT revolutionized the diagnostic evaluation of interstitial lung disease. HRCT utilizes x-ray technology, along with computerized algorithms, to construct images of virtual thin axial slices through the chest. These high-fidelity images allow a detailed examination of the pulmonary parenchyma. Subsequently, interobserver agreement and overall diagnostic accuracy has been improved with this technology. HRCT permits identification of alternate patterns of diseases. The primary role of HRCT during the diagnostic evaluation of interstitial lung diseases has become the discrimination of "typical" radiographic IPF from that of other ILD.
The "typical" appearance of IPF on HRCT consists of patchy, predominantly peripheral, predominantly subpleural and necessarily bibasilar reticular opacification (Figure ). Ground glass infiltrates can occupy no more than scant, limited areas of the images. Regions of dense reticulation may demonstrate secondary involvement of medium-sized airways that is known as "traction bronchiectasis". The presence of subpleural honeycombing (defined on HRCT as palisades of small, round translucencies), traction bronchiectasis and thickened interlobular septae increases specificity for a diagnosis of IPF. Together, these findings constitute a radiographic pattern that is termed "confident" or "certain" IPF [67
Figure 2 Computed tomography scan illustrates the "classic" features of IPF. Bilateral, peripheral, subpleural reticular infiltrates are evident. The presence of advanced fibrosis is indicated by honeycomb changes (arrowhead) and traction bronchiectasis (arrow). (more ...)
Certain characteristics need be absent from HRCT in order to label the images as consistent with the "confident" IPF pattern. The HRCT cannot feature predominant ground glass opacities, nodular infiltrates or significant lymphadenopathy. A predominance of upper lobe infiltrates is also inconsistent with "confident" IPF. These findings may suggest alternative diagnoses. For example, ground glass implies heart failure, non-specific interstitial pneumonia, cryptogenic organizing pneumonia, desquamative interstitial pneumonia, respiratory bronchiolitis-associated interstitial lung disease or hypersensitivity pneumonitis. A pattern of fine nodules might suggest hypersensitivity pneumonitis, granulomatous infection or lymphangitic spread of malignancy. Upper lobe disease is found in pulmonary Langerhans' cell histiocytosis, hypersensitivity pneumonitis, several pneumoconioses, sarcoidosis, ankylosing spondylitis, rheumatoid nodules and eosinophilic pneumonia syndromes. Significant hilar lymphadenopathy is associated with sarcoidosis, infection and malignancy.
Several studies examined the accuracy of HRCT utilizing histopathology as the "gold standard" [68
]. Studies have demonstrated specificity exceeds 90% for the "confident" HRCT pattern. Thus in the right clinical setting, it is possible to make the diagnosis of IPF by HRCT alone. In such cases HRCT obviates the need for lung biopsy.
However, testing for the "confident" HRCT pattern is not a sensitive tool for case finding of IPF. The full spectrum of the "confident" HRCT pattern can only be found in 4-out-of-5 cases of biopsy-proven IPF. In other biopsy-proven IPF cases, a less specific reticular pattern is seen on HRCT which has been called "possible" IPF (Figure ). The radiographic pattern of "possible" IPF requires surgical lung biopsy to confirm the diagnosis. Sometimes biopsy identifies an alternative diagnosis.
Figure 3 Computed tomography scan of an 81-year old man with biopsy-proven IPF. A peripheral distribution of reticular opacities is demonstrated. Honeycombing and traction bronchiectasis are notably absent. In the absence of specific findings, a surgical lung (more ...)
Findings of bronchoscopy and surgical lung biopsy
The role of bronchoalveolar lavage (BAL) in the diagnosis of IPF remains limited. While the cell count of BAL fluid from patients with IPF has an expected differential distribution (increased numbers of neutrophils and/or eosinophils), the diagnosis of IPF can not made solely on the basis of BAL fluid analysis. Though much effort has been invested in evaluating the clinical utility of BAL, the results of many studies are contradictory [70
]. Still, BAL fluid analysis, and sometimes transbronchial biopsy (TBB), can be helpful in excluding alternative diagnoses. Bronchoscopic exam may demonstrate tumor, infection, Langerhans' cells or occupational dust exposures.
A surgical lung biopsy is close to the "gold standard" for diagnosis and is recommended to confirm all suspected cases of IPF. Biopsy from two sites is recommended based on data indicating a substantial risk of sampling era unless specific effort is made to reflect the gamut of gross disease [72
In cases presenting with a "confident" HRCT pattern, biopsy can be avoided because the results of biopsy can be predicted [68
]. A sizable tissue specimen is required in order to distinguish patterns of IIP, one from the other. Therefore, surgical biopsy is needed and transbronchial biopsy is inadequate. A surgical lung biopsy can be performed by either open thoracotomy or a video-assisted thoracoscopy (VATS) approach. VATS is preferred since this procedure is associated with lower morbidity and shorter hospital stay as compared with open thoracotomy [73
The decision to perform surgical lung biopsy must be carefully considered. Advanced lung disease, poor functional status and older age are relative contraindications to surgery. The absolute risk associated with biopsy is controversial and all of the evidence related to this issue is derived from retrospective data, thus colored by inherent selection bias. While some studies have noted a high short-term mortality, other studies have demonstrated that surgical lung biopsy can be performed safely [74
]. VATS is usually well tolerated and can provide useful information concerning the diagnosis, prognosis and treatment options.
The gross pathology of IPF may be normal, but often a distinctive nodular appearance of the pleural surface is found. This has been likened to cirrhosis of the liver. The histopathological lesion associated with IPF is known as usual interstitial pneumonia (UIP). This lesion is defined by a distinctly variegated pattern. UIP features normal lung architecture alternating with patchy areas of histologically apparent pulmonary parenchymal fibrosis (Figures and ). Fibrosis takes the form of alveolar septal thickening with marked involvement of the subpleural regions. The most severely involved areas of the lung demonstrate complete distortion of normal architecture, with sheets of dense collagen replacing normal lung tissue and occasional cystic structures known as microscopic honeycombs. When examined carefully under the microscope, the region of scarred lung tissue appears to encroach upon areas of preserved, normal lung tissue. This has been termed the "leading edge" of fibrosis and contains specialized structures known as fibroblast foci. Fibroblast foci are pale-staining whirls of loose extracellular matrix molecules, interspersed with numerous cells of the fibroblast type (Figures and ). Inflammation is mostly absent from the UIP pathologic pattern except for occasional lymphoid follicles that are confined to regions of end-stage fibrosis. UIP contains no hyaline membranes, granulomas or organized alveolar exudates. Sometimes emphysema or respiratory bronchiolitis is superimposed upon the UIP pattern when the patient is a former or active smoker. These pathological changes can complicate diagnostic interpretation.
Figure 4 a) Low-magnification photomicrograph of UIP showing the characteristic heterogeneous involvement of the parenchyma. Zones of interstitial fibrosis are seen alternating with areas of normal lung. b) Higher-magnification demonstrates enlarged cystic airspaces (more ...)
Figure 5 Scanning view of UIP demonstrates the characteristic variegated appearance of UIP. Note the honeycomb change (arrowheads) present in the region of dense fibrosis adjacent to the pleural surface. A fibroblast focus (arrow) is seen at the leading edge of (more ...)
It is important to note that the UIP pattern is found in several diseases and is not limited to IPF. UIP can be associated with connective tissue disease, asbestosis, hypersensitivity pneumonitis, the Hermansky-Pudlak syndrome and drug toxicities (e.g.: bleomycin, amiodarone and nitrofurantoin toxicity). Distinguishing IPF from other disorders that contain UIP requires correlation with the clinical history.
It is also important to realize that honeycomb change of itself is a non-specific manifestation of end-stage fibrosis. Microscopic honeycombs do not equate with the UIP pattern nor do they connote a diagnosis of IPF. Only the full spectrum of UIP is diagnostic for IPF (in the correct clinical setting, as noted above).
The actual "gold standard" diagnosis of IPF consists of clinical-radiological-pathological correlation and was defined by consensus conference in the year 2000 and adopted by the American Thoracic and European Respiratory Societies (ATS/ERS) in a statement of guidelines published in the same year [1
]. According to guidelines, the diagnosis of IPF can be considered definitive only in the presence of a surgical (not transbronchial) lung biopsy.
The definite diagnosis
of IPF requires all of the following: [1
• Surgical lung biopsy revealing a histologic pattern consistent with UIP
• Exclusion of other known causes of interstitial lung disease (e.g.: connective tissue disease, environmental exposure, etc.)
• Abnormal pulmonary physiology with evidence of restriction and/or impaired gas exchange (can exist during exercise alone)
• HRCT demonstrating a pattern of "confident" or "possible" IPF.
In the absence of a surgical biopsy, the diagnosis of IPF remains uncertain. Yet, a set of reproducible clinical criteria were developed to define the probable diagnosis
of IPF in cases in which a surgical biopsy is not possible. These clinical criteria were endorsed by the ATS/ERS consensus statement on IPF [1
]. By consensus opinion, IPF can be reasonably diagnosed if all four major criteria and three-out-of-four minor criteria are satisfied. They are as follows:
Major criteria (Must fulfill all four requirements)
• Exclusion of other known causes for interstitial lung disease (such as drug toxicity, environmental exposure and connective tissue disease)
• Abnormal pulmonary function testing that includes evidence of restriction (reduced VC often with an increased FEV1/FVC ratio)and/or impaired gas exchange (increased A-a gradient or decreased diffusion capacity)
• Bibasilar reticular abnormalities with minimal ground glass opacities on HRCT scans (a "confident" HRCT is preferred)
• Transbronchial lung biopsy or bronchoalveolar lavage (BAL) does not support an alternative diagnosis
Minor criteria (Must fulfill at least three; in addition to major criteria)
• Age > 50 yr
• Insidious onset of otherwise unexplained dyspnea on exertion
• Duration of illness ≥ 3 months
• Bibasilar, inspiratory crackles (dry or "Velcro" type in quality)
These criteria have never been subjected to a prospective analysis and, over time, diagnostic algorithms have continued to evolve. As HRCT technology has improved and the utility of this modality has been consistently demonstrated in clinical trials, HRCT has become a more important tool in diagnostic algorithms. Meanwhile, transbronchial biopsy and BAL have fallen from favor, mostly due to low diagnostic yield. A new ATS/ERS-sponsored consensus statement should address these issues and publication of such is expected in 2008.
Controversies regarding the diagnosis of IPF
The clinical-radiological-pathological "gold standard" diagnosis of IPF is flawed due to several issues, such as: 1) lack of standardized tests to exclude known causes of interstitial lung disease; 2) poor interobserver agreement regarding the interpretation of radiographic images; and, 3) poor interobserver agreement as regards the recognition of histological patterns. Interobserver agreement amongst radiologists reading HRCT has been reported to be no better than 80% [76
]. Agreement amongst pathologists has been shown to depend upon experience and training and can be as low as 50% [77
Since the mid-1990s, attention has focused on the division of idiopathic pulmonary fibrosis into histologically-defined subgroups. This practice stems from the description of NSIP in 1994 [79
]. NSIP pathology consists of homogeneously thickened interstitial spaces that contain accumulated fibrosis and inflammation. In NSIP, fibroblastic foci are scarce and focal areas of organizing pneumonia can be found but remain inconspicuous. It has been suggested that patients with NSIP live longer than patients whose biopsy contains UIP [9
]. However, a clinical description of NSIP-associated symptoms, along with vital statistics and risk factors for NSIP, have never been systematically recorded. By default, NSIP has come to represent a disease that exists in parallel to IPF. Whether NSIP truly represents a separate disease from IPF remains to be shown. Patients with NSIP are usually ten years younger than those with UIP. NSIP is also reported to be more sensitive to corticosteroids [2
]. Some authors suggest that NSIP represents an early stage of IPF but this issue is highly controversial [54
Three lines of evidence suggest that NSIP and UIP are different ends of a spectrum resulting from the same disease. The first piece of evidence is found in the examination of patients undergoing multiple surgical lung biopsies. One such study found that 26% of patients with IPF have NSIP pathology in one lobe while simultaneously displaying UIP in a sample from another lobe [72
]. The second piece of evidence is provided by a study that examined survival, histopathology and pulmonary functions trends [5
]. This study compared a cohort with 12-month interval physiologic stability to a cohort with declining 12-month physiology. It was found that physiology predicts mortality more strongly than any other measurable parameter, including histopathological distinction (i.e
.: UIP vs
. NSIP); in fact, pathological pattern conferred no independent prognostic value. The last bit of evidence comes from a cohort of families affected by familial pulmonary fibrosis where both NSIP and UIP were often found within a single family [16