OBJECTIVE. The objective of our study was to evaluate serial CT changes from normal or near-normal lungs to honeycomb lungs in dust-exposed patients who developed chronic interstitial pneumonia.
MATERIALS AND METHODS. From the records of the national hospital for pneumoconiosis, we retrospectively identified patients with chronic interstitial pneumonia who were under surveillance between 1986 and 2006. All patients occasionally underwent chest CT for evaluation of silicosis or exclusion of possible complications. Patients were included in this study only if the initial CT examination did not show obvious chronic interstitial pneumonia. Fourteen patients (all men; median age at initial CT, 58 years) were identified as meeting the inclusion criterion. Two independent reviewers randomly reviewed the CT scans of the study patients to score the extent of ground-glass opacity, reticulation, and honeycombing; to provide a summation of all interstitial opacities (fibrosis score); and to assess coarseness.
RESULTS. Autopsy findings were available for eight of the 14 patients and confirmed the usual interstitial pneumonia (UIP) pattern seen on CT. The median follow-up period was 15.4 years, and none of the patients experienced acute exacerbation. One hundred two CT scans were reviewed. The earliest CT abnormalities included faint ground-glass opacity limited to the lung bases (n = 13) or only coarse reticular opacity (n = 1). In 13 patients, fibrosis and coarseness progressed linearly, whereas the other opacities did not. The annual increase of the fibrosis score and coarseness ranged from 0.306% to 4.633% and 0.179 to 0.479, respectively. Honeycombing developed in all patients over a median period of 12.1 years (range, 3.7–19.1 years).
CONCLUSION. The coarseness best represented the progression of chronic interstitial pneumonia in dust-exposed patients. The earliest CT finding of a UIP pattern in dust-exposed patients was indistinguishable from other types of chronic interstitial pneumonia.
The natural history of chronic interstitial pneumonia with fibrosis, including idiopathic pulmonary fibrosis (IPF) and nonspecific interstitial pneumonia (NSIP), is not well understood. Clinically, patients present with dyspnea on exertion and thin-section CT shows a mixture of areas with diffuse reticular opacity, ground-glass opacity, and honeycombing of various proportions . In many patients, honeycombing is already present on CT at the time they first complain of symptoms and is considered the main CT manifestation of IPF. However, honeycombing is not considered a frequent CT feature of NSIP, and its absence is one of the factors used to differentiate between the two diseases using CT [2–5]. The 5-year survival of patients with IPF after diagnosis has been reported to be approximately 20–40%, and the median duration of symptoms before diagnosis is 1–2 years . Many previously reported cases of IPF appear to involve patients with late-stage disease; thus, their clinical outcome was poor.
Because CT is not performed in asymptomatic subjects, little is known about the CT progression in patients with chronic interstitial pneumonia during the preclinical period. In our country, because patients with specific occupational dust exposure are under surveillance with periodic follow-up including CT, we are able to evaluate coexisting chronic interstitial pneumonia in these patients. The purpose of the present study was to assess the time and progression between normal or near-normal lungs and end-stage-fibrosis lungs with honeycombing in patients with chronic interstitial pneumonia, mostly the usual interstitial pattern (UIP), who were under surveillance for silicosis using CT.
Materials and Methods
This study was performed at one of the national hospitals for occupational lung disease in our country. All patients had a current or previous occupational history of siliceous dust exposure and were under surveillance for silicosis. Patients were referred to CT for the following reasons: to confirm the presence of pneumoconiosis, evaluate serial changes of pneumoconiosis nodules, and exclude possible complications of lung cancer or tuberculosis. The institutional review board approved this study and waived the requirement for informed consent because of the retrospective nature of the study.
We retrospectively identified patients with chronic interstitial pneumonia using CT reports for 364 CT studies performed between 1999 and 2006; serial CT images had been available since 1986. After identifying candidate patients with chronic interstitial pneumonia, two chest radiologists with 16 and 18 years' experience in chest CT inter pretation, respectively, reviewed the CT images of each candidate to determine whether the patient should be included in the study. We included patients with chronic interstitial pneumonia based on the following CT findings: the presence of diffuse reticular opacities with or without ground-glass attenuation, traction bronchiectasis, honey combing, or subpleural homogeneous attenuation [7, 8]; or the presence of diffuse nonsegmental ground-glass opacities with or without reticular opacities that did not improve during a follow-up period of more than 1 year . They reviewed the serial CT images of each patient, and only patients whose initial CT examination showed no findings or only subtle findings compatible with chronic interstitial pneumonia, either ground-glass or reticular opacities, were finally included in the study. Decisions about which patients to include in the study were made by consensus of the two reviewers.
Within 1 month of the initial CT examination, pulmonary function tests were performed using computerized pulmonary function instruments (Chestac 11 or 25, Chest Company). Forced vital capacity (FVC), forced expiratory volume in 1 second (FEV1), forced expiratory flow rate at 50% of FVC (FEF50), forced expiratory flow rate at 25% of FVC (FEF25), and forced expiratory flow rate at 75% of FVC (FEF75) were determined by spirometry. Each result was expressed as a percentage of the predicted value based on the patient's sex, age, and height.
Occupational and smoking histories were reviewed from the charts by one of the authors.
CT scans were obtained using a TCT-900 scanner (Toshiba Medical Systems) between 1986 and 1996 and a Somatom Plus 4 scanner (Siemens Medical Solutions) after 1996. The scanning parameters with the TCT-900 scanner were as follows: 5- or 10-mm collimation combined with a nonhelical technique, a 10-mm table feed, tube current of 300 mA, 120 kV, and 1 second per rotation. The scanning parameters with the Somatom Plus 4 scanner were as follows: 10-mm collimation, 1:1 helical pitch, 280 mA, 120 kV, and 0.75 second per rotation. Additional thinsection images were obtained for CT examinations performed after 1998. The parameters for the thin sections were 2-mm collimation, nonhelical technique, 15-mm interval, 240 mA, 120 kV, and 0.75 second per rotation. The lung images were reconstructed using a high-spatial-frequency algorithm and were photographed with a window width of 1,500 HU and at a level of –650 or –700 HU. The mediastinal images were photographed with a window width of 350 HU and at a level of 50 HU.
CT Image Analyses
The two chest radiologists reviewed the CT images separately and independently. To assess the serial changes in the CT findings, all the available CT images of the patients were reviewed in random order with no knowledge of the date of CT. The extent and distribution of the CT abnormalities at the upper, middle, and lower lung zones were scored independently by the two reviewers according to the following methods .
First, in each zone, the extent of pulmonary fibrosis (fibrosis score) defined as the combined areas of reticular opacity, ground-glass attenuation, consolidation, and honeycombing was visually estimated to the nearest 5%. The mean value of the three zones was considered the extent of fibrosis (fibrosis score).
Second, in each zone, each area with reticular opacities, ground-glass attenuation, consolidation, and honeycombing was visually estimated separately to the nearest 5% on CT, and the mean value of each opacity in the three zones was recorded.
Third, in each zone, the coarseness of fibrosis was classified as follows: grade 0, ground-glass opacification with no reticular element or cysts; grade 1, predominantly reticular opacity without cysts; grade 2, predominantly microcystic reticular pattern (i.e., definable air spaces < 1 cm in diameter); or grade 3, predominantly macrocystic reticular pattern (i.e., air spaces > 1 cm in diameter). The sum of the scores at each zone was recorded. In patients with no disease in one or two lung zones, the coarseness score was adjusted proportionately to a three-zone score.
The scores of the two reviewers were averaged and the averages were used as the final scores.
We also determined the predominant distribution of fibrosis as subpleural, peribronchial, or random after reviewing the CT sections of each patient. Finally, after completing the scoring sessions, the observers were asked to judge whether the findings on the last CT examination were typical of the published criteria for the UIP pattern seen in IPF [3, 7, 8, 11, 12].
Clinical Course and Pathologic Data
The episodes of acute exacerbation were recorded, and the cause of death was investigated [13, 14]. Autopsy lung specimens were available for eight of the patients who had died. The lungs were inflated with formalin, and a pathologist with 27 years' experience in pathology prepared the tissue specimens. Hemotoxylin-eosin and elastic-Goldner stains were usually used. The pattern of fibrosis was assessed as UIP pattern, NSIP pattern, or otherwise.
Observer variations were evaluated using the kappa coefficient for semicategorical variables or the single-determined SD for quantitative data . The declines in spirometry values were evaluated using a paired Student's t test. A linear regression model was used to evaluate the temporal increases in the CT scores compared with the values for the initial CT. Statistical analyses were performed using SPSS software (version 11.01J, SPSS).
Thirty-eight patients reported to have chronic interstitial pneumonia were submitted to the review session. Among them, 14 patients showed normal or near-normal findings on the initial CT scans. The follow-up period ranged from 3.7 to 20.1 years (median, 15.4 years). The median age at the initial CT examination was 58 years (range, 49–71 years). Twelve patients had silicosis, whereas the two others had no such nodules.
Thirteen patients were smokers and one had never smoked. Smoking pack-years ranged from 0 to 96 (median, 31 pack-years). Eight patients previously worked as metal ore miners, two as tunnel workers, two as stonemasons, and one as a tunnel worker and a metal ore miner; the remaining patient worked in a factory that dealt with siliceous materials. No patient had significant asbestos or hard metal exposure, and none was taking steroids during the follow-up period.
Pulmonary function tests performed at the time of initial CT examination showed normal values for FVC and borderline normal values for FEV1 (Table 1). The follow-up study at the time of the last CT examination revealed a significant decline in the FVC from a median of 96.8% (range, 73.2–127.8%) to 83.1% (range, 61.9–112.3%) (p = 0.002), whereas other parameters remained relatively constant (all, p > 0.05).
TABLE 1: Patient Age and Spirometry Results for 14 Patients with Chronic Interstitial Pneumonia at the Initial and Final CT Examinations
Median Value (Range)
At Initial CT
At Final CT
Mean of Difference (SD)
Median of Difference (Range)
Patient age (y)
-12.4 (-39.4 to 7.6)
-0.3 (-33.5 to 10.5)
Maximal mid-expiratory flow
1.7 (-33.2 to 28.7)
Forced expiratory flow
At 25% of FVC
5.1 (-18.2 to 12.8)
At 75% of FVC
Note—All spirometry data are reported as a percentage of the predicted value based on patient's sex, age, and height. FVC = forced vital capacity, FEV1 = forced expiratory volume in 1 second.
A total of 102 CT scans were reviewed in all patients. CT was performed three to 11 times (median, 8) in each patient. Sixty-six CT examinations were performed early in the study period, so only thick-section images were available for review; in the remaining 36 CT examinations, which were performed later in the study period, thick- and thin-section images were available. In all patients, at least one thin-section scan was obtained.
Interobserver agreement for the presence of each CT abnormality ranged from 0.79 to 0.98, and interobserver variations, expressed as the single-determination SD, were all less than 2 (Table 2).
TABLE 2: Interobserver Agreement for the Presence of Abnormalities on CT in 14 Patients with Chronic Interstitial Pneumonia
Note—Kappa value was calculated for the presence or absence of each finding. Single-determination SDs were calculated for the difference reported by two reviewers. NA = not available.
When the CT scores were evaluated in each patient, significant and consistent linear correlations were identified only for the fibrosis score and coarseness (Figs. 1 and 2). There was a significant linear relationship between time and the fibrosis score in all patients except one. The unstandardized regression coefficients and correlation coefficients (r) for the fibrosis scores ranged from 0.306 to 4.633 and 0.568 to 0.964, respectively, indicating that the annual increase in the fibrosis score ranged from 0.306% to 4.633% above baseline CT. The coarseness scores in each patient showed a significant linear relationship with time in 12 patients. The unstandardized coefficients and correlation coefficients (r) ranged from 0.179 to 0.479 and 0.815 to 0.981, respectively.
In 11 patients, initial chest CT scans showed faint ground-glass opacity with (n = 4) or without (n = 7) reticular opacities in the periphery of both lung bases (Figs. 3A, 3B, 3C and 4A, 4B). In only one patient, the initial abnormality was a reticular opacity. In the remaining two patients, the initial CT showed no abnormal opacities compatible with interstitial pneumonia; however, subpleural faint ground-glass opacity developed within 2 and 3 years, respectively. The median initial fibrosis score for all 14 patients was 3.3 (range, 0–8.4; SD, 2.2).
Honeycombing developed in all 14 patients over a median of 12.1 years (range, 3.7–19.1 years; SD, 4.1). The distribution of fibrosis on the last CT examination was predominantly in the lower lobes (n = 14) and either peripheral (n = 12) or random (no predominance) (n = 2). The interobserver agreement was almost perfect (κ = 1 and 0.826, respectively). The pattern at the last CT examination was typical of the UIP pattern in nine patients, but the five other patients showed an atypical UIP pattern. Interobserver agreement was almost perfect (κ = 1).
Clinical Course and Pathologic Diagnosis
Ten of the 14 patients had already died at the start of this study. The cause of death was as follows: malignancy (n = 3), pneumonia (n = 2), cardiac failure (n = 2), or respiratory failure due to pneumoconiosis and coexisting interstitial pneumonia (n = 3). No patient experienced acute exacerbation during the follow-up period.
Autopsy observations were available for eight patients. The interval between the last CT examination and autopsy ranged from 1 to 30 months (median, 8 months). At pathology, the lungs of all patients showed honeycombing and were diagnosed as having a typical UIP pattern. At CT, the lungs of five patients were interpreted as showing a typical UIP pattern, whereas the lungs of the three other patients were interpreted as showing an atypical UIP pattern.
In our series of patients with chronic interstitial pneumonia, we found that honeycombing required a median of 12 years to develop after no or minimal interstitial abnormalities were observed on CT. The CT extent of fibrosis and coarseness linearly increased compared with those at baseline CT in most patients. For example, according to this model, the annual increase of the fibrosis score was expected to go from 0.306% to 4.633%, with a mean increase of 1%. It should be noted that the speed of progression differed more than 10 times among the patients, indicating a wide range of differences among the cases. From the recent observations of several series, the 5-year survival from the time of the initial IPF diagnosis of patients is only 20–40% [16–22], and disease progression after clinical presentation or diagnosis is expected to be rather fast.
On the basis of the findings from our series, we believe that patients who present with symptoms and die after a few years with the disease are just a subgroup of patients with chronic interstitial pneumonia. Most of these cases are late in the course of the disease or might show relatively fast progression. However, as shown in our cases, there are patients who do not show symptoms because of the indolent disease course and limited extent of disease.
In a recent report, disease progression was described as stepwise rather than linear in some IPF patients . Our series also revealed a small stepwise progression, but overall the progression was close to linear as confirmed statistically in our series. Patients with IPF may experience the so-called “acute exacerbation,” in which severe respiratory failure accompanied by new radiographic opacities occurs within 1 month of diagnosis and no identifiable cause is found [13, 14]. The reported incidence of acute exacerbation depends on the series studied—for example, 9.6% in 2 years , 57% in 3 years , or 22% for a mean of 49.5 months . None of our patients experienced this type of exacerbation during the study period, and this may partly explain the linear progression of disease extent in our series.
Besides the extent of fibrosis, the coarseness score showed a linear progression with time in 12 patients. The coarseness score considers progression from solely ground-glass opacity to honeycombing as well as progression from the lower lobe to the upper lobe. Akira et al.  indicated that the honeycomb area was always preceded by ground-glass opacity that, in turn, disappears with the development of honeycombing. The variation of coarseness scores among the patients in our study was much smaller than that of the fibrosis score (0.179–0.479 vs 0.306–4.633, respectively); thus, the coarseness score may be a better indicator of disease progression in patients with chronic interstitial pneumonia.
In our series, the diagnoses at the last CT scan were compatible with the UIP pattern in nine patients, whereas the lungs of the five other patients were regarded as showing an atypical UIP pattern. The pathologic diagnoses of eight patients—including the three patients with a CT diagnosis of an atypical UIP pattern—all showed a UIP pattern. In a series of 28 patients with chronic interstitial pneumonia with silicosis or mixed-dust pneumoconiosis, the CT pattern was reported to be atypical of the UIP pattern in one quarter of the patients despite the fact that many showed a pathologic UIP pattern . Considering that the initial CT examinations showed only ground-glass opacity in the periphery of both lower lobes in all cases except one in the present study, the evolution of the CT findings might be unexpected because the initial pattern was suggestive of NSIP [1, 3, 7, 9, 28]. Whether there are cases of chronic interstitial pneumonia that evolve from the NSIP pattern to the UIP pattern is not known ; however, the CT findings in our series suggest that this may be possible. In addition, this might just indicate that the earliest CT appearance of the UIP pattern is ground-glass opacity with or without reticulation and that coarseness gradually increases as the disease progresses. From the results of our study, we argue that the CT appearance of chronic interstitial pneumonia depends on the stage of the disease, and this may partially explain why some IPF cases with a UIP pattern were interpreted as NSIP or as IPF with an atypical UIP pattern on CT in many of the previous reports [4, 7, 30–32] as well as in the present series.
There are several limitations in our study. First, the CT images were not always obtained with thin-section CT. The initial CT examinations were performed in the 1980s when the first CT scanner was introduced in our country and at that time high-resolution CT was not widely performed. Although our initial CT scanner was specially tuned through cooperation with the manufacturer, detailed parenchymal changes might have been underestimated . This technical drawback is an inherent problem of radiographic imaging in research for long-term follow-up because machines improve dramatically over relatively short time intervals. Fortunately, however, we obtained additional thin-section images when the parenchymal abnormalities in our patients became complex. Furthermore, the initial abnormality was mostly ground-glass opacity and collimation was considered to not substantially affect the interpretation of the earlier images.
Second, 12 of our patients showed coexisting silicotic nodules in their upper lobes. Although epidemiologic evidence has revealed the etiologic role of occupational dust exposure in patients with IPF [34–37], whether disease progression of chronic interstitial pneumonia associated with silicosis is the same as that of IPF without dust exposure remains controversial and further research is required.
Third, the number of patients in our study was small, and pathologic confirmation was obtained in just eight of the 14 patients. Although we may not understand the whole history of the preclinical stage of chronic interstitial pneumonia from our small sample, we believe that the results of this study still add important information about disease progression.
In conclusion, we identified 14 dust-exposed patients with chronic interstitial pneumonia whose initial CT examination showed no or limited subpleural ground-glass opacity, often with reticulation, that was indistinguishable from the NSIP pattern on CT. During a median observation period of 15.4 years, the extent of fibrosis and the coarseness of the texture pattern on CT increased linearly with time in most patients and honeycombing resulted in all patients. It took a median of approximately 12 years (range, 3.7–19.1 years) for honeycombing to develop in these patients. In eight patients (57%), histologic diagnosis of a UIP pattern was obtained.
The results of this study confirm that the CT appearance of chronic interstitial pneumonia changes substantially with time and suggests that CT may not be able to differentiate the UIP pattern from other types of chronic interstitial pneumonia in the early stage of the disease.
The Rosai Hospital for Silicosis was closed at the end of March 2006 by the decision of the Japanese Government. We thank Reiko Hirayama and Hiromi Nakayama for their assistance.
Kondoh Y, Taniguchi H, Kawabata Y, Yokoi T, Suzuki K, Takagi K. Acute exacerbation in idiopathic pulmonary fibrosis: analysis of clinical and pathologic findings in three cases. Chest 1993; 103:1808 –1812
Collard HR, King TE Jr, Bartelson BB, Vourlekis JS, Schwarz MI, Brown KK. Changes in clinical and physiologic variables predict survival in idiopathic pulmonary fibrosis. Am J Respir Crit Care Med 2003; 168:538 –542
Nicholson A, Colby T, DuBois R, Hansell D, Wells A. The prognostic significance of the histologic pattern of interstitial pneumonia in patients presenting with the clinical entity of cryptogenic fibrosing alveolitis. Am J Respir Crit Care Med 2000; 162:2213 –2217
Arakawa H, Johkoh T, Honma K, et al. Chronic interstitial pneumonia in silicosis and mix-dust pneumoconiosis: its prevalence and comparison of CT findings with idiopathic pulmonary fibrosis. Chest 2007; 131:1870 –1876
Daniil ZD, Gilchrist FC, Nicholson AG, et al. A histologic pattern of nonspecific interstitial pneumonia is associated with a better prognosis than usual interstitial pneumonia in patients with cryptogenic fibrosing alveolitis. Am J Respir Crit Care Med 1999; 160:899 –905