The medical records of patients hospitalized and diagnosed with OP at Kyungpook National University Hospital (KNUH), a tertiary referral center, in Daegu, South Korea between March 2002 and December 2012 were retrospectively reviewed. OP was diagnosed by multidisciplinary discussion of clinician, radiologists, and pathologists. The pathologic pattern of OP was defined as patchy airspace organization with fibroblastic tissue within alveoli and alveolar ducts with or without bronchiolar involvement and variable interstitial changes, such as type 2 alveolar epithelial cell proliferation and interstitial chronic inflammatory infiltrates2. Pathologic specimens were obtained by surgical lung biopsy, transbronchial lung biopsy, or percutaneous needle biopsy. Patients diagnosed based on a non-surgical biopsy specimen required typical computed tomographic (CT) findings of OP, which included patchy consolidation or ground glass opacity (GGO) with a peribronchial or subpleural distribution without extensive reticulation or honeycombing4. The exclusion criteria applied were as follows: (1) focal OP2 (only ILD form was included) and (2) cases which responded to antibiotic therapy. As described in a previous study5, relapse was defined as the appearance of characteristic new infiltrates by chest radiography with compatible clinical symptoms after corticosteroid administration was stopped or the dose was reduced. This study was approved by the Institutional Review Board of the KNUH, which waived the requirement for written informed consent because of the retrospective nature of the study.

Demographic data including age, gender, and body mass index, and smoking history at presentation were checked, and comorbid conditions and duration of corticosteroid therapy were reviewed. Blood laboratory findings, including leukocyte count, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), protein, albumin, liver function test, procalcitonin, N-terminal-pro-B-type natriuretic peptide (NT-proBNP), lactate dehydrogenase (LDH), were included in the check list. Bronchoalveolar lavage (BAL) fluid analysis included total and differential cell counts and lymphocyte subset data. Arterial blood gas analysis data, such as partial pressure of oxygen (PaO₂) and carbon dioxide (PaCO₂) in arterial blood, and PaO₂/fraction of inspired oxygen (FiO₂) and lung function data, such as percent predicted forced vital capacity (FVC) and diffusing capacity for carbon monoxide (DLCO), were reviewed.

Chest CT findings were reviewed by two chest radiologists (K.-M.S. and J.-K.L.) unaware of clinical information. When the readings of the two radiologists were discrepant, a final decision was reached by consensus. The presences of consolidation, GGO, and reticulation were checked, and the extents of these lesions were estimated at three levels, resulting in a total six lung zones8. Average extents of GGO, consolidation, and reticulation were scored to the nearest 5% in each zone, as previously described9. The upper zone was defined as at or above the aortic arch, the middle zone was defined as between the aortic arch and pulmonary veins, and the lower zone was defined as at or below the pulmonary veins9.

Statistical analysis was performed using SPSS version 20.0 (IBM Co., Armonk, NY, USA). Data are expressed as means with standard deviations or medians with interquartile ranges (IQR) for continuous variables and numbers with percentages for categorical variables. The Student's t test or Mann-Whitney U test was used to compare continuous variables between relapse and non-relapse groups, and the chi-squared test or Fisher exact test was used to compare categorical variables. Multivariable logistic regression analysis was used to identify predictors of relapse in OP patients. The Hosmer-Lemeshow test was used as a test to assess the goodness-of-fit of logistic regression models. p-values of <0.05 were considered to indicate statistical significance.

Seventy-three patients with OP were identified. Mean patient age was 57±11 years with female preponderance (n=45, 62%). Never-smokers (n=50, 68%) were more common than ever-smokers. Pathologic OP patterns were identified by surgical lung biopsy in 42 patients (58%), transbronchial lung biopsy in 22 (30%), or percutaneous needle biopsy in 9 (12%). Secondary OP (n=24, 33%) was associated with collagen-vascular disease (n=8; rheumatoid arthritis, n=4; systemic lupus erythematosus, n=2; and polymyositis, n=2), radiation therapy (n=7), drugs (n=6), hematopoietic stem cell transplantation (n=1), ulcerative colitis (n=1), and mycoplasma pneumonia (n=1).

With the exception of one patient with polythiouracil-induced OP, all patients received corticosteroid treatment with or without immunosuppressant including azathioprine (n=1) and cyclosporine (n=1). Response to therapy could not be assessed in three patients due to follow-up loss (n=2) or no available medical record (n=1). Of the 70 remaining patients, 66 (94%) improved with or without treatment, and of these, 20 (30%) experienced relapse. The other four patients died from disease progression (n=3) or pneumonia (n=1). Median follow-up duration was 19 months (IQR, 7-48 months), median duration of corticosteroid administration before relapse was 6 months (IQR, 4-12 months), and median time from corticosteroid discontinuation to relapse was 2 months (IQR, 0.2-6 months).

Baseline and clinical characteristics are presented in Table 1. Age, gender, body mass index, smoking history, comorbid conditions, and the prevalence of secondary OP were not significantly different in the relapse and non-relapse groups. However, the duration of corticosteroid treatment before relapse was significantly longer in the relapse group, as compared with the total duration of corticosteroid therapy in the non-relapse group (186 days [129-365 days] vs. 137 days [81-195 days], p=0.036). When patients were classified by the duration of corticosteroid administration (within 3 months, 4-6 months, 7-9 months, and more than 9 months), the relapse rate during each time interval was 19%, 29%, 25%, and 58%.

ESR, CRP, procalcitonin, NT-proBNP, and LDH levels were not significantly different between the two groups (Table 2). However, mean serum protein level was significantly lower and mean serum albumin level tended to be lower in the relapse group (6.8±0.67 g/dL vs. 4±0.8 g/dL, p=0.003; and 3.5 g/dL [3.2-3.8 g/dL] vs. 3.7 g/dL [3.5-4.0 g/dL], p=0.070). No significant intergroup difference was found in other parameters of liver function test. PaO₂/FiO₂ tended to be higher in the relapse group (396±85 mm Hg vs. 356±71 mm Hg, p=0.077), but no significant differences were observed for PaO₂ and PaCO₂.

Of BAL fluid analysis data, total and differential cell counts and lymphocyte subsets showed no significant differences. However, the relapse group had significantly lower percentage of FVC predicted value and tended to have a higher percentage of predicted DLCO value than the non-relapse group (70±15% pred. vs. 86±21% pred., p=0.009; and 62±25% pred. vs. 77±27% pred., p=0.057, respectively).

Between the two groups, no significant differences in CT findings were found, and extents of GGO or consolidation and reticulation and total disease extent were not significantly different (Table 3).

Multivariate logistic regression analysis was used to identify predictors of relapse (Table 4). Age, female gender, and variables with p-values of <0.1 by univariate analysis were subjected to multivariate analysis. However, serum albumin was excluded because of its significant correlation with serum protein. Percent predicted FVC, PaO₂/FiO₂, and serum protein level were found to be independent predictors of relapse in OP patients (odds ratio [OR], 0.82; 95% confidence interval [CI], 0.70-0.97, p=0.018; OR, 1.02; 95% CI, 1.00-1.04; p=0.042; and OR, 0.06; 95% CI, 0.01-0.87; p=0.039, respectively). The Hosmer-Lemeshow test indicated that the overall model fit was good (p=0.883).