Of the total of 392 patients who underwent surgical resection for lung nodules/masses, 40 patients (10.2%) were excluded due to 1) lack of a satisfactory quality CT scan of the chest or lung tissue and 2) presence of non-lung cancer related metastatic lesions (Figure 1). The final cohort included 352 patients (Figure 1). The prevalence of PBM in the cohort was 9.1% (32/352 patients). The remaining 320 patients (90.9%) constituted the comparison group without PBM (Figure 1).
The study cohort was elderly with a mean age of 66.15±10.19 years, predominantly white (96.3%), had approximately equal numbers of male and female patients (57.1% female), and was overweight with mean body mass index (BMI) of 28.35±6.88 kg/m2 (Table 1). Demographic variables were not different between the groups with and without PBM. Ninety percent of the study cohort was ever-smoker with mean pack years of 43.23±31.86 and median pack years of 40 (IQR 32.5). This reflects the distinct, heavy smoking habit in the Appalachian region. Approximately two-thirds of patients (61.4%) carried a clinical diagnosis of chronic obstructive pulmonary disease (COPD), whereas only a minority of patients carried the clinical diagnosis of ILD (n=5, 1.5%). There were no significant differences between the groups with and without PBM in smoking behavior and prevalence of COPD. All five patients with a history of ILD were in the group without PBM. The PBM group had higher proportion of patients with gastro-esophageal reflux disease (GERD, 19/32 (59.4%) vs. 124/320 (38.7%), p=0.036). The other notable comorbidities for the cohort included hypertension (69.9%), hyperlipidemia (59.3%), coronary artery disease (37.2%), anxiety (32.7%), diabetes (24.4%), mood disorders (40.6%), pain disorders (31.2%) and hypothyroidism (18.1%) with no differences between the groups. Baseline O2 requirement was documented in 14.5% patients, predominantly due to a clinical COPD diagnosis, with no difference between the groups (Table 1).
Pre-surgical PFTs were performed for most patients in the cohort (n=336, 95.4%). Corresponding to the diagnoses of COPD, the study cohort showed PFTs consistent with obstructive ventilatory impairment (mean FEV1/FVC ratio of 67.74±12.17), air trapping (mean percent predicted RV of 141.13±51.41), and mildly decreased percent predicted DLCO (69.06±21.95). None of the indices of pulmonary function showed a statistically significant difference between the groups with and without PBM. Finally, mortality did not differ between the groups (15.6% vs. 10.9, p=0.387) (Table 1).
Radiographic emphysema was extremely frequent in the study cohort (70.4%). The group with PBM showed a significantly higher prevalence of any form of emphysema relative to the group without PBM (100% vs. 67.5%, p<0.0001). Additionally, the group with PBM had a higher prevalence of both CL and PS emphysema (90.6% vs. 59.4%, p<0.0004 and 43.7% vs. 25.6%, p=0.036, respectively). About half of the study cohort (52.9%) noted presence of subclinical ILA (34.3%) and ILD patterns (17.6%) on the CT chest. Additional findings of ILA included subpleural reticular changes (14.8%), CL-GGO (8%), non-emphysematous cysts (8%) and mixed CL-GGO with subpleural reticular changes (7.4%). ILD patterns recognized on the CT scan of the chest included UIP (0.6%), probable UIP (0.6%), NSIP (0.9%), RB-ILD (3.1%), LCH (0.6%), DIP (1.1%), CPFE (3.1%), OP (2.8%) and unclassifiable (4.8%). A combination of ILA and ILD patterns were observed in a greater proportion of the group with PBM compared to the group without PBM (87.5% vs. 49.4%, p=0). The PBM group demonstrated a higher prevalence of CL-GGO ILA pattern (18.7% vs. 6.8%, p=0.031) and any ILD patterns (43.7% vs. 15%, p=0.002). Lastly, isolated traction bronchiectasis on the CT scan was more common in the group with PBM (28.1% vs. 13.1% p=0.032) (Table 2).
Histologically, primary lung malignancy was noted in 92% of resected nodules/masses in the study cohort with no difference between the groups with and without PBM. There were several smoking-associated pathologic findings observed in greater frequency in the group with PBM including emphysema (100% vs. 48.7%%, p<0.0001), any pulmonary fibrosis (56.2% vs. 13.1%, p<0.0001), DIP (12.5% vs. 2.8%, p=0.022), anthracosis (78.1% vs. 32.2%, p<0.0001) and honeycomb changes (15.6% vs. 0.9%, p<0.0001). While organizing pneumonia was solely present in non-PBM group (0 vs. 6.5%, p=0.238), respiratory bronchiolitis was noted in both groups without significant difference (15.6% vs. 12.5%, p=0.581) (Table 2). CT chest images and corresponding pathologic findings of a representative case with PBM are displayed in figure 3 and 4, respectively.
Univariate analysis of significant clinical, radiographic and pathologic predictors for PBM are provided (Figure 2). Comorbid GERD predicted PBM in the study cohort (OR 2.31, 95% CI [1.10 – 4.84]). Subtypes of CL and PS emphysema were associated with PBM (OR 6.61, 95% CI [1.97 – 22.16] and OR 2.25, 95% CI [1.07 – 4.74], respectively). Of the various ILA patterns, CL-GGO increased PBM approximately three-fold (OR 3.13, 95% CI [1.16 – 8.39]). Combined ILA/ILD correlated strongly with PBM (OR 7.17, 95% CI [2.46 – 20.93]). Additional histologic findings associated with PBM included honeycomb changes, any fibrosis, anthracosis and DIP (OR 19.56, 95% CI [4.43 – 86.33]; OR 8.51, 95% CI [3.94 – 18.38]; OR 7.52 95% CI (3.15 – 17.96); and OR 4.93 95% CI [1.43 – 17.05], respectively).
Based on above mentioned variables, a logistic regression model was developed to predict PBM (Table 3). Considering mutually existing prevalence of both radiologic and histologic emphysema in all PBM patients, they were excluded from the model. Radiologic findings of CL-GGO ILA and isolated traction bronchiectasis were highly predictive of PBM (OR 5.72, 95% CI [1.53 – 21.32] and OR 3.45, 95% CI [1.21 – 9.79], respectively). Histologic findings of fibrosis, DIP and anthracosis were significantly associated with PBM (OR 5.83, 95% CI [2.17 – 15.68], OR 13.60, 95% CI [2.74 – 67.52] and OR 6.66, 95% CI [2.50 – 17.70], respectively). A radiographic ILD pattern and a histologic finding of honeycomb changes trended towards significance.