Prediction of histopathologic usual interstitial pneumonia among ILD cases with a non-diagnostic high-resolution computed tomography scan

Rationale: It remains unclear whether a possible usual interstitial pneumonia (UIP) pattern on high-resolution computed tomography (HRCT) scan provides sufficient diagnostic specificity in selected cases to forgo surgical lung biopsy in the diagnostic work-up of idiopathic pulmonary fibrosis (IPF). Methods: Patients with biopsy-proven interstitial lung disease (ILD) and non-diagnostic HRCT scans (i.e. patients with “possible UIP” and “inconsistent with UIP” patterns) were identified (n = 406). Test characteristics were determined for possible UIP pattern as a predictor of UIP pattern on surgical lung biopsy, with and without additional consideration of age, sex, and presence of traction bronchiectasis. Results: Of 406 patients included, 115 (28%) had histopathologic UIP pattern. The most common clinical diagnoses were hypersensitivity pneumonitis (101/406, 24.9%) and IPF (96/406, 23.6%). On HRCT, 65/406 (16%) had possible UIP on HRCT; the remainder was inconsistent with UIP. Possible UIP pattern had a positive predictive value (PPV) of 61.5% (95% CI 48.6-73.3) for UIP pattern on surgical lung biopsy. The addition of age, sex, and severity of traction bronchiectasis to an HRCT possible UIP pattern improved the PPV for histopathologic UIP, as did increasing the population prevalence of histopathological UIP pattern (Table). Conclusions: Possible UIP pattern on HRCT demonstrated poor positive predictive value for histopathologic UIP in our patient population. This is in contrast to recent studies with much higher population prevalence of IPF.1-3 Whether possible UIP pattern on HRCT is sufficient to forgo surgical lung biopsy depends on local prevalence of histopathologic UIP and other clinical and radiographic features.