Background CT and Thoracoscopy are useful tools in the evaluation of patients with pleural diseases. Previous studies have shown that certain features on CT may help in differentiating malignant and benign pleura and these may correlate with thoracoscopic findings. We aimed to assess whether a simple scoring system based on CT appearance criteria could be used to predict the likelihood of pleural malignancy.

Methods 104 patients who underwent semi-rigid thoracoscopy between January 2008 and April 2012 were studied. The scans were reviewed by the Chest Radiologist who was blinded to the thoracoscopic findings and pleural biopsy results. Five CT features were given a score of 1–3, where 3 suggested a feature was present, 1 not present and 2 if uncertain. An overall impression of likelihood of malignancy was then scored 1–3, 3 being malignant, 1 benign and 2 indeterminate. Data was analysed using chi-square and logistic regression.

Results N=104, mean age 74 years (range 40–93); female 35(34%), male 69(66%). Pleural biopsy histology was available in 90 of which 57 (63%) were malignant and 33 (37%) benign; 7 did not have biopsy and 7 were either insufficient or non-diagnostic. In those with malignancy, pleural thickening was noted in 43 (sensitivity 88%, specificity 34%) and nodularity in 29 (sensitivity 52%, specificity 82%) on thoracoscopy.

Using a score of 1 and 3, specificity and sensitivity for circumferential/diffuse pleural thickening was 39% and 65%, nodular pleural thickening 94% and 53%, parietal pleural thickening >1cm 94% and 26%, mediastinal involvement 76% and 63%, smooth pleural thickening 33% and 81%, and overall impression score was 84% and 63%. Total score ≥8 for all 5 features and the overall impression (p<0.0001) was significantly associated with malignancy.

A score of 3 for nodular thickening (p<0.0001), parietal thickening (p=0.03), mediastinal involvement (p=0.0011) and a combined score ≥ 4 for these 3 features (sensitivity 79%, specificity 73%, p=0.0001) was highly suggestive of malignancy.

Conclusion These results suggest that a CT scoring system based on the above features can be reliably used to predict pleural malignancy prompting more intensive evaluation such as thoracoscopy, although this will require further validation.