In this prospective cohort study, we investigated whether MVBP could be a preoperative predictor of pleural adhesions. We found a significant relationship between MVBP and pleural adhesions, and that the PPV, NPV, and accuracy per lung segment were 26.9%, 96.0%, and 90.0%, respectively. Prediction by MVBP had a sensitivity of 70.0% per patient, which was higher than that of other predictors, and would be suitable for screening for pleural adhesions. MVBP is a CT finding of increased blood flow beneath the visceral pleura, with the micro-vessels connected to pulmonary vessels for perfusion. Therefore, MBVP can be easily distinguished from pleural thickening and fibrosis. In addition, pleural adhesions with BVLC have MVBP on preoperative CT near the adhesion site because of this pulmonary venous perfusion associated with increased blood flow beneath the visceral pleura. In a previous retrospective study, MVBP were detected near the pleural adhesion site on preoperative CT in 79% of patients with BVLC based on surgical findings [6]. Because MVBP could also enable prediction of the pleural adhesion site, in this study, MVBP were evaluated in each pulmonary segment to carry out the site-specific assessment. The PPV of MVBP for predicting pleural adhesions per area was 26.9% and the sensitivity was 39.1%. Moreover, the PPV of MVBP for predicting pleural adhesions with BVLC specifically, was 23.1% and the sensitivity was 45.6%. This means that a pleural adhesion exists in about a quarter of patients whose preoperative CT shows MVBP. PPV was relatively lower than other prediction factors and the false positive rate was high. This may be due to the fact that MVBP are caused not only by BVLC, but also as a result of the gravitational effect, partial pulmonary venous stagnation, and partially increased blood flow from old inflammatory changes.

There have been several studies on how to predict pleural adhesions [7,8,9,10,11,12,13,14]. Dynamic examinations such as ultrasonography, dynamic CT, and dynamic MRI to detect sliding of the visceral and parietal pleura during the inspiratory and expiratory phases are performed to help predict pleural adhesions. These tests have a relatively higher PPV than MVBP, with ultrasonography between 44 and 73% [7, 13] and dynamic CT at 39% [14]. However, pleural adhesions are not frequent (13–35%) [7, 13,14,15]. Therefore, performing these additional examinations for the sole purpose of predicting adhesions is controversial in daily clinical practice because of the burden on patients and medical staff, as well as the financial costs. Evaluation by MVBP is performed on conventional CT and involves no patient burden and low examiner effort. Other methods to predict pleural adhesions include evaluation of the blunted costophrenic angle on chest X-ray (PPV = 39%) or pleural thickening on CT (PPV = 58%) [15, 16]. Moreover, a medical history of inflammatory diseases such as pneumonia, pleuritis, and pyothorax, or a history of surgery on the ipsilateral side have also been recognized as predictors of adhesions. In this study, univariable analysis identified that MVBP, pleural thickness on CT, and interstitial change on CT were significant factors for predicting pleural adhesions. A history of pneumonia and thoracic surgery (ipsilateral side of the surgery) had relatively high odds ratios (7.72 and 5.04, respectively). Although not significant because of the small number of cases, these factors should be considered as strong predictors of pleural adhesions. Multivariable analysis identified that MVBP, pleural thickness on CT, and interstitial change on CT were independent factors. MVBP could be considered a predictor of pleural adhesions distinct from inflammatory change just below the pleura, as expressed by pleural thickness. Interstitial changes and pleural thickening have higher hazard ratios than MVBP. Therefore, pleural thickening and interstitial changes are more likely to be associated with pleural adhesions than MVBP.

This study had several limitations. First, pleural adhesions without BVLC cannot be predicted by MVBP. Second, in cases of extensive pleural adhesions, blood flow on the pleural surface is dispersed even with BVLC, and identification of MVBP would be difficult on CT. Therefore, it is difficult to predict pleural adhesions over the whole thoracic cavity, which are the most troublesome during thoracic surgery. Third, the detection of MVBP is difficult in patients with strong interstitial or emphysematous changes that make it hard to distinguish between pulmonary vessels and linear opacity. In this study, MVBP were evaluated by two thoracic surgeons. This is because MBVP are expected to be identified in daily clinical practice by the thoracic surgeons who will perform the surgery when reviewing the preoperative CT findings. However, a coincidence rate of the evaluation among observers has not been analyzed; therefore, reproducibility is an issue for the future.

The prediction of pleural adhesions by MVBP has the following benefits. Evaluation of MVBP by CT is simple and takes less than a minute, allowing the thoracic surgeon to check MVBP easily. Furthermore, the evaluation can be done with conventional preoperative CT and does not require additional examination or special abilities. Finally, it involves no extra economic burden on the patient or the medical system, and the burden on the evaluator is negligible. In contrast, the frequency of pleural adhesions is so low (13–35%) that the value of performing additional examinations just to predict pleural adhesions is questionable.

The detection of MVBP does not affect major treatment strategies. However, in the presence of MVBP on CT, pleural adhesions with BVLC could exist around the findings. Blood flow within the adhesion suggests dense adhesions and a careful approach to the thoracic cavity and an alteration of the skin incision site could reduce lung injury and bleeding. Moreover, if the presence of adhesions is known before surgery, it is possible to prepare for surgery associated with the removal of pleural adhesions; for example, by reserving more operating time and preparing the necessary surgical equipment. This study revealed the predictive ability of MVBP, although the PPV of 26.9% was relatively low. A combined assessment of MVBP and other predictors may be more valuable, especially when the pretest probability based on clinical findings is high.