Pleural fluid cytology was positive on 82.5% of the non-bloody effusions and on 82.4% of the bloody ones. The number of red blood cells (220.5 × 103/μL vs. 12.3 × 103/μL) and LDH values (1914 IU/dl vs. 863 IU/dl) were statistically higher in bloody pleural effusions. The presence or absence of blood in pleural effusions cannot predict their etiology in patients with cancer and recurrent symptomatic pleural effusions.
Our study was subject to the limitations of any retrospective study. However, the potential effect of any information bias is minimal, as we used pathologic and laboratory data that were not influenced by the clinician or the personnel conducting the research. Of more concern is the external validity of our data. Our results are applicable only to patients with known primary extrapleural malignancies presenting with a recurrent symptomatic pleural effusion, not to the general patient population with a lower pretest probability of having a MPE. In addition, our patient sample did not include any findings of "purulent" fluids, so our conclusions do not encompass this possibility.
It would be extremely useful if an easily assessed parameter like pleural fluid appearance could be prospectively used to identify patients with positive cytology or to estimate the inflammatory or tumor burden on the pleural space in patients with a previous diagnosis of cancer. It would potentially reduce the number of interventions performed in patients with MPE prior to definitive therapy. Timely identification of the inflammatory pleural response would also be of great interest, as some authors have hypothesized that the results of chemical pleurodesis could be predicted using cellular and chemical characteristics of the pleural fluid[9, 10].
Several published studies using patients presenting with pleural effusions without a prior history of cancer have found a correlation between bloody pleural effusion and malignancy. In a series of 163 patients with large or massive pleural effusions, Porcel and Vives  reported significantly higher RBC counts in patients with MPEs compared with patients with nonmalignant effusions (median value 18 × 109 cells/L versus 2.7 and 109 cells/L, respectively; p < 0.001).
In another prospective study of 334 consecutive patients with chronic pleural effusion reported by Martensson and colleagues, 86% of the 44 bloody fluids and 57% of the 65 blood-tinged fluids were cancerous on cytology or biopsy (p < 0.01) . Villena and coworkers found that the presence of a bloody fluid slightly increased the probability of a malignant pleural effusion (Odds Ratio = 1.73). In her series of 715 patients, 47% of the bloody effusions were MPE. The authors concluded that fluid appearance should not be overemphasized as a diagnostic tool . However, our study does not support an extrapolation of these reports to patients with a known history of cancer. In this study, we found no association between pleural fluid appearance and chemical laboratory analysis, cell counts (except for LDH and RBC), or presence of malignant cells on cytology in patients with a previous diagnosis of cancer and a high pretest probability of having MPE.
In our results, although 82.5% of bloody fluids showed a positive cytology, this percentage was not significantly different than that observed in the non-bloody fluids (82.4%). However, our population comprised patients with a prior diagnosis of cancer and a higher pretest probability of having an MPE. So, while bloody pleural effusions may be suggestive of positive cytology in a general population of patients presenting with pleural effusions, it does not appear to be useful in this regard in cancer patients.