Table 1.
Demographic information on all the subjects used in the study.
Table 2.
Summary of the clinical signs and symptoms in the test population.
Fig 1.
A: Typical thoracic endometriosis simulating native proliferative endometrium. H&E staining, original magnification X300. B: Dilated endometrioid gland with intraluminal pseudoxanthoma cells; the surrounding stroma is inconspicuous. H&E staining, original magnification X150. C: Lymphocytic aggregate and dilated and congested blood vessels in oedematous nonspecific stroma; a distorted endometrioid gland is also present. H&E staining, original magnification X100. D: Oedematous hyalinised stroma with congested capillaries and siderophages with golden-brown pigments. H&E staining, original magnification X300. E: Ceroid macrophages with brownish finely granular cytoplasm. H&E staining, original magnification X300. F: Foamy macrophages in the stroma. H&E staining, original magnification X150. G: Cholesterol crystals and multinucleated foreign body-type giant cells. H&E staining, original magnification X300. H: Bundles of smooth muscle cells. H&E staining, original magnification X150. I: Pleural fibrosis. H&E staining, original magnification X150.
Fig 2.
Barplot showing the frequency distribution of the various histological features in the test (thoracic endometriosis) population.
The stripped red line marks the point of separation between the frequent or Group A features (darker bars) affecting more than 13 (>13/26; 50%) patients from the features (lighter bars) affecting less than 13 (<13/26; 50%) patients.
Table 3.
Comparison of percentage frequency distribution of histological features in the test and control populations, along with the corresponding Chi-squared test statistic of independence, its P-value and the Cramér’s V coefficient for strength of association.
Fig 3.
Lollipop plot showing the strength of association with disease of the various histological features using the Cramér’s V coefficient.
The dashed lines divide the histological features into those of high association (brick red), medium association (coral) and low association (blanched almond).
Fig 4.
Boxplot of distribution of scores based on 3 groups of features: all features (Group Omnibus), those with association to disease by Chi-squared test (Group B) and those with high strength of association by Cramér’s V coefficient (Group C).
The tips of the whiskers in these boxplots represent the minimum and maximum scores of the various groups.
Table 4.
The traditional criterion and two new suggested alternative criteria for the diagnosis of thoracic endometriosis.
These new alternative criteria are based on the features of Groups C and B.
Fig 5.
Venn plot showing the overlap of the 6 frequent features (Group A), the 8 features with significant association to disease by Chi-squared test (Group B), and the 4 features with high strength of association to disease by the Cramér’s V coefficient (Group C).