Discussion

This study is the first comprehensive report of the prevalence and relative prognostic value of vascular, cardiac, lung parenchymal and mediastinal changes on CT in PAH and its major subgroups. Within subgroups, different vascular and cardiac configurations are observed, reflecting the heterogeneity of PAH. We have demonstrated that cardiac and vascular morphology can predict outcome in PAH and CT measures such as IVC area, and pleural effusion/septal lines are independent prognostic markers.

In addition to established measures, quantifiable CT findings including elevated RV to LV ratio, deviated IV septum, dilated RA, RV free wall hypertrophy, pericardial/pleural effusions, contrast regurgitation into the hepatic veins and GGO occur frequently in PAH and major subgroups.

Different CT patterns were observed in PAH subgroups but none were diagnostic. IPAH and PAH-CHD-Eisenmenger have a higher proportion of severe RA dilatation, contrast regurgitation into hepatic veins and ground glass opacities. While more patients with IPAH have a deviated IV septum and greater RV to LV ratio, the Eisenmenger group had a higher PA to Ao ratio, PA calcification, mural thrombus and dilated collateral vessels. Although the higher RV to LV ratio in IPAH may appear counterintuitive, this may reflect better preserved RV function in the conditioned RV of the Eisenmenger patient and the off-loading effect of a right-left shunt. We noted that central distribution of ground glass opacity was frequent in patients with PAH-SSc whereas a centrilobular pattern was more common in IPAH and PAH-CHD-Eisenmenger.

CT features predict outcome in treatment naive PAH including RV to LV ratio, right atrial size, IV septal position, inferior vena caval size, pericardial/pleural effusions, mediastinal lymphadenopathy and septal lines. Most of these parameters predict survival in major subgroups (see online supplementary table S4). IVC area the presence of pleural effusions/septal lines were predictors of outcome independent of pulmonary haemodynamics and WHO functional class, highlighting the prognostic value of CT.2 In PAH, RV function is a prognostic determinant. IV septal displacement and dilatation of RV as shown by a high RV to LV ratio on CT reflect a failing RV and it is not surprising they have prognostic significance. A failing RV and tricuspid regurgitation elevate right atrial pressure. Inferior vena caval size was demonstrated to be a strong predictor of outcome. Vena caval diameter reflects the pressure in the RA2 which is an important prognostic marker in PAH. Elevated right atrial pressure impedes mediastinal lymphatic and venous drainage resulting in the development of septal lines, mediastinal lymphadenopathy and pleural and pericardial effusions. These features are not surprisingly associated with a poor outcome.

Ground glass attenuation is described in patients with PH.3 Our results show that GGO is seen in all major subgroups of PAH. Centrilobular pattern of GGO correlated with poor outcome following treatment in patients with PAH in a previous study.4 In our study, this prognostic observation was restricted to IPAH and was not seen in other major subgroups. We found a lower prevalence of GGO/septal lines in PAH-SSc than may have been expected but this reflects the rigorous way that patients with SSc with coexisting lung disease are excluded in the ASPIRE Registry, with this group having an FVC of 95±14% (see online supplement).

So how does an appreciation of CT characteristics in PAH help the physician? First, although there is overlap in the different forms of PAH, the presence of a ground glass pattern affecting the central part of the lung and sparring the periphery with oesophageal dilation should raise the possibility of underlying systemic sclerosis, whereas a centrilobular ground glass pattern with a very large PA and luminal calcification should suggest the possibility of an underlying congenital heart problem and prompt further investigation to confirm or refute these diagnoses. Second, the presence of adverse prognostic features on CT such as the presence of a pleural effusion, septal lines or a dilated IVC should alert the physician that the patient falls into a poor prognostic group prompting urgent referral for specialist assessment and treatment. Given that CT results in radiation exposure and there are other validated biomarkers which are sensitive to change, its use in follow-up is not recommended. Whether a reduction in IVC area or resolution of pleural effusions/septal lines is of prognostic importance is not answered by this study. Finally, a centrilobular ground glass pattern is very common in PAH and its presence should alert the physician to the possibility of PAH if the scan was performed for unexplained breathlessness and be a prompt to look for other CT features such as PA and RV enlargement, which are seen in PH.

There are limitations to our study. We used non-gated axial images. However, the purpose of our study was to evaluate uncomplicated measurements in a non-gated CT frequently performed as a first line examination. Second, pulmonary veno-occlusive disease (PVOD) has several overlapping clinical and pathological features with IPAH, which are difficult to recognise without lung biopsy or transplantation. It is possible that patients with PAH in our cohort may have had PVOD, although the number of patients found to have histological evidence of PVOD in a previous study at postmortem was low.5 In this cohort of patients, the majority of patients were in WHO functional capacity III and IV, comparable with the majority of Registry data from this era. Some more recent reports have identified a higher proportion of patients with earlier stage disease.

In conclusion, we report features of PAH and its major subgroups on a pretreatment CT and demonstrate that CT measures including the presence of pleural effusion/septal lines and inferior vena caval area predict outcome. Finally, GGO is commonly seen in PAH and its presence on a CT performed for breathlessness should raise the possibility of this diagnosis.