Assessment of the risk of post-operative pulmonary complications and mortality using analytic morphometry in patients undergoing major non-cardiac surgery

Abstract

Some 20% of patients undergoing major abdominal surgery develop post-operative pulmonary complications (PPC). This is a nebulous group of respiratory pathologies that occur frequently after major abdominal surgery, significantly impacting functional recovery and survival. Effective preoperative risk stratification and prevention are crucial for optimal clinical outcomes. Thorough pre-operative assessments usually include full-body imaging. However, the valuable information obtained is not fully utilised. This untapped quantitative data provides an opportunity to identify new biomarkers (radiomarkers, morphomarkers) associated with PPC and survival, enhancing preoperative risk assessment understanding and improving patient care.

Two cohorts of patients were retrospectively analysed in this thesis: a cohort of patients with aortic pathology and a cohort of patients with colorectal cancer. The morphometric body composition parameters were analysed prospectively using Terarecon software.

I established the incidence of PPC which were higher than reported in the literature for the vascular cohort (24.2%) and within expected range for the colorectal patients (9.2%). I also established factors associated with PPC and their relation to mid-term survival. The presence of peripheral arterial occlusive disease (PAOD) and aneurysm diameter in vascular patients, as well as pre-existing chronic lung conditions (CLD), lymph node involvement at level N1 and N2, dementia, atrial fibrillation (AF), white cell count (WCC), male gender in colorectal cancer patients, are the most important independent risk factors for the development of PPC. It was established that PPC development was independently associated with survival in colorectal cancer patients.

I investigated the associations between quantitative computed tomography (CT)-derived structural lung morphology with PPC and survival. This resulted in the identification of potential new morphomarkers that could be useful in preoperative stratification of the patient’s risk. Lung tissue volumes measured at certain Hounsfield Units (HU) thresholds (high, normal, and low attenuation areas measured at HAA(VOL) -850 to -500 HU, NAA(VOL) -850 to -750 HU, and LAA(VOL) -950 HU) and airway volume were associated with PPC development and survival following vascular and colorectal cancer surgery, respectively.

Evaluation of indices of body composition including psoas, diaphragm muscles, and abdominal adipose tissue identified diaphragm thickness as an important morphomarker that may be useful in identifying patients at risk of developing PPC. I also analysed these in the context of survival, demonstrating a significant association between diaphragm thickness and total psoas muscle area (TPA) and survival in colorectal cancer patients.

The inclusion of lung morphology and body composition morphomarkers to clinical regression models alongside clinical patient-related factors improved their overall performance and predictive ability for PPC and survival. Lung morphology and body composition morphomarkers can form a useful, otherwise not used, set of factors that should be used in preoperative assessment. Further work is warranted to build predictive, rather than exploratory models using morphomarkers described in this thesis. Machine learning models could be useful in not only facilitating this, but also helping in developing new lung analysis techniques.

Defined PPC by the European Joint Task-Force for Perioperative Clinical Outcomes (EPCO), are a heterogenous group of complications with different aetiology. It is therefore unsurprising that despite their important association with mortality, the relation with lung structure and composition is not uniform. Further research is required to determine usefulness of assessment of lung morphometry and body composition in stratification of perioperative risk in