Back to Annual Meeting Program
Use of Morphomic Analysis for Preoperative Risk Stratification in Patients Undergoing Major Head and Neck Cancer Surgery
Kavitha Ranganathan, MD, Jonathan Peterson, BS, Oluwatobi Eboda, BS, Shailesh Agarwal, MD, Steven Buchman, MD, Paul Cederna, MD, Stewart Wang, MD, PhD, Benjamin Levi, MD
University of Michigan, Ann Arbor, MI, USA.
Introduction: Post-operative complications following major head and neck cancer surgery (MHNCS) are associated with significant morbidity and mortality. The ability to stratify pre-operative risk based on already available objective measures could improve pre-operative counseling and decision making. Currently, pre-operative risk is assessed based on comorbidities. Traditionally, patient assessment has been performed using comorbidities. However, an emerging field of patient assessment involves the evaluation anatomic morphology to measure patient frailty. We hypothesize that the morphologic characteristics of the temporalis region in patients undergoing MHNCS can serve as a reliable marker for risk of complications.
Methods: Patients who underwent MHNCS from 2004 to 2012 with available pre-operative CT scans were included in this study. All CT scans were performed as part of routine pre-operative planning, and were not initially ordered to evaluate the temporalis region. Archived images were extracted and the three-dimensional characteristics of the bilateral temporalis regions were mapped from both sides using a previously validated method to determine the zygomatic arch thickness and TFPV. Charts were reviewed for patient characteristics (age, gender, BMI, diabetes history, smoking status, and ASA score) and post-operative complications (e.g. MI, VTE, wound breakdown, hematoma, infection). Univariate and multivariate statistical analysis was performed to determine the relationship between risk of complications and zygomatic arch thickness and/or temporal fat pad volume. We then developed receiver-operator curves (ROC) to determine the reliability of arch thickness and TFPV for evaluating complication risk.
Results: A total of 70 patients undergoing MHNCS had available CT scans. Mean zygomatic arch thickness was 2.7 mm (s.d. 0.6 mm) and mean TFPV was 1.37 cm3 (s.d. 0.92 cm3). Major post-operative complications were noted in 28% (20/70) of our patients. Patients with post-operative complications had 33% less TFPV (1.50 v. 1.01 cm3, p<0.02), and 13% smaller zygomatic arch thickness (2.82 v. 2.46 mm, p<0.02); interestingly, we noted no significant differences in other characteristics (e.g. age, BMI, diabetes history, smoking status, or ASA score), between patients which did or did not experience complications. The area-under-the-curve (AUC) for zygomatic arch and temporal fat pad ROCs were 0.709 and 0.674 respectively (Fig 1), suggesting a slightly better test for complications using zygomatic arch thickness. Multivariate analysis showed that the odds of complications was 0.43 for each cm3 increase in TFPV, and 0.24 for each mm increase in zygomatic bone thickness.
Conclusion: Head and Neck Morphomics is a powerful technique which can provide pre-operative risk assessment for major complications in Head and Neck Cancer patients based on anatomic variables as a proxy for frailty and fitness. Specifically, we demonstrated that MHNCS patients with significantly lower zygomatic arch thickness and fat pad volume had higher major complication rates. These data, either alone, or in combination with co-morbidities hold the promise of more accurately determining pre-operative risk stratification in order to better council patients and prepare surgeons to deliver the best possible clinical care.
Back to Annual Meeting Program