L. Y. Smucker1,2, L. R. Henry2, U. W. Von Holzen1,2, A. N. Hardy2, M. J. Minarich2, R. E. Schwarz1,2 1Indiana University School Of Medicine,South Bend, IN, USA 2Goshen Center for Cancer Care,Goshen, IN, USA
Introduction: Patients requiring visceral resections for cancer therapy are at risk for complicated postoperative outcomes, dependent on disease-specific, procedure-related, constitutional and comorbidity factors. Predicting an individual's morbidity risk has been traditionally challenging, and exploration of predictive risk assessment tools is worthwhile to optimize care strategies.
Methods: Data from patients undergoing major visceral resection within a cancer center-based surgical practice were queried to generate scores or parameters of potential value for risk prediction, including clinicopathologic variables, ASA score, radiographic scan-based morphometrics, Charlson comorbidity index (CCI), modified frailty index (MFI), P-POSSUM scores and NSQIP scores. Association with untoward outcomes was tested via nonparametric and parametric group comparisons, chi-square statistics and logistic regression analyses.
Results: Two hundred forty patients who underwent visceral resection during 4 years were analyzed. There were 107 women and 133 men, with a median age of 67 (range: 17-98) and a malignancy in 90%. Procedures included 28 upper GI (12%), 62 lower GI (26%), 69 pancreatic (29%), 54 hepatobiliary (22%), and 27 other resections (11%); eight percent of resections were multivisceral, and 4% emergent. Mean POSSUM morbidity risk (63%) and death risk (10.8%) were high. Above average NSQIP scores were observed in 49% for any complication, 46% for severe complication, and 57% for risk of death. There were 125 complications (53%), of which 54 were severe (grade 3+, 23%) including 5 lethal events (2%). The median length of stay (LOS) was 8 days (IQR: 5). Discharge to home failed in 16%. Univariate score associations with any complication (in descending order of significance, with p values) were seen for POSSUM (p=0.001), NSQIP (0.001), visceral fat (0.006) CCI and MFI (both 0.03), but not for ASA, sarcopenia index or other morphometrics. Serious complications correlated with NSQIP (p=0.0008), sarcopenia visceral fat obesity index (SOVFI, 0.004), visceral fat (0.02), MFI (0.02), POSSUM and CCI (both at 0.04), but not ASA. Scores that correlated with postoperative deaths included average NSQIP risk and ASA (both at p<0.0001), POSSUM (0.003), SOVFI (0.009), visceral fat (0.02), MFI (0.02), spleen volume index (0.04), abd. depth (0.04) and girth (0.05), but not the actual calculated NSQIP death risk or CCI scores. Failure of discharge to home was linked to both NSQIP and MFI scores and to functional status (at p <0.0001), CCI (0.007), POSSUM morbidity score (0.009) and marital status (0.03). When controlled for other variables, no score retained significant association for any complication; NSQIP score remained significant for serious complications (p=0.046), and POSSUM was predictive of death (p=0.02).
Conclusion: Preoperative morbidity risk scores correlate generally well with postoperative untoward outcomes and thus serve well for risk adjustments. Their performance is limited when other clinicopathologic variables are accounted for. NSQIP and POSSUM scores emerged as superior to other score metrics, but still appear to be insufficient to be utilized for prospective treatment decisions.