Background ContextSacral tumors are incredibly rare lesions affecting fewer than one in every 10,000 persons. Reported perioperative morbidity rates range widely, varying from 30% to 70%, due to the relatively low volumes seen by most centers. Factors affecting perioperative outcome following sacrectomy remain ill-defined. PurposeTo characterize perioperative outcomes of sacral tumor patients undergoing sacrectomy and identify independent risk factors of perioperative morbidity Study Design/SettingRetrospective cohort study at a single comprehensive cancer center Patient SampleConsecutively treated sacral tumor patients (primary or metastatic) undergoing sacrectomy for oncologic resection between April 2013 and April 2020 Outcome MeasuresPerioperative complications, hospital length of stay, non-home discharge, 30-day readmission, and 30-day reoperation MethodsDetails were gathered about tumor pathology and morphology, surgery performed, baseline medical comorbidities, preoperative lab data, and patient demographics. Stepwise multivariable regressions were conducted to identify independent risk factors of perioperative outcomes while evaluating predictive accuracy. Results57 sacral tumor patients were included (mean age 55.5±13.0 years; 60% female). The complication, non-home discharge, 30-day readmission, and 30-day reoperation rates were 39%, 56%, 16%, and 14%, respectively. Independent predictors of perioperative complications included ASA>2 (OR=10.7; 95%CI [1.3, 86.0]; p=0.026), radicular pain (OR=10.9; p=0.014), platelet count (OR=0.989 per 10³/μL; p=0.049), and instrumentation (OR=10.7; p=0.009). Independent predictors of length of stay included iliac vessel involvement (β=15.8; p=0.005), larger tumor volume (β=0.027 per cm³; p<0.001), a staged procedure (β=10.0; p=0.018), and S1 nerve root sacrifice (OR=15.8; p=.011). The optimal model predictive of non-home discharge included bilateral S3-S5 or higher nerve root sacrifice (OR=3.9; p=0.054), instrumentation (OR=8.6; p=0.005), and vertical rectus abdominis musculocutaneous flap closure (OR=5.3; p=0.067). 30-day readmission was independently predicted by history of chronic kidney disease (OR=26.7; p=0.021), radicular pain (OR=8.1; p=0.039), and preoperative saddle anesthesia (OR=12.6; p=0.026). All multivariable models achieved good discrimination (AUC>0.8 and R2>0.7). ConclusionClinical and operative factors were important predictors of complications and 30-day readmission, while tumor-related and operative factors accounted for most of the variability in length of stay and non-home discharge.
Read full abstract