Introduction: In contrast with pancreatoduodenectomy, the benefit of early drain removal (EDR) for distal pancreatectomy (DP) is debated. This study explores the results of EDR following DP in light of the current evidence on clinically-relevant fistula (CR-POPF) prediction. Method: Outcomes of DPs performed at two institutions (2013-2017) were compared between de facto "early" (≤POD5) and "late" (>POD5) drain removal groups. Multivariable regression and propensity-score matching were adjusted for previously identified CR-POPF risk factors after DP [age, BMI, albumin, pathology, vascular resection, splenectomy, and postoperative-day (POD) 1 drain fluid amylase (DFA)]. Results: Drains were removed early in 261 DPs (57.2%). Post-operative outcomes, including CR-POPF, were worse in the late-removal group (Table). At multivariable analysis, POD1 DFA>2000 (OR=1.96, p=0.037) and late drain removal (OR=7.37, p< 0.001) were the only significant predictors of CR-POPF. Propensity-score matching confirmed a lower CR-POPF rate in the early-removal group (8% vs 41.4%, p< 0.001). Among patients who did not develop CR-POPF, complications (Accordion≥1: 36.8% vs 71.6%, p< 0.001; Accordion≥3: 3.8% vs 11%, p=0.012) and duration of stay (9 vs 7 days, p< 0.001) were improved in the early-removal group. A POD5 DFA≤50 and >5000 had the greatest negative (86.8%) and positive (56.5%) predictive value for CR-POPF, respectively, while a cut-off of 100 provided the best CR-POPF overall prediction. Conclusions: This study represents the largest examination of drain management for DP, and substantiates EDR as a "best practice" following DP. POD5 DFA can aid in determining which patients will develop CR-POPF, and benefit from longer drain duration.Tabled 1FP27-02 TablePatients' Characteristics and Outcomes of the Early and Late Drain Removal Cohorts - Univariable AnalysisVariableOverall Outcomes (N - 456)Removal ≤ POD5 (N = 261)Removal > POD5 (N = 195)p-valueAge (median, IQR)62 (52-70)64 (53-71)59 (49-69)p = 0.020ASA I–II62.7% (286)66.1% (150)73.9% (136)p = 0.086 III–IV27.4% (125)33.9% (77)26.2% (48)BMI Normal (< 25)45.6% (208)45.1% (110)51% (98)p = 0.463 Overweight (25-29.9)37.1% (169)40.6% (99)36.5% (70) Obese (≥30)12.9% (59)14.3% (35)12.5% (24)POD1 DFA(median, IQR)1813 (470-5095)938 (319-2063)5055 (2061-7500)p< 0.001POD1 DFA > 200046.8% (182)26% (59)75.9% (123)p < 0.001POD Drain Removal (median, IQR)5 (3-10)4 (3-5)13 (7-22)p < 0.001Clinically-relevant fistula20.8% (95)6.9% (18)39.5% (77)p < 0.001Any complication (Accordion ≥1)54.4% (248)41.9% (95)83.2% (153)p < 0.001Severe complications (Accordion ≥ 3)15.8% (72)11.5% (26)25% (46)p< 0.001Duration of Stay (median, IQR)8 (6-11)7 (6-9)10 (8-17)p < 0.001Readmission12.7% (52)8.4% (19)18% (33)p - 0.003Reoperation7.1% (29)2.6% (6)12.6% (23)p < 0.001Death1% (4)0.4% (1)1.6% (3)p = 0.222POD: postoperative day, DFA: drain fluid amylase Open table in a new tab POD: postoperative day, DFA: drain fluid amylase

BACKGROUND:Histological subtype influences both prognosis and patterns of treatment failure in retroperitoneal sarcoma. Previous studies on the efficacy of neoadjuvant radiotherapy (NRT) have incorporated multiple histological types with heterogeneous tumour biology. The survival impact of NRT specifically for patients with retroperitoneal liposarcoma is poorly defined.METHODS:Patients who underwent resection with curative intent for retroperitoneal liposarcoma and who received NRT or surgery alone were identified in the US National Cancer Data Base (2004-2013). Cox regression was used to identify co-variables associated with overall survival. NRT and surgery-alone cohorts were matched 1 : 1 by propensity scores based on the survival hazard on Cox modelling. Overall survival was compared by Kaplan-Meier estimates.RESULTS:A total of 2082 patients with retroperitoneal liposarcoma were identified; 1908 underwent surgery alone and 174 received NRT before surgical resection. Median tumour size was 22·0 cm and 34·9 per cent of tumours were high grade. In the unmatched cohort, NRT was not associated with improved overall survival (χ2  = 3·49, P = 0·062). In the propensity score-matched cohort, NRT was associated with an improvement in survival (median overall survival 129·2 versus 84·3 months; P = 0·046; hazard ratio (HR) 1·54, 95 per cent c.i. 1·01 to 2·36). This effect appeared most pronounced for tumours with adjacent organ invasion (median overall survival not reached versus 63·8 months; P = 0·044; HR 1·79, 1·01 to 3·19).CONCLUSION:NRT improved survival in patients undergoing surgery for retroperitoneal liposarcoma, particularly those with high-risk pathological features.