Diagnostic value of abdominal drainage in individual risk assessment of pancreatic fistula following pancreaticoduodenectomy

BackgroundThe use of prophylactic abdominal drainage following pancreaticoduodenectomy (PD) is controversial as its therapeutic value is uncertain. However, the diagnosis of postoperative pancreatic fistula (POPF), the main cause of PD-associated morbidity, is often based on drain pancreatic amylase (DPA) levels. The aim of this study was to assess the predictive value of DPA, plasma pancreatic amylase (PPA) and serum C-reactive protein (CRP) for diagnosing POPF after PD. MethodsPatients undergoing PD with prophylactic drainage between 2008 and 2012 were studied prospectively. DPA, PPA and CRP levels were obtained daily. Differences between groups with clinically relevant POPF (International Study Group on Pancreatic Fistula (ISGPF) grade B/C) and without clinically relevant POPF (no POPF or ISGPF grade A) were evaluated. Receiver operating characteristic (ROC) analyses were performed to determine the value of DPA, PPA and CRP in prediction of POPF. Risk profiles for clinically relevant POPF were constructed and related to the intraoperative pancreatic risk assessment. ResultsFifty-nine (18<bold>7</bold> per cent) of 315 patients developed clinically relevant POPF. DPA, PPA and CRP levels on postoperative day (POD) 1-3 differed significantly between the study groups. In predicting POPF, the DPA level on POD 1 (cut-off at 1322 units/l; odds ratio (OR) 24<bold>61</bold>, 95 per cent confidence interval 11<bold>55</bold> to 52<bold>42</bold>) and POD 2 (cut-off at 314 units/l; OR 35<bold>45</bold>, 14<bold>07</bold> to 89<bold>33</bold>) was superior to that of PPA on POD 1 (cut-off at 177 units/l; OR 13<bold>67</bold>, 6<bold>46</bold> to 28<bold>94</bold>) and POD 2 (cut-off at 98 units/l; OR 16<bold>97</bold>, 8<bold>33</bold> to 34<bold>59</bold>). When DPA was combined with CRP (cut-off on POD 3 at 202 mg/l; OR 16<bold>98</bold>, 8<bold>43</bold> to 34<bold>21</bold>), 90<bold>3</bold> per cent of postoperative courses could be predicted correctly (OR 44<bold>14</bold>, 16<bold>89</bold> to 115<bold>38</bold>). ConclusionThe combination of serum CRP and DPA adequately predicted the development of clinically relevant pancreatic fistula following PD. Serum C-reactive protein and drain amylase are useful