Results
Four critically ill TTP patients with confirmed or suspected infections receiving cefoperazone/sulbactam monotherapy were enrolled. The median age was 55 years. Table 1 lists the demographic clinical features in these patients. Of the 4 patients, mean (±SD) serum albumin concentration was 34.75±1.64g L-1, mean total bilirubin (TBIL) was 49.2±13.53 μmol L-1 and mean creatinine clearance (Ccr) was 55.5±12.36 μmol L-1. Patient 2 and 3 had a Ccr of 36 and 54 ml min-1, respectively. The Ccr of patient 1 and 4 were above 60 mL min-1. Three patients recovered and one patient died of sepsis. Mean duration and plasma volume exchanged per session were 96±10 min (85∼112 min) and 1970±36 mL (1910∼2000 ml), respectively.
The QPE of CFP was 395±147 mg (268∼645 mg) and fe % was 11.38±3.18% (8.43~16.75 %). For SUL, the QPE was 35±11 mg (20~51 mg) andfe % was 2.74±1.13 % (1.35~4.28 %)(Table 2) . Noteworthy, a positive correlation was found between the amount of drug removed and plasma drug concentration of CFP and SUL before TPE, but there were no statistically significances (r= -0.895, p=0.056 for CFP, and r= -0.821, p=0.179 for SUL) (Figure 2).
KelTotal and Kelpatient were described in Table 2 , respectively. For CFP, KelTotal and Kelpatient were 0.4±0.14 and 0.26±0.13 h-1, which correspond to T1/2 of 3.73±1.27 and 3.24±1.57 h; estimated KelTotal and Kelpatient for SUL were 0.65±0.15 and 0.46±0.2 h-1, which correspond to T1/2 of 1.71±0.26 and 1.72±0.26 h, respectively. (Figure 1).The average KelTotal for CFP and SUL were estimated to be 0.5-fold higher than that for Kelpatient. However, no significant differences were found in T1/2a for CFP and SUL between the two sessions (P=0.974 and P=0.967). The fraction of CFP and SUL eliminated due to TPE (%CLPE) were 27.71±10.8% and 6.16±2.16%, respectively.
Pharmacokinetics of CFP and SUL were also compared on both sessions (Table 3).AUC0-8 of CFP on session I and session II were 1532.8±768.95 and 1411.4±789.43 mg×h L-1, respectively (p=0.855). AUC0-8 of SUL on session I and session II were 110.47±36.22 and 115.12±39.81mg×h L-1, respectively (p=0.887) (Table 3) . There were little differences in pharmacokinetic parameters (AUC0-8, Vd, and t1/2) between session I and session II. We noticed that the maximal peak concentration of session I was higher than that of session II, it was thought to be the cause of the pharmacokinetic instability. We did not observe a redistribution phenomenon occurred at two hours after the TPE procedure