The investigators hypothesized that pulse pressure variation (PPV) could predict fluid responsiveness in spontaneously breathing patients undergoing anesthesia induction.

Pulse pressure variation (PPV) is known to be a dynamic indicator to predict fluid responsiveness in mechanically ventilating patients. PPV is generated by cardiopulmonary interaction with respiratory cycles. In spontaneous breathing, the cardiopulmonary interaction may lead to PPV by inverted manner. In this study, the investigators will evaluate the ability of PPV to predict fluid responsiveness in spontaneously breathing patients. Before induction of anesthesia, radial artery cannulation is performed, and FloTrac/Vigileo (Edwards Lifesciences Corp., Irvine, USA) system is connected. And then, the investigators will measure baseline data including invasive blood pressure, pulse pressure, cardiac index, stroke volume variation in two different conditions: during tidal spontaneous breathing and during forced spontaneous breathing. All hemodynamic measurement will be repeated after infusion of fluid using 6mL/kg of hydroxyethyl starch solution (Voluven), and will be averaged over 3 minutes. Patients whose cardiac index increased by >15% to fluid infusion is defined as responders. The investigators will compare the hemodynamic parameters between responders and non-responders. To evaluate the ability to predict volume responsiveness, receiver operating characteristics curves of the PPV will be calculated.

Inclusion Criteria:

• patients undergoing thoracic surgery

Exclusion Criteria:

• preoperative arrhythmia
• left ventricular ejection fraction <40%
• intra-cardiac valve disease
• intra-cardiac shunt
• pulmonary artery hypertension
• severe peripheral vascular obstructive disease
• decreased pulmonary function