During surgery high concentrations of supplementary oxygen are routinely administrated. However, there is increasing evidence of potential harm with liberal oxygen therapy. The hypothesis of the present study is that oxygen therapy adjusted to a normal arterial oxygen target is feasible and will attenuate the side effects of supplementary oxygen therapy. The study design is a before-and-after study in which 25 patients will follow the standard regime with high concentrations of oxygen therapy and 25 patients will be treated with oxygen to achieve a normal value of arterial oxygenation.

Background Oxygen is essential to life and supplementary oxygen therapy is routinely administrated during surgery to avoid life-threatening hypoxemia. Supplementary oxygen therapy is often given as a high and fixed fraction of inspired oxygen (FiO2) ranging from 0.60 to 1.00. As a consequence, the partial pressure of arterial oxygen (PaO2) often exceeds normal levels. There is increasing evidence of potential harm with liberal administrated oxygen therapy. Therefore, the standard regimes for oxygen therapy during surgery have to be further investigated.

The purpose of the study is to evaluate whether a regime for oxygen therapy targeting normal levels of PaO2 during elective cardiac surgery with use of cardiopulmonary bypass (CPB) and in the first postoperative day in the intensive care unit (ICU) is feasible. Additionally, metabolic changes in exhaled breath condensate and in arterial blood collected prior to surgery, before and after CPB, the first postoperative day in the ICU and the third postoperative day at the surgical ward will be evaluated.

Study Design The study design is a before-and-after study in patients scheduled for elective cardiac surgery being coronary artery bypass grafting or valve replacement or a combination of both. The first 25 patients will follow our institutional the standard regime for supplementary oxygen therapy; i.e. a minimum FiO2 of 0.60 during mechanical ventilation and at least 3 liters oxygen per minute after weaning from the ventilator in the ICU. Thereafter, another 25 patients will receive supplementary oxygen therapy to achieve a normal PaO2 defined as 10-12 kPa (75-120 mmHg) during surgery and in the ICU.

Exhaled breath condensates will be collected via a mouthpiece during spontaneous breathing prior to surgery, first and third postoperative day while the condensates will be collected via the tracheal tube during mechanical ventilation. Arterial blood samples will be collected simultaneously. Exhaled breath condensates and serum samples will be divided into aliquots and stored in a -80 degree Celsius freezer until final analyses after all patients have been included.

Metabolic changes in exhaled breath condensates and in serum samples will be analyzed using nuclear magnetic resonance (NMR) spectroscopy and mass spectrometry (MS). The spectra's will give a snapshot of all metabolites in a given sample and elucidate changes in the metabolic profiles in response to oxygen exposures.

Hypothesis The hypothesis is that oxygenation targeting is feasible and that changing the supplementary oxygen therapy to a oxygenation targeting within the normal range will reduce the FiO2 and thus the PaO2/FiO2 ratio.

Outcomes Primary outcome is the PaO2/FiO2 ratio on the third postoperative day. Secondary outcomes will be changes in metabolic profiles in exhaled breath condensates and serum samples, changes in oxygenation parameters, PaO2 and oxygen saturation, changes in time to weaning from the ventilator in the ICU, changes in FiO2 and in supplementary oxygen therapy after weaning from the ventilator, and changes in postoperative complications defined as pneumonia, arrhythmias, re-operation, and re-admission to the ICU.

Statistical analyses This is a pilot study being a feasibility before-and-after study. Based on institutional data a power calculation is performed estimating an increase in the PaO2/FiO2 ratio from 30 ± 7 kPa with the standard regime for oxygen therapy to 36 ± 7 kPa with a oxygenation targeting approach and with an alpha of 0,05 and a power of 80% (1-beta), 23 patients are needed in each group. To cover to dropouts at total of 50 patients, 25 patients in each group, will be recruited.

Data will be assessed for normality. Continuous normally distributed variables will be compared using Student t tests and reported as mean and standard deviation, while non-normally distributed data will be compared using Wilcoxon rank-sum tests and reported as median and interquartile range. All analyses will be performed by using Stata (Metrika Consulting AB, Stockholm, Sweden), Matlab.

The composition of metabolomics will be compared by multivariable analysis. For multivariate analysis, various unsupervised and supervised analyses will be carried out in Matlab (The MathWorks Inc., Natick, United States) and Python (Python Software Foundation, Delaware, United States) software with in-house scripts. Principal component analysis will be used to obtain a preliminary outlook of the data, to screen for outliers and detect clusters. Eventually, partial least squares discriminant (PLS-DA) analyses coupled to different vv rs selections algorithms will be employed to perform data reduction and classification of treatments. For validation purposes, the Venetian-Blinds cross-validation approach will be applied.

Ethics Patients will be included after an informed signed content is obtained. FiO2 of 0.60 is given routinely to patients undergoing cardiac surgery with CPB at Aalborg University Hospital. Targeting a physiological PaO2 level between 10 and 12 kPa is considered to be safe. Patients undergoing cardiac surgery is routinely maximal monitored during surgery and in the ICU. The cardiac anaesthetist in charge can charge the PaO2 target at any time.

The collection of exhaled breath condensate is painless and without any discomfort for the patient. The arterial blood sample on the third postoperative day will be performed by a cardiac anaesthetist experienced in this procedure, who will apply a subcutaneous analgesia with Lidocain 1% prior to the puncture and thus avoiding pain during the procedure. Thereafter, a manual compression will be performed for 2 minutes to minimise the risk of a haematoma. In total, four extra blood samples are collected for the metabonomics and one extra blood gas analyses. The total amount of blood extracted in the study is 12.5 mL.

All other procedures and treatments follow the standard regimes.

Inclusion Criteria:

• Scheduled for elective cardiac surgery either coronary artery bypass grafting and/or valve replacement performed with use of cardiopulmonary bypass

Exclusion Criteria:

• Left ventricular ejection fraction below 30 percent