Prognostic Value of Effort-independent Cardiopulmonary Exercise Test Variables for Postoperative Complications After Elective Colorectal Surgery

After resection for colorectal carcinoma, >30% of the patients develop a complication during admission or within 30 days after surgery. Several studies indicate that preoperative aerobic fitness, as objectively measured by a maximal cardiopulmonary exercise test, is associated with postoperative complications, in which a lower aerobic fitness indicates a higher risk for complications. The most used CPET variables, oxygen uptake (VO2) at peak exercise (VO2peak) and VO2 at the ventilatory anaerobic threshold (VO2VAT), have specific limitations. For a valid VO2peak, a maximal effort is required and VO2VAT determination is subjective and cannot be determined in all patients. Therefore, this study aims to explore the association of submaximal (effort-independent) preoperative CPET variables that are determinable in all patients, specifically the relation between minute ventilation and carbon dioxide production (VE/VCO2-slope) and the oxygen uptake efficiency slope (OUES), and postoperative outcomes in patient undergoing colorectal surgery.

Participants An explorative study will be carried out using retrospectively collected data from patients who underwent preoperative CPET in Medisch Spectrum Twente (MST), Máxima Medical Center (MMC), Maastricht University Medical Center+ (MUMC+), and VieCuri Medical Center (VMC).

Patient characteristics and outcome measures The following baseline patient characteristics will be collected: sex, age, body height, body mass, body mass index (BMI), nutritional status assessed by the short nutritional assessment questionnaire (SNAQ) score, smoking status (current, former, never), use of beta-blocker (yes/no), veterans-specific activity questionnaire score, location, type and stage of the tumor, American Society of Anesthesiologists (ASA) score (I-IV), Charlson comorbidity index (divided into three groups: 0, 1, and 2+), and type of surgical resection.

CPET data will be interpreted by two trained and experienced clinical exercise physiologists. The variables VO2peak, VO2VAT, VE-VCO2-slope, and OUES will be determined.

Outcome measures of interest are postoperative complications within 30 days after surgery and length of hospital stay. The severity of any postoperative complication will be scored using the Clavien-Dindo classification of complications (grade 1-5). A postoperative complication is defined as a Clavien-Dindo grade of 1 or higher. A grade 3-5 complication is defined as a severe complication.

Statistical analysis Receiver operator curve (ROC) analysis will be used to assess the independent ability of the VO2peak (mL/kg/min), VO2VAT (mL/kg/min), VE/VCO2-slope, and OUES normalized for body mass (OUES/kg) to discriminate between patients with and without 30-day postoperative complications. The optimal cut-off point is based on our preference to have primarily a high sensitivity (>0.8) with a reasonable specificity (>0.5), as we aim to detect almost all high-risk patients that might benefit from a preoperative intervention (e.g., exercise prehabilitation). Forward stepwise multivariable logistic regression analyses will be performed to investigate the prognostic value of VO2peak, VO2VAT, VE/VCO2-slope, and OUES/kg for 30-day postoperative complications separately for all complications (Clavien-Dindo grade of 1 or higher) and severe complication (Clavien-Dindo grade of 3-5). In case the baseline demographics are associated with 30-day postoperative complications (p<0.200), they will be tested for their association with VO2peak, VO2VAT, VE/VCO2-slope, and OUES/kg (p<0.200) using Pearson's r or Spearman's rho correlation coefficients, or paired sample t-tests, Mann Whitney U test, or one-way ANOVA, as appropriate. The logistic regression models will be adjusted for the potential confounders in a forward stepwise procedure.

Based on the optimal cut of points extracted from the ROC curves for each CPET variable, Kaplan-Meijer curves will be constructed to evaluate the univariate relationship between VO2peak, VO2VAT, VE/VCO2-slope, and OUES/kg and length of hospital stay. The Log-rank test will be used to compare survival curves.

Data will be analyzed with the Statistical Package for the Social Sciences for Windows (version 23.0; IBM, SPSS Inc., Chicago, IL, USA). Continuous data will be presented as mean with standard deviation, or as median with interquartile range (IQR), as appropriate. Categorical data will be summarized by frequency and percentage. A p-value <0.05 will be considered statistically significant.