Mortality Risk in Mild-to-Moderate Heart Failure
Mortality Risk in Mild-to-Moderate Heart Failure
We have shown that a composite risk score using variables from CPET outperforms the traditional single variable approach in predicting outcome in patients with mild-to-moderate HF. We validated our model against the HFSS and found that variables collected solely from CPET significantly outperformed the well-established HFSS. The HFSS includes only one CPET-related variable, peak oxygen uptake. The other six variables are derived from a combination of standard investigative methods including echocardiography, electrocardiography, blood pressure monitoring, blood biochemistry, and patient history (aetiology of disease). Recently, Goda et al showed that the HFSS outperformed peak oxygen uptake alone for stratifying risk in chronic heart failure (CHF) patients in the presence of implantable cardioverter-defibrillators and/or cardiac resynchronisation therapy. Risk stratification models should investigate the efficacy of combinations of CPET-related variables in the era of device therapy.
Historically, most exercise-related composite risk scores have been developed in patients with coronary artery disease prior to the widespread adoption of CPET. Perhaps the most accepted integrated exercise risk score is the Duke Treadmill Score (DTS) which has both prognostic and diagnostic predictive power. The DTS combines exercise time (using a Bruce protocol) with ECG abnormalities (ST segment depression) and symptoms of angina. It was originally described in patients with coronary artery disease, though successful validation in other subgroups has been reported. Few composite risk scores have been developed which have specifically included CPET variables for risk stratification among HF patients.
Risk models using both non-invasive and invasive data (with and without catheterisation data) in combination with peak VO2 was developed in 268 patients with advanced HF. The models were prospectively validated on 199 similar patients, and the non-invasive model performed well in both samples, and model performance did not improve with the addition of invasive catheterisation-related data. The authors concluded that the selection of candidates for cardiac transplantation may be improved by using a non-invasive risk-stratification model which included peak VO2. Myers et al recruited 710 patients with HF (80% men; 56±13 years; LVEF 33±13%) from four different institutions in Italy and the USA. CPET-derived variables included in the risk score included peak VO2, VE/VCO2 slope, OUES, resting PETCO2, HR recovery and chronotropic index. The VE/VCO2 slope (optimal cut point ≥34) was the strongest predictor of risk, and attributed a relative weight of seven points. A cumulative CPET score >15 points was associated with an annual mortality of 27% and a relative risk of 7.6. The authors recently published a validation study of their original work in a larger sample size using a different statistical approach. Our study extends these findings by including other independent predictors of mortality in our risk algorithm such as ventilatory variables, EOV and VEqCO2 nadir, and circulatory-related variables including HRR and PCP.
Guazzi et al used peak VO2, VE/VCO2 slope, EOV, to develop a prognostic risk score in 695 patients with HF. EOV was the strongest single predictor of cardiac mortality. The greatest contribution to the risk score was provided by EOV, followed by VE/VCO2 slope, whereas peak VO2 added minimal prognostic value. However, one of the major limitations of the study was that it only included three variables and did not include other potentially important predictive variables such as OUES, VEqCO2 nadir, and PCP. Recently, Italian clinicians published data from a multicentre study designed to build a new risk score for patients with systolic HF, integrating CPET measures with established clinical, laboratory and echocardiographic risk factors in order to identify patients at risk of cardiovascular death and urgent heart transplant. The MECKI score combined % predicted peak VO2, VE/VCO2 slope, LVEF, haemoglobin, sodium, and modification of diet in renal disease. A ROC analysis of the MECKI score for predicting cardiovascular death and heart transplant was 0.804 at year 1 which decreased to 0.760 by year 4. The MECKI score appears to further advance holistic risk models such as the HFSS and the HF-Action Predictive Risk Score Model by the inclusion of a well-established CPET-related risk algorithm.
Discussion
We have shown that a composite risk score using variables from CPET outperforms the traditional single variable approach in predicting outcome in patients with mild-to-moderate HF. We validated our model against the HFSS and found that variables collected solely from CPET significantly outperformed the well-established HFSS. The HFSS includes only one CPET-related variable, peak oxygen uptake. The other six variables are derived from a combination of standard investigative methods including echocardiography, electrocardiography, blood pressure monitoring, blood biochemistry, and patient history (aetiology of disease). Recently, Goda et al showed that the HFSS outperformed peak oxygen uptake alone for stratifying risk in chronic heart failure (CHF) patients in the presence of implantable cardioverter-defibrillators and/or cardiac resynchronisation therapy. Risk stratification models should investigate the efficacy of combinations of CPET-related variables in the era of device therapy.
Historically, most exercise-related composite risk scores have been developed in patients with coronary artery disease prior to the widespread adoption of CPET. Perhaps the most accepted integrated exercise risk score is the Duke Treadmill Score (DTS) which has both prognostic and diagnostic predictive power. The DTS combines exercise time (using a Bruce protocol) with ECG abnormalities (ST segment depression) and symptoms of angina. It was originally described in patients with coronary artery disease, though successful validation in other subgroups has been reported. Few composite risk scores have been developed which have specifically included CPET variables for risk stratification among HF patients.
Risk models using both non-invasive and invasive data (with and without catheterisation data) in combination with peak VO2 was developed in 268 patients with advanced HF. The models were prospectively validated on 199 similar patients, and the non-invasive model performed well in both samples, and model performance did not improve with the addition of invasive catheterisation-related data. The authors concluded that the selection of candidates for cardiac transplantation may be improved by using a non-invasive risk-stratification model which included peak VO2. Myers et al recruited 710 patients with HF (80% men; 56±13 years; LVEF 33±13%) from four different institutions in Italy and the USA. CPET-derived variables included in the risk score included peak VO2, VE/VCO2 slope, OUES, resting PETCO2, HR recovery and chronotropic index. The VE/VCO2 slope (optimal cut point ≥34) was the strongest predictor of risk, and attributed a relative weight of seven points. A cumulative CPET score >15 points was associated with an annual mortality of 27% and a relative risk of 7.6. The authors recently published a validation study of their original work in a larger sample size using a different statistical approach. Our study extends these findings by including other independent predictors of mortality in our risk algorithm such as ventilatory variables, EOV and VEqCO2 nadir, and circulatory-related variables including HRR and PCP.
Guazzi et al used peak VO2, VE/VCO2 slope, EOV, to develop a prognostic risk score in 695 patients with HF. EOV was the strongest single predictor of cardiac mortality. The greatest contribution to the risk score was provided by EOV, followed by VE/VCO2 slope, whereas peak VO2 added minimal prognostic value. However, one of the major limitations of the study was that it only included three variables and did not include other potentially important predictive variables such as OUES, VEqCO2 nadir, and PCP. Recently, Italian clinicians published data from a multicentre study designed to build a new risk score for patients with systolic HF, integrating CPET measures with established clinical, laboratory and echocardiographic risk factors in order to identify patients at risk of cardiovascular death and urgent heart transplant. The MECKI score combined % predicted peak VO2, VE/VCO2 slope, LVEF, haemoglobin, sodium, and modification of diet in renal disease. A ROC analysis of the MECKI score for predicting cardiovascular death and heart transplant was 0.804 at year 1 which decreased to 0.760 by year 4. The MECKI score appears to further advance holistic risk models such as the HFSS and the HF-Action Predictive Risk Score Model by the inclusion of a well-established CPET-related risk algorithm.
