Results
The derivation cohort included 273 consecutive SE with 90 females (33%) and 183 males. Mean ± standard deviation (SD) age was 60±11 years. The derivation cohort included 250 patients with non-ischemic SE and 23 with tests suggestive of myocardial ischemia. All patients were followed up for a mean 18±20 months.
The WMSI was calculated for both groups. The median WMSI for the non-ischemic group was 1 (1-1) and was 1.65 (1.50-1.83) for the ischemic cohort (p<0.0001 for the difference). The SV was estimated at rest and post exercise and the ΔSV calculated. The mean ΔSV for the non-ischemic group was 24.0ml (22.6 to 25.4) compared to 3.4ml (-1.1 to 8.2), p<0.0001 for the difference. A receiver operator curve (ROC) analysis comparing non-ischemic and ischemic SE by ΔSV showed that a ΔSV of less than 11ml had an optimal area under the curve (AUC) of 0.95, with a sensitivity of 95% and a specificity of 99%, p < 0.0001) for myocardial ischemia. See Figure 3. Examples of the Doppler tracings used to estimate the normal and abnormal ΔSV values are shown in Figures 1 & 2.
Indexing for SV (SV divided by the body surface area) was also performed. The ROC analysis using the Δ indexed SV (ΔSVi) resulted in an optimal AUC of 0.78, with a sensitivity of 72% and a specificity of 85%, p=0.001 (See Figure 4).
Based on these data, a cut-off value for a “normal ΔSV” was set as >10ml. The ΔSVi was not used, as it appeared to be a less discriminative measure.
To confirm there was no change in the LVOT diameter at the level of the aortic valve before and after exercise, these measurements were taken for all patients in the derivation cohort (n=273). The mean pre-exercise diameter was 2.25cm (95% CI 2.24-2.27) and was 2.26cm (2.24-2.27) post-stress, p=0.33, confirming that there was no significant change.
The validation cohort consisted of 1093 consecutive SE that were available for analysis, with 376 females (34%) and 717 males. Mean ± standard deviation (SD) age was 59±12 years. The baseline characteristics are detailed in Table 1. There were 1000 patients with non-ischemic SE, and 93 patients with SE suggestive of myocardial ischemia. Total follow up was for 20,460 patients-months (up to 5 years per patient, for a mean 37±23 months). The SV and CO data for non-ischemic and ischemic stress echocardiograms are listed in Table 2. The estimated SV was able to be performed in approximately 94% of stress echocardiograms. Limiting factors included post exercise image quality, inadequate Doppler envelopes (with subsequent inadequate VTI measurement) and irregular electrical rhythms, especially bigeminy.
The mean maximum heart rate achieved by patients in the validation cohort was 152±20bpm, which represented 98±19% of the MPHR. The mean heart rate at the time the SV post was acquired was 108±38bpm, which was 68±25% of the maximum predicted heart rate. For the non-ischemic patients, the heart rate was 108±39, and for the ischemic patients it was 105±15, p=0.19 (see table 2).
Prognostic events for the follow-up period in the validation cohort were subsequently analysed based on patient SV response to exercise, using the ΔSV. Utilising the criteria from the derivation cohort, patients were divided into two (2) groups: Group 1 had a normal ΔSV response to exercise (>10ml) and group 2 had an abnormal ΔSV response (≤10ml). The baseline characteristics of these two groups are detailed in Table 3. The vast majority of the ischemic SE in were in group 2 (98% of the ischemic SE).
Multivariate Cox proportional hazard regression analysis demonstrated that an abnormal ΔSV response (ΔSV ≤10ml) resulted in a hazard ratio (HR) adjusted for age, sex, exercise capacity and Framingham risk score of 10.3, with 95% confidence intervals (CI) of 5.6 to 19.1, p<0.0001, for the combined adverse cardiac endpoints, adjusted for age, gender, exercise capacity, ejection fraction and Framingham risk score. The Kaplan Meier curves show early and on-going separation (see Figure 6).
Cardiac deaths were rare events during follow up. None were seen in the normal ΔSV group and three (3) noted in the abnormal ΔSV cohort. Analysis of these limited data suggested a trend to increased mortality in the abnormal ΔSV group with a HR of 12.4 (95% CI 0.85 to 18.0), p=0.066.
The HR for subsequent invasive catherization based on a ΔSV ≤10ml was 4.92 (95% CI 3.76 to 6.45) p<0.0001, noting that the overwhelming majority of these patients had an ischemic SE. These patients were more likely to proceed to invasive catheterization on a clinical basis, due to the SE result.
There were 99 patients who had an abnormal ΔSV response who had a non-ischemic SE. These patients were shown to have increased risk of an adverse cardiac event when compared to the 867 patients with a non-ischemic SE who had a normal ΔSV exercise response. The HR was 3.2 with 95% CI 1.1-9.3, p=0.035. (see Figure 7) The majority of events appeared to be ischemic in nature (angina pectoris or need for percutaneous coronary intervention). Using a lack of adverse cardiac outcomes as the baseline for a true negative test, the overall negative predictive value for a normal SV response (ΔSV > 10ml) was 94%.