PWV measurements

Using the TT technique PWV was measurable in all individuals, however in 5 individuals, one of the sides produced grossly inaccurate results (PWV excessively high or negative), this error was slightly more common on the left (n=3) than the right (n=2).  Using the high temporal resolution QA technique, PWV was measurable in 29/30 of the study participants at the MPA, 29/30 at the RPA, and 29/30 at the LPA. Using the high spatial resolution QA technique, PWV was measurable in 29/30 participants.

Using the TT technique MPA-RPA PWV was 2.7±0.1 ms-1 while the MPA-LPA PWV was 3.3±0.2 ms-1.  There was a significant difference between the PWV using the right pulmonary artery and the left pulmonary artery: mean (95% CI) of the differences -0.55 (-1.1- -0.03), P=0.038.   

The results of the three QA techniques for both the high temporal resolution and low temporal resolution sequences are described in Table 1.  Using the high temporal resolution sequences the QATrad produced significantly higher results than the QA3 (p<0.001) or the QAInv (p<0.001), whilst there was no difference between the QA3 and QAInv techniques (p=0.41). Similar findings were observed with the high spatial resolution sequence with the QATrad produced significantly higher results than the QA3 (p=0.004) or the QAInv (p<0.001), whilst there was no significant difference between the QA3 and QAInv techniques (p=0.47).There was no difference in PWV between the two sequences using QATrad, however the higher temporal resolution yielded consistently lower PWV than the high spatial resolution sequence for both the QA3 (p=0.028) and QAInv (p=0.001).  Using the QATrad method the PWV was 1.99 ± 0.14 ms-1 in the MPA, 1.49 ± 0.11 ms-1 in the RPA and 1.37 ± 0.11 ms-1 in the LPA.  The PWV was significantly higher in the MPA compared to the RPA (mean (95% CI) of the differences: 0.49 (0.15-0.89), P=0.006) and the LPA (mean (95% CI) of the differences 0.63 (0.36-0.89), P<0.001).  There was no significant difference between the RPA and LPA (mean (95% CI) of the differences 0.12 (-0.11-0.34), P=0.29).

Table 1: Comparison of PWV between the two phase contrast sequences and 3 post processing techniques

Within scan reproducibility

Within scan reproducibility was assessed in the 20 OHVs.  Bland-Altman plots of the between scan differences are shown in Fig. 3 .  

Using the TT technique, the MPA-RPA PWV had better precision but lower accuracy than the MPA-LPA PWV (mean (95% CI) of PWV differences = 0.21 (-1.01-1.42) ms-1 and 0.13 (-3.57-3.83) ms-1 for the MPA-RPA and MPA-LPA respectively).  Using the QA technique, the QAInv technique again consistently yielded improved accuracy and precision over the QATrad and QA3 technique (mean (95% CI) of PWV differences = -0.46 (-2.39-1.47) ms-1, 0.05 (-1.68-1.77) ms-1, and 0.01 (-1.23-1.25) ms-1 for the QATrad, QA3, and QAInv of the MPA respectively; 0.17 (-0.68-1.02) ms-1, 0.19 (-1.43-1.81) ms-1 and 0.06 (-0.42-0.55) ms-1 for the QATrad, QA3, and QAInv of the RPA respectively; and -0.29 (-0.91-0.32) ms-1, -0.01 (-0.78-0.76) ms-1 and -0.06 (-0.69-0.56) ms-1 for the QATrad, QA3, and QAInv of the LPA respectively).  The high spatial resolution yielded poorer reproducibility than the high temporal resolution sequence (mean (95% CI) of QAInv PWV differences = 0.01 (0.63) ms-1, and 0.3 (0.86) ms-1 respectively).  A combination of a high temporal resolution sequence through the RPA combined with the QAInv post processing yielded the best reproducibility.

Inter-scan reproducibility

The scans were repeated at 6 months in 9 of the 10 YHVs.  This resulted in reproducible results with no significant differences in the two measurements (P>0.5).  Bland-Altman plots of the between scan differences are shown in Fig. 4 .  

Using the TT method, the MPA-RPA PWV had better precision but lower accuracy than the MPA-LPA PWV (mean (95% CI) of PWV differences = 0.56 (-0.86-1.99) ms-1 and 0.19 (-1.47-1.84) ms-1 for the MPA-RPA and MPA-LPA respectively).  Using the QA technique, the QAInv technique again consistently yielded improved accuracy and precision over the QATrad and QA3 technique (mean (95% CI) of PWV differences = 0 (-3.24-3.24) ms-1, -0.07 (-2.11-1.96) ms-1, and -0.43 (-2.18-1.32) ms-1 for the QATrad, QA3, and QAInv of the MPA respectively; -0.5 (-3.18-2.18) ms-1, -0.13 (-1.22-0.96) ms-1 and -0.52 (-1.46-0.43) ms-1 for the QATrad, QA3, and QAInv of the RPA respectively; and 0.31 (-1.51-2.14) ms-1, -0.59 (-1.84-0.67) ms-1 and -0.24 (-1.22-0.75) ms-1 for the QATrad, QA3, and QAInv of the LPA respectively).  A combination of a high temporal resolution sequence through the RPA combined with the QAInv post processing yielded the best reproducibility.

Intra-observer reproducibility

Bland-Altman plots of the between scan differences are shown in Fig. 5 .

Using the TT method, the MPA-RPA PWV had better precision but lower accuracy than the MPA-LPA PWV (mean (95% CI) of PWV differences = 0.28 (-1.63-2.19) ms-1 and 0.19 (-3.11-3.48) ms-1 for the MPA-RPA and MPA-LPA respectively).  Using the QA technique, the QAInv technique again consistently yielded improved accuracy and precision over the QATrad and QA3 technique (mean (95% CI) of PWV differences = -1.41 (-4.29-1.48) ms-1, -0.36 (-2.01-1.28) ms-1, and -1.51 (-3.74-0.73) ms-1 for the QATrad, QA3, and QAInv of the MPA respectively; -0.31 (-1.12-0.5) ms-1, -0.95 (-3.1-1.2) ms-1 and -0.58 (-1.06- -0.1) ms-1 for the QATrad, QA3, and QAInv of the RPA respectively; and -0.74 (-2.12-0.64) ms-1, -0.89 (-2.93-1.15) ms-1 and -0.78 (-1.79-0.22) ms-1 for the QATrad, QA3, and QAInv of the LPA respectively).  The high spatial resolution yielded better reproducibility than the high temporal resolution sequence (mean (95% CI) of QAInv PWV differences = -1.51 (-3.74-0.73) ms-1, and -0.70 (-2.31-0.9) ms-1 ms-1 for the high temporal resolution and high spatial resolution sequences respectively). A combination of a high temporal resolution sequence through the RPA combined with the QAInv post processing yielded the best intra-observer reproducibility.

Inter-observer reproducibility

Bland-Altman plots of the between scan differences are shown in Fig. 6 .

Using the TT method, the MPA-RPA PWV had better precision but lower accuracy than the MPA-LPA PWV (mean (95% CI) of PWV differences = 0.23 (-0.64-1.1) ms-1 and 0.02 (-0.97-1.01) ms-1 for the MPA-RPA and MPA-LPA respectively).  Using the QA technique, the QAInv technique again consistently yielded improved accuracy and precision over the QATrad and QA3 technique (mean (95% CI) of PWV differences = -0.2 (-1.86-1.47) ms-1, 0.02 (-1.25-1.29) ms-1, and 0.01 (-1.47-1.5) ms-1 for the QATrad, QA3, and QAInv of the MPA respectively; 0.01 (-1.67-1.69) ms-1, -0.67 (-2.24-0.9) ms-1 and -0.08 (-1.58-1.42) ms-1 for the QATrad, QA3, and QAInv of the RPA respectively; and 0.03 (-1.57-1.63) ms-1, -0.32 (-2.04-1.41) ms-1 and 0.14 (-1.13-1.42) ms-1 for the QATrad, QA3, and QAInv of the LPA respectively).  The high temporal resolution yielded better reproducibility than the high spatial resolution sequence (mean (95% CI) of QAInv PWV differences = 0.01 (-1.47-1.5) ms-1, and -0.18 (-2.61-2.25) ms-1 for the high temporal resolution and high spatial resolution sequences respectively).  The TT method through the right pulmonary artery yielded the best overall inter-observer reproducibility, while the combination of a high temporal resolution sequence through the RPA combined with the QAInv post processing yielded the best inter-observer reproducibility for the QA technique.

Inter-observer and intra-observer variability account for the majority of the inter-scan variability, with the impact more pronounced in the QA technique than in the TT technique.

Age

There was no difference between the 2 groups in terms of age, sex, smoking status, height, weight, BMI, or resting heart rate (See Table 2).  However the older population had a significantly higher resting systolic (YHV 113 ± 1 vs OHV 128 ± 3 mmHg, p<0.001) and diastolic blood pressure (YHV 68 ± 2 vs OHV 75 ± 2, p=0.028).  The pulmonary PWV did not differ between the young healthy volunteers and older healthy volunteers using either the TT technique (YHV 2.4 ± 0.3 vs OHV 2.9 ± 0.2, p=0.1) or the QAInv technique (YHV 0.95 ± 0.1 vs OHV 1.15 ± 0.1, p=0.2).  However a significant difference was observed between the two groups for the aortic arch pulse wave velocity (YHV 7.4 ± 1.6 vs OHV 10.7 ± 1.6, p=0.014) (See Table 3).

Table 2: Comparison of the demographic and anthropomorphic measures of the two study groups

Table 3: Effects of age on Pulmonary and Aortic PWV

Exercise

The isometric calf exercises resulted in a significant and sustained rise in heart rate from 65 ± 2 to 87 ± 3 (P<0.0001) and in blood pressure from 113/68 to 130/84 (P<0.0001).  A significant increase in cardiac output was also observed from 5.5 ± 0.4 to 6.7 ± 0.6 L/min (P=0.004) which was mediated through an increase in heart rate rather than an increase in stroke volume (P=0.98).  Exercise resulted in erroneous measures in 2 participants using the TT technique, and in 1 using the QA technique.

Using the TT technique: the MPA-RPA PWV changed from 2.43 ± 0.26 at rest to 2.86 ± 0.25 during exercise (mean (95% CI) of PWV differences = -0.42 (-1.2-0.4) ms-1, P=0.24); while the MPA-LPA PWV changed from 3.03 ± 0.27 at rest to 3.35 ± 0.39 during exercise (mean (SD) of PWV differences = -0.32 (-1.6-1) ms-1, P=0.57).

Using the QAInv technique: the MPA PWV went from 0.95 ± 0.09 at rest to 0.91 ± 0.1 during exercise (mean (95% CI) of PWV differences = 0.03 (-0.2-0.3) ms-1, P=0.77); the RPA PWV went from 1.03 ± 0.06 at rest to 0.92 ± 0.17 during exercise (mean (95% CI) of PWV differences = 0.11 (-0.3-0.5) ms-1, P=0.53); the LPA PWV went from 1.03 ± 0.16 at rest to 1.05 ± 0.11 during exercise (mean (95% CI) of PWV differences = -0.02 (-0.4-0.3) ms-1, P=0.9).