Page 105 - 89Zr-Immuno-PET:Towards a Clinical Tool to Guide Antibody-based Therapy in Cancer

Noise-induced variability of 89Zr-immuno-PET
Repeatability coefficients (RC) in percent were calculated according to equation 1:
, [1]
where ∆ is (AC1-AC2) per VOI, MV is the mean value 0.5*(AC1+AC2) per VOI, 100% is multiplication by 100, SD the standard deviation over all percentage differences for n VOI in the group, and CVw the coefficient of variation within scan.
RC were expressed as a percentage instead of absolute value, as the difference
between AC1 and AC2 scaled linearly with the mean value of AC1 and AC2. Furthermore, the mean percentage difference was calculated as well by taking the
average ∆ over all VOIs in the group. The mean percentage difference and the RC
combined define the limits of agreement (LoA). As a result the LoA were directly 5 related to the coefficient of variation. Use of a relative unit allows for comparison
with other studies, irrespective of the measurement unit used (e.g. type of normalization used to calculate SUV).
Reliability analysis
To estimate the contribution of noise induced variability to the observed differences between patients or tumor lesions, the ICC was calculated in addition to the RC. The ICC was calculated as the proportion of the total variance that is due to the true variance. True variance reflects biological differences, for example between patients or between tumor lesions, while the total variance comprises both true variance as well as the measurement variance (equation 2).
, [2]
where σvoi2 is the variance between the n VOI per group, σ∆2 is the variance over the differences between AC1 and AC2. For a reliable measure, a high ICC is expected, as an ICC of 1 reflects that all measured variance can be attributed to biological differences (the contribution of measurement variability is negligible). An ICC of 0 signifies that all measured variance can be attributed to measurement variability (no detection of biological differences beyond measurement variability). In general, an ICC of > 0.7 is considered acceptable (14). The 95% confidence interval of the ICC was obtained to estimate the precision of the ICC. ICCs and 95%
103