Repeatability of 18F-FDG uptake in metastatic bone lesions of breast cancer patients and implications for accrual to clinical trials

EJNMMI Research

Table 2 SUV Repeatability metrics for all n = 38 lesions

Parameter	SUVmax	SULpeak
SUVmin, SUVmax	2.0, 13.9	0.9, 6.4
\(\stackrel{-}{\varvec{d}}\) ± SD	0.20 ± 0.43	0.14 ± 0.20
d_i Min, Max	-0.96,0.88	-0.26, 0.77
*\(\stackrel{-}{\varDelta }\) [-CI, +CI] (%)	4.4 [0.4, 8.3]	6.4 [1.1, 11.7]
-RC [-CI, +CI] (%)	-14.7 [-17.9, -11.3]	-14.0 [-18.6, -9.1]
+RC [-CI, +CI] (%)	17.2 [12.7, 21.8]	16.3 [10.0, 22.9]
\({\varvec{w}\mathbf{C}\mathbf{V}}_{\varDelta }\) [-CI, +CI] (%)	5.9 [4.4, 8.3]	5.6 [3.5, 7.7]
-RC₀ [-CI, +CI] (%)	-16.3 [-19.7, -12.8]	-17.5 [-25.3, -8.8]
+RC₀ [-CI, +CI] (%)	19.4 [14.6, 24.5]	21.2 [9.7, 33.9]
\(\varvec{w}{\mathbf{C}\mathbf{V}}_{\varDelta 0}\) [-CI, +CI] (%)	6.6 [5.0, 8.2]	7.2 [3.4, 11.1]

^*The 95% confidence interval of the average difference of log-transformed data,\({\text{C}\text{I}}_{\stackrel{-}{\varDelta }}\), was found to not contain zero. This was hypothesized to be a sampling effect and RCs were recalculated without subtracting the sample mean, as described in the Supplementary materials. This approach led to a slightly more conservative (i.e. larger) estimates of the repeatability coefficients and coefficient of variation, denoted in Supplementary materials Table S2 as ± RC₀ and wCV_∆0.