Table 4.

Multivariate analysis of PET variables to PFS and OS (p-values, HR).

† PET Parameter	PFS			OS
	Days	SE	p-Value	HR	SE	p-Value
Lesion-Level
SUVmaxavg1	26.5	13.1	0.043	1.135	0.856	0.875
ΔSUVmaxavg	−2.1	21.5	0.923	0.800	1.081	0.853
SUVpeakavg1	32.0	13.5	0.018	1.421	2.770	0.879
ΔSUVpeakavg	−10.4	21.9	0.635	1.142	3.486	0.968
Index SUVmax1	17.3	14.2	0.222	1.296	2.552	0.908
Index ΔSUVmax	0.7	18.3	0.971	1.196	4.299	0.964
Index SUVpeak1	21.5	15.4	0.163	1.443	3.250	0.892
Index ΔSUVpeak	−2.4	17.4	0.888	0.874	0.874	0.885
Patient-Level
qSUVmax1	17.6	19.2	0.359	1.321	2.759	0.908
ΔqSUVmax	0.9	27.2	0.972	1.341	4.186	0.935
qSUVpeak1	36.9	18.3	0.044	1.646	3.191	0.840
ΔqSUVpeak	−15.4	18.8	0.413	1.003	1.609	0.999
qSUVtotal1	0.7	20.9	0.972	2.911	6.061	0.753
ΔqSUVtotal	14.1	25.4	0.580	0.635	1.396	0.794
qVF1	−11.2	21.6	0.606	1.977	4.017	0.808
ΔqVF	15.7	21.2	0.458	0.708	1.063	0.783

^† The multivariate model used age, ln(BAP) and the PET parameter. For association with PFS multiple linear regression was used as the data were not censored. PFS days are the number of days corresponding to a 1-SD increase in the PET parameter, and SE is the standard error of Days. Cox proportional hazard modeling was used to determine association of the multivariate model to OS, where 4 patients were censored. The hazard ratio (HR) is the associated hazard ratio corresponding to a 1-SD increase in the PET parameter. The lesion-level analyses were performed on 17 patients at baseline (indicated by 1 after the parameter), while change on-dasatinib was determined on 14 of the 17 patients. The patient-level whole-body QTBI analyses were performed on 16 patients at baseline, while change was determined on 12 of the 16 patients. Boldface type indicates a significant (p ≤ 0.05) association with outcome.