Table 1 Analysis of tumour growth in a generalised linear-mixed model

From: Ewing sarcoma dissemination and response to T-cell therapy in mice assessed by whole-body magnetic resonance imaging

Parameter

Estimated volume (mm3) (95% confidence limits)

P -value

Group A vs B

−7.95 (−17.33 to 1.43)

P=0.0965

Time (per day)

+1.44 (1.04 to 1.85)

P<0.0001

Localisation (compared with kidney tumours)

 

P<0.0001

 Lung

−2.03 (−15.76 to 11.69)

P=0.7707

 Femur/tibia

+11.06 (−1.84 to 23.95)

P=0.0925

 Pelvis/Vertebral column

+80.67 (61.52 to 99.83)

P<0.0001

 Suprarenal gland

−5.71 (−28.35 to 16.94)

P=0.6203

 Soft tissue

−7.32 (−20.94 to 6.29)

P=0.2905

  1. Shown are estimates for tumour volume (mm3) with 95% confidence limits and P-values for the three parameters with influence on tumour volume: group, time, localisation. Owing to varying time points of follow-up scans, resulting in missing values and nested data structure, generalised linear-mixed models were fitted. The explanatory variables, such as group, time, and localisation, were modelled as fixed effects. The parameter identifying each mouse (compound symmetry covariance structure), time in days after tumour inoculation (spatial covariance structure) were modelled as random effects. Lesions were modelled as subject effect. The chosen underlying distributions were Gaussian, with identity as link function.
Back to article page