Risk of Chemotherapy Toxicity in Older Patients With Blood Cancer or Non-small Cell Lung Cancer

Will fit a Cox regression model for time to toxicity (grades 3-5) containing the baseline chemo-toxicity risk score as the only predictor. The overall ability of the model to distinguish grade 3-5 from grade 1-2 toxicity will be evaluated using Harrell's C-statistic, which can be viewed as a generalization of the area under the curve measurement for receiver operator characteristics (ROC's) curves based on binary outcome data. A 95% confidence interval will be constructed for the C-statistic and if the lower bound is greater than 0.5 (expected value if the risk score is not predictive of toxicity) we will conclude that risk score is significantly able to distinguish between toxicity grades.

Will fit a Cox regression model for time to toxicity (grades 3-5) containing the baseline chemo-toxicity risk score as the only predictor. The overall ability of the model to distinguish grade 3-5 from grade 1-2 toxicity will be evaluated using Harrell's C-statistic, which can be viewed as a generalization of the area under the curve measurement for ROC's curves based on binary outcome data. A 95% confidence interval will be constructed for the C-statistic and if the lower bound is greater than 0.5 (expected value if the risk score is not predictive of toxicity) we will conclude that risk score is significantly able to distinguish between toxicity grades.

Will explore the relationship of RDI and MAX2 (reduced and prescribed) with clinical and biologic factors of frailty. RDI (>= 85% versus [vs.] < 85%) will be treated as a continuous and dichotomous variable. The distribution of RDI and MAX2 will be examined graphically and transformed to normality as appropriate. Linear regression for continuous RDI and logistic regression for dichotomized RDI will be used to understand the relationship with RDI and independent variables (e.g. GA scores, age, short physical performance battery [SPPB] etc.) Stepwise regression will be used to identify significant clinical and biologic factors (e.g. OSU_Senescence) that are independently associated with continuous or dichotomized RDI.

Linear mixed models will be used in the analyses. Each model will include effects of each prognostic variable of interest, time (baseline vs. end of treatment), and time-by-prognostic variable interactions to determine if the associations between the prognostic variables and the outcome differs at baseline and end of treatment. An unstructured covariance matrix will be used to model the relationship between outcome measurements from the same patient.

Linear mixed models will be used in the analyses. Each model will include effects of each prognostic variable of interest, time (baseline vs. end of treatment), and time-by-prognostic variable interactions to determine if the associations between the prognostic variables and the outcome differs at baseline and end of treatment. An unstructured covariance matrix will be used to model the relationship between outcome measurements from the same patient.

Will determine the association of molecular markers of aging with risk of chemotherapy toxicity using the Chemo-Toxicity calculator and other prognostic factors (e.g. age, disease, stage, body composition by imaging). Linear mixed models will be used in the analyses. Each model will include effects of each prognostic variable of interest, time (baseline vs. end of treatment), and time-by-prognostic variable interactions to determine if the associations between the prognostic variables and the outcome differs at baseline and end of treatment. An unstructured covariance matrix will be used to model the relationship between outcome measurements from the same patient.