Results from scenario analyses (LYs gained)
| Scenario number and description . | Rationale . | Per patient . | US population . | |
|---|---|---|---|---|
| Base case | 3.20 | 18 875 | ||
| 1 | Probability of infusion not affected by V2VT | In this scenario, V2VT only impacts postinfusion survival (ie, not the proportion of patients that receive an infusion). | 1.98 | 11 706 |
| 2 | Postinfusion survival not affected by V2VT (Bachy et al9) | In this scenario, postinfusion survival is informed by Bachy et al9 which does not differentiate survival by V2VT. | 0.82 | 4826 |
| 3 | Switch non-infused survival source (Kuhnl et al8) | As above, except using an alternative source for postinfusion survival: Kuhnl et al9 | 3.19 | 18 832 |
| 4 | Switch HR cutoffs (<28 d vs 28-40 d vs ≥40 d) | In the base-case analysis, HR cutoffs of <36 and ≥36 d were used, as a simple means to dichotomize the Locke et al6 cohort in terms of their survival experience linked to V2VT. In this scenario, alternative cutoffs are used, which breaks the cohort into 3 groups instead of 2. | 3.47 | 20 500 |
| 5 | Change long V2VT to be 37 d | Alternative long V2VT specified to reflect a smaller reduction for the short V2VT group. | 2.46 | 14 526 |
| 6 | Change short V2VT to be 30 d | Alternative short V2VT specified to reflect a smaller reduction from the long V2VT group. | 2.82 | 16 661 |
| 7 | Assume half of the US population | Sensitivity of the population results stress-tested by assuming half of the estimated eligible cohort. | 3.20 | 9438 |
| 8 | Assume CIBMT registry population of 1294 patients | Sensitivity of the population results stress-tested by assuming same population per latest data from CIBMT registry. | 3.20 | 4138 |
| 9 | Postinfusion survival model: lognormal | 1.82 | 10 761 | |
| 10 | 1 knot(s) normal spline | 2.34 | 13 801 | |
| 11 | MCM: Weibull | Choice of an alternative survival extrapolation for patients that receive CAR T. | 3.53 | 20 813 |
| 12 | MCM: log-logistic | 3.29 | 19 435 | |
| 13 | Non-infused survival model | Choice of an alternative survival extrapolation for patients that do not receive CAR T. | 3.20 | 18 861 |
| 14 | Log-logistic | 3.20 | 18 865 | |
| 15 | 1 knot(s) odds spline | 3.06 | 18 042 | |
| 16 | MCM: lognormal MCM: log-logistic | 3.06 | 18 067 | |
| 17 | V2VT regression model: | Choice of an alternative regression model for estimating the proportion of patients who were infused, based on V2VT. | 3.14 | 18 529 |
| 18 | weighted-linear | 3.07 | 18 102 | |
| 19 | logistic | 2.68 | 15 802 | |
| 20 | method of moments Expectation maximization algorithm | 2.44 | 14 420 | |
| 21 | Iterative V2VT sampling | In the base-case analysis, all patients were assumed to have the same V2VT. In this scenario, V2VT is sampled from a distribution, with the mean results taken. Further details of this approach are provided in a supplemental Appendix. | 2.79 | 16 475 |
| Scenario number and description . | Rationale . | Per patient . | US population . | |
|---|---|---|---|---|
| Base case | 3.20 | 18 875 | ||
| 1 | Probability of infusion not affected by V2VT | In this scenario, V2VT only impacts postinfusion survival (ie, not the proportion of patients that receive an infusion). | 1.98 | 11 706 |
| 2 | Postinfusion survival not affected by V2VT (Bachy et al9) | In this scenario, postinfusion survival is informed by Bachy et al9 which does not differentiate survival by V2VT. | 0.82 | 4826 |
| 3 | Switch non-infused survival source (Kuhnl et al8) | As above, except using an alternative source for postinfusion survival: Kuhnl et al9 | 3.19 | 18 832 |
| 4 | Switch HR cutoffs (<28 d vs 28-40 d vs ≥40 d) | In the base-case analysis, HR cutoffs of <36 and ≥36 d were used, as a simple means to dichotomize the Locke et al6 cohort in terms of their survival experience linked to V2VT. In this scenario, alternative cutoffs are used, which breaks the cohort into 3 groups instead of 2. | 3.47 | 20 500 |
| 5 | Change long V2VT to be 37 d | Alternative long V2VT specified to reflect a smaller reduction for the short V2VT group. | 2.46 | 14 526 |
| 6 | Change short V2VT to be 30 d | Alternative short V2VT specified to reflect a smaller reduction from the long V2VT group. | 2.82 | 16 661 |
| 7 | Assume half of the US population | Sensitivity of the population results stress-tested by assuming half of the estimated eligible cohort. | 3.20 | 9438 |
| 8 | Assume CIBMT registry population of 1294 patients | Sensitivity of the population results stress-tested by assuming same population per latest data from CIBMT registry. | 3.20 | 4138 |
| 9 | Postinfusion survival model: lognormal | 1.82 | 10 761 | |
| 10 | 1 knot(s) normal spline | 2.34 | 13 801 | |
| 11 | MCM: Weibull | Choice of an alternative survival extrapolation for patients that receive CAR T. | 3.53 | 20 813 |
| 12 | MCM: log-logistic | 3.29 | 19 435 | |
| 13 | Non-infused survival model | Choice of an alternative survival extrapolation for patients that do not receive CAR T. | 3.20 | 18 861 |
| 14 | Log-logistic | 3.20 | 18 865 | |
| 15 | 1 knot(s) odds spline | 3.06 | 18 042 | |
| 16 | MCM: lognormal MCM: log-logistic | 3.06 | 18 067 | |
| 17 | V2VT regression model: | Choice of an alternative regression model for estimating the proportion of patients who were infused, based on V2VT. | 3.14 | 18 529 |
| 18 | weighted-linear | 3.07 | 18 102 | |
| 19 | logistic | 2.68 | 15 802 | |
| 20 | method of moments Expectation maximization algorithm | 2.44 | 14 420 | |
| 21 | Iterative V2VT sampling | In the base-case analysis, all patients were assumed to have the same V2VT. In this scenario, V2VT is sampled from a distribution, with the mean results taken. Further details of this approach are provided in a supplemental Appendix. | 2.79 | 16 475 |
CIBMT, Center for International Blood and Marrow Transplant Research; HR, hazard ratio.