Analysis of Time-to-treatment Discontinuation of Targeted Therapy, Immunotherapy, and Chemotherapy in Clinical Trials of Patients With Non-Small-Cell Lung Cancer

G. M. Blumenthal; Y. Gong; K. Kehl; P. Mishra-Kalyani; K. B. Goldberg; S. Khozin; P. G. Kluetz; G. R. Oxnard; R. Pazdur


Ann Oncol. 2019;30(5):830-838. 

In This Article


We first reviewed results for two trials of EGFR TKI versus doublet chemotherapy, where one control arm included maintenance chemotherapy and the other did not. When maintenance chemotherapy was given (Figure 2A), the PFS and TTD curves exhibit the expected steady decline over time. However, without maintenance chemotherapy (Figure 2B), PFS drops dramatically while TTD drops in an even more precipitous fashion, resulting in a clear discordance between PFS and TTD in the control arm. This confirmed our concern over evaluating TTD when there is a planned termination of therapy after a defined number of cycles, and our omission of such trial arms from the remainder of the analysis (Figure 1). Reviewing results from two other trials comparing two TKI therapies (Figure 2C) or comparing immune checkpoint inhibition with single agent chemotherapy (Figure 2D), these regimens maintain a visually consistent trajectory for the PFS and TTD curves, supporting the inclusion of such regimens in the remainder of the TTD analysis.

Figure 2.

Selected K-M analysis of TTD and PFS. (A) EGFR TKI versus doublet chemo with maintenance in EGFR mutant population. (B) EGFR TKI versus doublet chemo (no maintenance) in EGFR mutant population. (C) EGFR TKI versus EGFR TKI in EGFR mutant population. (D) Immune checkpoint inhibitor versus monotherapy chemotherapy.

In all, 8947 patients met the criteria for the pooled analysis (Figure 1). Table 2 provides a demographic summary of the patients included in the pooled analysis. The overall patient-level correlation between TTD and PFS was 0.87 (Table 3), ranging from 0.78 to 0.98 across the six therapeutic subgroups (Table 3). Correlation was lowest for chemo monotherapy (0.78, 95% CI 0.77–0.79). The patient-level correlation between TTD and OS was 0.68, ranging from 0.55 to 0.79 across the six therapeutic subgroups (Table 3). This correlation was highest for ALK TKIs where the TTD/OS correlation (0.79, 95% CI 0.76–0.81) was stronger than the PFS/OS correlation (0.65, 95% CI 0.61–0.68). Similarly, for ICI therapy the observed TTD/OS correlation (0.72, 95% CI 0.70–0.74) appeared stronger than the PFS/OS correlation (0.63, 95% CI 0.61–0.66). The results did not substantially change when the censoring window from the data cut-off date was moved forward 14 and 21 days, respectively (supplementary Table S1, available at Annals of Oncology online). In addition, the results did not substantially change when increased exposure time was accounted for intravenously administered drugs (supplementary Table S2, available at Annals of Oncology online) or when excluding patients who were censored on both endpoints (supplemental Table S3, available at Annals of Oncology online).

For the oncogene-directed targeted therapy subgroups (EGFRm and ALK), the median TTD exceeded the median PFS: 13.4 months (95% CI 12.4, 14.0) versus 11.4 months (95% CI 11.1, 12.8) for EGFRm and 14.1 months (95% CI 12.3, 15.4) versus 11.3 months (95% CI 10.9, 12.9) for ALK. In contrast, median TTD was shorter than median PFS in the ICI, chemo doublet, and chemo mono subgroups: 3.5 months (95% CI 3.4, 3.6) versus 4.2 months (95% CI 4.0, 4.3) for ICI; 3.8 months (95% CI 3.7, 4.0) versus 5.6 months (95% CI 5.5, 5.7) for chemo doublet; and 2.2 months (95% CI 2.1, 2.3) versus 4.1 months (95% CI 3.9, 4.2) for chemo monotherapy. Figure 3 and supplementary Figures S1 and S2, available at Annals of Oncology online show the TTD-PFS, TTD-OS, and PFS-OS scatter plots by sub-group.

Figure 3.

Patient-level TTD-PFS scatter plots by subgroup.

Supplemental Figure 1.

TTD- OS scatter plots

Supplemental Figure 2.

PFS- OS scatter plots

The oncogene-targeted therapy subgroups had high rates of late TTD (Figure 4)—12.4% of patients for the EGFRm subgroup and 22.9% of patients in the ALK subgroup—indicating that these patients often continued treatment beyond RECIST progression. Early TTD was rare for the EGFRm and ALK subgroups (2.0% and 1.7%). The ICI subgroup had a mixture of both early (5.9%) and late (8.5%) TTD. The chemo doublet and chemo mono subgroups had subsets of patients with early TTD (7.7% and 15% respectively) but little late TTD (0.8% and 0.5% respectively), indicating patients stopped therapy before progression but rarely continued post-progression.

Figure 4.

TTD-PFS histogram.