Data Support TMB as Marker for Durvalumab Activity in NSCLC

Author(s):

Tumor mutational burden identified patients who obtained a survival benefit with the PD-L1 inhibitor durvalumab, as initial therapy versus chemotherapy for advanced non–small cell lung cancer, even though the primary analysis of the randomized trial showed no difference between treatment groups, according to a new analysis.

Solange Peters, MD, PhD

Tumor mutational burden (TMB) identified patients who obtained a survival benefit with the PD-L1 inhibitor durvalumab (Imfinzi), as initial therapy versus chemotherapy for advanced non—small cell lung cancer (NSCLC), even though the primary analysis of the randomized trial showed no difference between treatment groups, according to a new analysis.

Patients with tissue (t) TMB ≥10 mutations/megabase of DNA (mut/Mb) had a median overall survival (OS) of 18.6 months with durvalumab, 16.6 months with the combination of durvalumab and the anti—CTLA-4 antibody tremelimumab, and 11.9 months with chemotherapy. However, the difference between the durvalumab groups and chemotherapy did not achieve statistical significance. Patients with a TMB <10 mut/Mb fared better with chemotherapy alone.

TMB measured in blood (bTMB) had good correlation with tTMB, and subsequent analyses showed a consistent advantage for durvalumab-containing therapy across the bTMB range of ≥4 to ≥20 mut/Mb, as reported at the 2019 AACR Annual Meeting.

“TMB was predictive of an OS benefit with first-line durvalumab and durvalumab plus tremelimumab in metastatic non—small cell lung cancer,” said Solange Peters, MD, PhD, of University Hospital Center Vaudois and Lausanne University in Switzerland. “In tissue TMB greater than or equal to 10, overall survival was improved with durvalumab with or without tremelimumab versus chemotherapy; however, the small dataset limits interpretation.

“A blood TMB threshold greater than or equal to 20 was associated with improved overall survival with durvalumab versus chemotherapy. We observed markedly improved overall survival and greater clinical benefit with durvalumab plus tremelimumab versus chemotherapy, indicating the potential contribution of tremelimumab in this setting.”

Results of the exploratory analysis warrant prospective investigation of bTMB as a predictive biomarker for immunotherapy, she added.

The findings came from an analysis of data from the phase III, randomized MYSTIC trial involving patients with untreated stage IV NSCLC. Otherwise eligible patients were enrolled irrespective of PD-L1 expression status. Investigators randomized 1118 patients to receive durvalumab, durvalumab plus tremelimumab, or platinum-containing chemotherapy.

The primary analysis included patients with PD-L1 expression ≥25% in tumor cells. Primary endpoints were the comparison of OS with durvalumab versus chemotherapy, with durvalumab plus tremelimumab versus chemotherapy, and progression-free survival with the combination versus chemotherapy. Overall survival by bTMB and tTMB were exploratory endpoints.

The data showed a median OS of 16.3 months with durvalumab, 11.9 months with the combination, and 12.9 months with chemotherapy. Neither durvalumab-containing treatment arm significantly improved OS as compared with chemotherapy. OS in the intention-to-treat (ITT) population was 11 to 12 months in all three treatment groups.

TMB has demonstrated potential as a predictive biomarker for immunotherapy, independent of PD-L1 expression, Peters said. Across a variety of tumor types, higher tTMB has been associated with improved OS with immunotherapy. However, an OS benefit versus other standard-of-care treatment had yet to be shown.

Tumor response data from the CheckMate-558 trial identified a tTMB cutoff of ≥10 mut/Mb for the PD-1 inhibitor nivolumab (Opdivo). Use of that tTMB cutoff in CheckMate-227 showed significant improvement in PFS with nivolumab-treated patients.

Baseline characteristics and OS in the MYSTIC trial were similar between the ITT population and the subgroup of patients evaluable for tTMB (N = 460). Analysis of the tTMB cutoff of ≥10 mut/Mb showed hazard ratios of 0.70 for durvalumab and 0.72 for the combination versus chemotherapy, neither of which achieved statistical significance.

Peters noted that tTMB assessment is invasive, requires an adequate tissue sample, and has a significant turnaround time. On the other hand, bTMB determined from circulating tumor DNA is faster, less invasive, and possibly more representative of the heterogeneity of metastatic lesions.

Investigators evaluated bTMB for 809 MYSTIC participants by means of a commercially available 500-gene panel. In the subset of patients with matched blood and tissue samples, bTMB showed good correlation with tTMB. Comparison with the ITT population showed similar baseline characteristics and OS for the bTMB-evaluable subgroup.

A preliminary analysis of bTMB with a cutoff of ≥16 mut/Mb (N = 319) yielded a median OS of 16.5 months for the combination arm, 11.0 months for durvalumab alone, and 10.5 months with chemotherapy. The combination was associated with a statistically significant 38% reduction in the survival hazard versus chemotherapy, but single-agent durvalumab did not distinguish itself from chemotherapy. The 24-month OS was 39.1% with the combination, 29.6% with durvalumab alone, and 18.2% with chemotherapy.

Evaluation of a range of bTMB cutoffs led to the selection of ≥20 mut/Mb for further analysis because of the observed effect size achieved with the combination and the size of the patient population that benefited, said Peters.

Patients with bTMB in blood ≥20 mut/Mb (N = 211) had a median OS of 21.9 months with the combination, 12.8 months with single-agent durvalumab, and 10.0 months with chemotherapy. The combination reduced the survival hazard by 51% versus chemotherapy (HR, 0.49; 95% CI, 0.32-0.74). Durvalumab monotherapy reduced the survival hazard by 28%, which did not achieve statistical significance (HR, 0.72; 95% CI, 0.50-1.05). The combination arm had a 24-month OS of 48.1% versus 33.8% with durvalumab alone, and 19.4% with chemotherapy.

Patients with bTMB <20 mut/Mb (N = 598) had similar median and 24-month OS, regardless of assigned treatment.

Analysis of PFS produced a hazard ratio of 0.53 for the combination versus chemotherapy (95% CI, 0.34-0.81) and 0.77 for single-agent durvalumab versus chemotherapy (95% CI, 0.52-1.13). The 12-month PFS values were 38.6% with the combination, 30.3% with durvalumab alone, and 2.3% with chemotherapy. Among patients with bTMB <20 mut/Mb, median PFS did not differ by treatment group.

Investigators also compared the relationship between PD-L1 expression and bTMB ≥20 (N = 799). Peters said 488 (44%) patients had PD-L1 expression ≥25%, and 211 (19%) had bTMB ≥20. Only 100 patients (9%) had both PD-L1 expression ≥25% and bTMB v20.

“The values did not correlate,” she said. “PD-L1 and blood TMB are independent biomarkers.”

Peters S, Cho BC, Reinmuth N, et al. Tumor mutational burden (TMB) as a biomarker of survival in metastatic non-small cell lung cancer (mNSCLC): Blood and tissue TMB analysis from MYSTIC, a Phase III study of first-line durvalumab ± tremelimumab vs chemotherapy. Presented at: 2019 AACR Annual Meeting; March 29-April 3, 2019; Atlanta, GA. Abstract CT074.

<<< 2019 AACR Annual Meeting