Background: Preclinical studies have identified molecular correlates of sensitivity to ATR inhibition. This translational study was designed to test the ATR inhibitor berzosertib in patients with advanced solid tumors carrying alterations in ATRX, ATM, genes conferring replication stress (RS), or SDH.
Methods: Patients were recruited to 4 cohorts: T1: ATRX-mutant leiomyosarcoma; T2: ATM-mutant solid tumors; T3: solid tumors with mutations in RS-associated genes; and T4: SDH-deficient GIST. Patients were treated with berzosertib 240 mg/m2 IV twice per week. Pre and on-treatment biopsies were obtained in cohorts T1-T3.
Results: Patients with SDH-mutant GIST had the longest median progression-free survival (PFS) (229 days) with stable disease as the best response. Patients in the other cohorts experienced progressive disease within 4 months. There was no significant difference in PFS comparing outcomes in patients with/without mutations in ATM or RS genes. Decreased pS345-CHK1 in on-treatment biopsies indicated target engagement by berzosertib and were accompanied by substantial increases in levels of DNA damage (g-H2AX) and RS (pKAP1) markers in a subset of patients. However, these biomarker changes did not translate to clinical benefit. In contrast, in cohorts T1-T3, increased expression of SFLN11 on treatment correlated with clinical benefit (HR = 0.045; 95%CI 0.005-0.400).
Conclusions: Across cohorts, only SDH-mutant GIST patients experienced prolonged disease control. Despite evidence of target engagement, patients enrolled to all other cohorts had short PFS, suggesting rapid adaptation to ATR inhibitor monotherapy. Among these patients, those with tumors expressing SLFN11 during berzosertib exposure derived the most clinical benefit.