Introduction

Cumulative telomeric attrition with each cellular division leads to replicative senescence1. To bypass replicative senescence and achieve replicative immortality, cancer cells preferentially exploit one of two known telomere maintenance mechanisms to reconstitute telomere length—either (i) the de-silencing of hTERT telomerase gene expression, occurring in the majority of human cancers, or (ii) through ALT2. While most prevalent in mesenchymal tumors, ALT has been described in up to 18% of melanomas3.

ALT+ cells synthesize telomeric DNA through a homology-directed, recombination-dependent replication pathway similar to break-induced replication, generating chromosomal instability and replication stress2. Furthermore, ALT is highly associated with the loss-of-function of related chromatin remodelers (such as ATRX and DAXX), which leads to replication stress through the promotion of aberrant HR at repetitive elements, including telomeres4. Though clinical experience is limited, preclinical studies have suggested that the high replication stress state arising through ALT is therapeutically exploitable by targeting ATR, a key mediator of cellular response to replication stress, in a synthetic lethal approach5,6. We now present our experience treating a patient with ALT+ metastatic melanoma with the novel, potent, and selective ATR inhibitor, camonsertib7.

Results

Patient treatment history

The patient, a 63-year-old Caucasian male, was diagnosed in July 2019 with metastatic cutaneous melanoma and treated with nivolumab and ipilimumab with initial radiological partial response (PR) after 3 months, which was confirmed on subsequent nivolumab maintenance. Ipilimumab was resumed following suspected pseudoprogression in March 2020 and continued until confirmed disease progression in June 2020 (Fig. 1A). The patient then received additional off-label targeted treatments with the poly(ADP-ribose) polymerase (PARP) inhibitor, olaparib, initially in combination with nivolumab, before switching to olaparib plus trametinib. However, no anti-tumor benefit was observed with both combination regimens and disease progression was noted in April 2021, prior to enrollment on the current study (Fig. 1A).

Fig. 1: Prior treatment, clinical genomics, and response to camonsertib.
figure 1

A Treatment history from diagnosis through TRESR trial. B TVR counts derived from Telomere Hunter analysis of WGS data in the tumor compared to blood normal sample collected from this patient. C A representative confocal microscopy image of the ssTeloC foci (green) detected in the retroperitoneal metastatic biopsy (scale bar = 20 µm). The pie chart shows the proportion of cells with various numbers of the foci from the total number of cells analyzed (N = 44,616). Scans of the peritoneal mass at (D) baseline (3.3 cm × 3.2 cm), and 11 months on treatment (0.0 cm × 0.0 cm). E RECIST v1.1 measurements at baseline and on-treatment. Green box indicates −30% change from baseline. F Longitudinal ctDNA monitoring was performed using tumor uninformed variant-based approach using Tempus xF and methylation-based circulating tumor fraction estimation using Guardant Infinity™, and a tumor-informed dPCR based approach from Tracer Biotechnologies. Approximate LOQs for each assay type are marked in the dotted lines.

Full size image

Tumor genomic findings

Contemporaneous genomic analyses of liver biopsy and paired blood specimens collected at the time of disease progression after front-line nivolumab plus ipilimumab revealed pathogenic somatic alterations in TP53 (p.S127F, p.R65fs), NF1 (p.K1263*), BRCA2 (p.R2482C), ATRX (p.R418*), high tumor mutational burden, microsatellite stable status, and a germline alteration in CHEK2 (p.T367fs) (Table 1). Furthermore, retrospective genomic analysis of a retroperitoneal metastatic biopsy taken after disease progression on the combinations of olaparib plus nivolumab and olaparib plus trametinib revealed the previously detected alterations, as well as an additional pathogenic germline NF1 alteration, and provided further molecular information on allele zygosity and mutational signatures (Table 1, Fig. S1). Importantly, these analyses demonstrated that the BRCA2 and gCHEK2 mutations were monoallelic, which was consistent with an HR proficient (HRP) genomic signature, as well as the lack of anti-tumor clinical benefit from olaparib. Compound germline and somatic NF1 pathogenic mutations point to a potentially early initiating event for this patient’s tumor.

Table 1 Characterization of genomic structure and signatures
Full size table

Assessment of telomere maintenance mechanisms

Whole genome sequencing (WGS) confirmed biallelic loss of ATRX, which was consistent with the presence of an ALT+ genomic signature (Table 1), driven by markedly higher telomere variant repeats (TVR) counts compared with normal tissue (Table 1, Fig. 1B). To confirm the ALT+ status in tumor cells, Native Fluorescence In Situ Hybridization (FISH) was performed on a tissue section from the retroperitoneum metastasis (same specimen as WGS). Single-strand telomeric C-rich (ssTeloC) foci were observed in 28.3% (12,648/44,616) of tumor cells (Fig. 1C), providing functional confirmation for ALT activity in the patient’s tumor. In total, we provide 3 independent lines of evidence supporting ALT+ status from the same tumor tissue specimen.

Safety, dosing, and adverse events

The patient was enrolled on the TRESR trial based on his previously identified BRCA2 alteration and received camonsertib monotherapy at a dose of 120 mg given on Days 1–3 of each week of a 21-day cycle. The patient tolerated treatment well and continued at the same dose for 13 cycles (9 months), after which an intermittent weekly schedule (2 weeks on/1-week off) was adopted due to the development of grade 3 anemia requiring blood transfusion. Anemia was the only reported toxicity considered related to study drug treatment. Following the change in dosing schedule, the patient’s hemoglobin levels remained stable for the rest of the duration of therapy (a further 13 months).

Radiological and molecular response

The patient achieved a PR (RECIST v1.1; sum of target lesions -36% from baseline) after 3 months of treatment which was confirmed and sustained with further deepening of the target lesion response (sum of target lesions -74% from baseline) after 11 months of camonsertib (Fig. 1D, E). Circulating tumor DNA (ctDNA) clearance was observed by three weeks after camonsertib initiation and remained low throughout treatment monitored using variant-, methylation- and personalized digital polymerase chain reaction (dPCR)-based assays (Fig. 1F). The cancer began to progress at 19 months (+105% from nadir, -47% from baseline), but due to ongoing clinical benefit, treatment was continued for a total duration of 22 months.

Discussion

We describe the first clinical report of synthetic lethal targeting of ALT+ melanoma with an ATR inhibitor. This patient’s tumor was characterized by the biallelic loss of ATRX, markedly elevated TVR counts and ssTeloC foci, consistent with an ALT+ phenotype. Upon treatment with camonsertib, dosed at 120 mg weekly on Days 1–3 of each week, the patient experienced marked and sustained RECIST PR and ctDNA molecular response, and had a total of 22 months of treatment with a safe and well-tolerated oral targeted therapy.

Camonsertib was well-tolerated with anemia being the only treatment-related toxicity observed, which is a known class effect related to orally available ATR inhibitor. ALT is an emerging predictive biomarker of response for ATR inhibitors, but to the best of our knowledge, has never been formally assessed or demonstrated in the clinic5,6. Indeed, in ALT+ cells, telomeres are prone to replication stress caused by a variety of physical impediments to replication such as RNA-DNA hybrids, G-quadruplexes, or repetitive DNA sequence8. ATRX loss, which is highly associated with the ALT phenotype, further exacerbates replication stress at telomeres through defective nucleosome assembly9. In ALT+ cells, DNA double-stranded breaks at telomeres arising from cleaved replication forks encourage inter/intra-telomeric HR and break-induced telomere synthesis leading to marked cell-to-cell telomere length heterogeneity and chromosomal instability5. ATRX loss has also been described to lead to persistent association of replication protein A (RPA) with telomeres after DNA replication, creating a recombinogenic RPA-single-stranded DNA structure, which serves not only as an HR intermediate but a recruiter of ATR, which itself regulates HR5. Treatment of ALT+ cell lines with an ATR inhibitor was demonstrated to trigger chromosomal fragmentation and apoptosis5, providing preclinical rationale for the targeting of ATR in ALT+ tumors.

While this patient’s tumor harbored other genomic alterations of interest (namely, somatic BRCA2 and germline CHEK2 mutations), which have also been described as predictive biomarkers for ATR inhibitor sensitivity, both alterations were monoallelic and associated with an HRP genomic signature, consistent with the patient’s lack of clinical benefit from PARP inhibitor-based therapy given as a prior line of treatment. The HRP genomic signature detected in the patient’s tumor is furthermore consistent with its ALT dependence, as ALT+ cells actively rely on intact HR to sustain replicative immortality. Furthermore, the identification of compound heterozygous NF1 loss of function is notable, but not predicted to predispose to ATR inhibitor sensitivity, and more likely represents an early event in tumorigenesis.

Several molecular markers may be used to provide evidence of ALT activity in cancers, each with their own inherent advantages and disadvantages. These markers include Native FISH10,11, Telo-FISH4, colocalization of telomeres with promyelocytic leukemia protein (PML) bodies9, and extrachromosomal telomeric repeats including circles of partially double-stranded C-rich telomeric DNA (C-circles) assay12. A C-circle assay is a highly specific and sensitive method to detect ALT activity and may be amenable to testing in plasma samples, although this has not yet been clinically validated13. However, C-circle detection requires multiple DNA extraction, enzymatic, and detection steps that may not be convenient for a routine histopathological laboratory14. Detection of ‘ultrabright’ telomeric foci by Telo-FISH and telomere localization in PML bodies have been used to detect ALT in archival FFPE samples15, although, especially in the case of ALT-associated PML bodies, routine use is challenged by technical limitations and potential for false positives or negatives10. In this study, whole-genome sequencing analysis using a published bioinformatic classifier3,16 was combined with a Native FISH assay, providing orthogonal lines of evidence for ALT activity. The Native FISH assay is a convenient readout for ALT detection that is compatible with routine FISH-based assays in clinical histopathological laboratories10. This assay was shown to reliably detect ALT+ tumor models when a threshold was applied of as few as 10% of cells showing ≥1 nuclear focus, highlighting its specificity and sensitivity. We believe that further diagnostic development of Native FISH, including a robust validation of the relevant cut point, could enable the prospective selection of patients harboring ALT+ tumors. Computational sequence analysis approaches are also showing promise for the stratification of telomere maintenance mechanisms3,16,17. The classifier3 adapted in this study has been shown to robustly detect ALT+ tumors with a C-circle assay, based on the quantification of relative telomere content and the presence of specific TVRs (e.g., TTTGGG, TTCGGG, TTGGGG, and GTAGGG). However, the requirement of whole-genome sequencing data may be yet limiting for routine prospective diagnostic use. We envision that a future diagnostic assay for ALT status may combine multiple orthogonal readouts, allowing for the highest specificity and sensitivity.

Another important area of future investigation will be to understand the dynamics of ALT activity over the course of cancer development and in response to therapy. In this study, genomic analysis of ctDNA at progression did not identify any emergent resistance mechanisms, but was limited by low circulating tumor fraction. While little is currently known about acquired resistance mechanisms to ALT+ targeted therapies, one could speculate that reactivation of telomerase-mediated telomere elongation through hTERT could render ALT+ tumor cells resistant to ATR inhibitors.

In conclusion, treatment with the ATR inhibitor, camonsertib, led to impressive and durable radiological and ctDNA molecular responses in a patient with ATRX mutant/ALT+ heavily pre-treated metastatic melanoma. While the clinical experience of ATR inhibitor therapy in patients with ALT+ tumors remains limited and broad conclusions about the efficacy of ATR inhibitor therapy in ALT+ melanoma cannot be made from this single-patient case report, our report provides the first clinical example of the potential utility of ALT+ as a predictive biomarker of response. Further investigation into ALT+ as a patient selection biomarker is warranted in future ATR inhibitor clinical trials.

Methods

TRESR study of camonsertib in advanced solid tumors

Camonsertib is a highly potent (50% inhibitory concentration, 0.33 nM in cell-based assays), orally bioavailable, and highly selective ATR inhibitor, which is undergoing clinical investigation in several biomarker-selected early phase clinical trials as monotherapy or in combination with other targeted- or chemo-therapies18. Clinical responses to camonsertib have been reported in patients harboring biallelic loss-of-function mutations in genes including ATM, BRCA1/2, RAD51C, SETD2, and CDK127. While key toxicities of clinical grade ATR inhibitors have included a substantial risk of hematological side effects, namely anemia and neutropenia19, the high selectivity of camonsertib for ATR18 and the adoption of intermittent drug dosing schedules have led to camonsertib being associated with acceptable tolerability on recent trials7,20. Doses of camonsertib greater than 100 mg/day were projected as biologically active and determined to be clinically efficacious7. A dose of 160 mg on Days 1–3 for the first 2 weeks of a 3-week cycle was selected as the optimized monotherapy dosing regimen based on a combination of safety, pharmacokinetics, pharmacodynamics, and efficacy data20.

The patient reported herein was enrolled in a monotherapy arm of TRESR (NCT04497116), a first-in-human, Phase 1/2a, nonrandomized study of camonsertib in patients with advanced solid tumors harboring genomic biomarkers predicted to sensitize tumors to ATR inhibitors, including somatic or germline pathogenic alterations in selected genes (ATM, ATRIP, BRCA1/2, CDK12, CHTF8, FZR1, MRE11, NBN, PALB2, RAD51B/C/D, RNASEH2A/B, RAD17, REV3L, RAD50, SETD2). The patient provided written informed consent which allowed for in-depth genomic investigation of tumor and blood specimens, and consented to the publication and presentation of study results. The protocol was approved by the Institutional Review Board at The University of Texas, MD Anderson Cancer Center (Federalwide Assurance Documentation number FWA00000363, trial number 2020-0284) and complied with the Declaration of Helsinki and Good Clinical Practice Guidelines.

Tumor genomic analyses

Available next-generation sequencing (NGS) reports were reviewed at the time of patient enrollment to the TRESR trial for evidence of a genomic alteration to fulfill study eligibility criteria (Tempus xT, Tempus) (Table 1). Subsequently, longitudinally collected, archival, formalin-fixed, paraffin-embedded (FFPE) tumor samples from the patient were retrospectively subjected to detailed genomic analyses (Table 1), including central NGS testing using Synthetic Lethal Interactions for Precision Diagnostics (SNiPDx™), and WGS. SNiPDx™ is a novel, targeted sequencing panel consisting of 26 DNA damage response (DDR) genes, capable of distinguishing monoallelic and biallelic loss-of-function alterations21. WGS was performed as previously described22. Genes were considered to have biallelic loss if one of the following criteria was met: (1) homozygous deletion; (2) compound heterozygous mutation; (3) mutation and loss of heterozygosity; or (4) mutation and non-overlapping heterozygous loss. Genes were considered to have monoallelic loss if the mutation was detected without loss of heterozygosity. VAF, inferred tumor purity and ploidy were also considered in characterizing gene allelic status. Mutational signatures were decomposed using SigProfileExtractor23. HRD classification was determined using CHORD24. Telomere variant repeats and other telomeric features were quantified using TelomereHunter16, and ALT status was classified according to a model trained on known ALT+ samples from Pan-cancer Analysis of Whole Genomes3.

Efficacy and safety outcomes

Radiological tumor response was assessed by the investigator according to Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST v1.1). Camonsertib adverse events (AEs), assessed using National Cancer Institute Common Terminology Criteria for Adverse Events, version 5.0, were recorded and graded from the start of study treatment until 30 days after the last administration of trial therapy. A comprehensive list of secondary outcomes is provided in the protocol.

Molecular response analysis

Molecular response by ctDNA level was assessed, defined as ≥50% decline in circulating tumor fraction compared with study baseline for at least one timepoint. For ctDNA monitoring, plasma samples were collected from the patient at baseline and at each cycle of treatment. ctDNA analysis was performed using Tempus xF (Tempus), a genomic-variant detection and methylation-based tumor fraction assessment using Guardant Infinity™ (Guardant Health), and TracerDx (Tracer Biotechnologies), a tumor-informed dPCR-based approach using 48 variants selected from the patient WGS profile based on sequence context, sequencing quality metrics, and multiplex compatibility.

Native fluorescence in situ hybridization analysis

The protocol was adapted from Claude et al.,10. The FFPE tissue section was deparaffinized in HistoClear II (Electron Microscopy Sciences, 64111) three times for 3 minutes each and rehydrated in 100%, 95%, 70%, and 30% ethanol for 3 min each. The tissue section was then treated with permeabilization buffer (20 mM Tris-HCl pH 8.0, 50 mM sodium chloride (NaCl), 3 mM magnesium chloride, 300 mM sucrose, 0.5% Triton X-100) for 1 hour at 37 °C, followed by treatment with 500 µg/mL RNase A in RNase A blocking buffer (1 mg/ml bovine serum albumin, 3% normal goat serum, 1 mM EDTA, 0.1% Triton X-100 in phosphate-buffer saline) for 1 hour at 37 °C. Subsequently, the section was serially dehydrated in 70%, 85%, and 100% ethanol for 2 min each and air-dried.

Next, the tissue section was incubated with hybridization solution (100 nM TelG-Alexa488 (PNA Bio, F1008), 50% deionized formamide (Millipore, S4117), 2x Saline-Sodium Citrate (SSC), 1x Blocking Reagent (Roche, 11096176001) at room temperature for 2 h. It was then washed twice for 15 min in 50% formamide, 2x SSC, 20 mM Tris-HCl pH 7.4, and three times for 5 min in 150 mM NaCl, 0.05% Tween-20, 50 mM Tris-HCl pH 7.4. The section was serially dehydrated again, air-dried, and mounted with ProLong Gold Antifade Mountant with DNA Stain DAPI (Invitrogen, P36931).

The confocal microscopy images were acquired at 40X water immersion using the PreciScan function in Opera Phenix Plus (Revvity), and the ssTeloC foci were automatically quantified using Spot Count in Harmony (Revvity, version 5.2). Reporting of ssTeloC foci was restricted to tumor cells as annotated by a board-certified pathologist.