Phase II Study of Adavosertib in Patients With Tumors Containing BRCA1 and BRCA2 Mutations: Results From the NCI-MATCH ECOG-ACRIN Cancer Research Group (EAY131) Subprotocol Z1I

Abstract

PURPOSE

Inhibition of WEE1, a tyrosine kinase responsible for G2 arrest, results in premature mitotic entry and double-strand DNA breaks. Adavosertib is a selective, ATP-competitive, and small-molecule WEE1 kinase inhibitor. In an adavosertib single-agent, Phase I trial, partial responses (PRs) were observed in patients with solid tumors carrying pathogenic variants (PVs) in BRCA1/2. In this trial, we further evaluated adavosertib in patients with PV in BRCA1/2 solid tumors.

PATIENTS AND METHODS

Eligible patients met criteria for the National Cancer Institute-Molecular Analysis for Therapy Choice master protocol and had a diagnosis of advanced BRCA-mutated solid tumor (germline or somatic mutations were accepted); 33 patients were enrolled, and 30 received study treatment. Adavosertib was administered orally, 300 mg once daily, with 5 days on and 2 days off over a 3 week cycle of 2 weeks on and 1 week off, until disease progression or unacceptable toxicity. Radiologic assessment was performed every three cycles. The primary end point was overall response rate (ORR); secondary end points included 6-month overall survival (OS6) and 6-month progression-free survival rate (PFS6).

RESULTS

The ORR was 3.3% (90% CI, 0.2 to 14.9); the OS6 was 57.3% (90% CI, 41.9 to 72.7); the PFS6 was 23.4% (90% CI, 10.7 to 36.2). One PR (fallopian tube serous carcinoma) and six cases of stable disease (SD, >6 months) were observed. In patients with SD <6 months or progressive disease (nonresponders), significantly increased PI3K/AKT/mTOR signaling pathway gene transcripts suggested activation of this resistance mechanism. The primary reason for treatment discontinuation was disease progression. Common side effects included mye-losuppression, fatigue, nausea, anemia, vomiting, and diarrhea.

CONCLUSION

In heavily pretreated, advanced solid tumor patients with PV in BRCA1/2, adavosertib treatment resulted in low ORR, and this trial did not meet the primary end point.

INTRODUCTION

Balance between cyclins and cyclin-dependent kinases (CDKs) is critical for cell cycle progression. CDK activity is tightly regulated by mitogenic signals and cell cycle checkpoints that respond to DNA damage. WEE1, a checkpoint tyrosine kinase, enforces the G2/M checkpoint by inhibiting CDK1 and CDK2 activity through phosphorylation.^1,2 This regulation maintains replication fork stability under stress and prevents cells with DNA damage from entering mitosis, thereby avoiding mitotic catastrophe.^1,2 Inhibiting WEE1 disrupts these checkpoints, drives cells into mitosis with unresolved DNA damage,^3,4 and represents an appealing therapeutic approach to enhance replication stress and induce apoptosis.

Adavosertib is a selective, ATP-competitive, small-molecule WEE1 inhibitor (IC₅₀ = 5.18 nmol/L) that blocks CDK1 phosphorylation, inducing G2/M checkpoint escape and DNA damage in vitro.^5,6 WEE1 inhibition was initially hypothesized to be effective in p53-deficient tumors, which rely on the G2 checkpoint after DNA damage, and indeed, preclinical studies in p53-deficient pancreatic and colon cancer xenografts demonstrated that adavosertib increased cytotoxicity of DNA-damaging agents.^5–7 In addition, early-phase clinical trials in advanced solid tumors^8,9 and phase II trials in TP53-mutant ovarian cancer and pancreatic cancer reported promising activity of adavosertib in combination with chemotherapy.^10–12 However, toxicities have limited the clinical utility of these adavosertib combinations.

A phase I study of single-agent adavosertib revealed that it may be effective in tumors with BRCA1/2 pathogenic variants (PVs).¹³ In that study, partial responses (PRs) were observed in patients with germline BRCA1/2 PVs, motivating this phase II study of single-agent adavosertib in patients with BRCA1/2 PV+ solid tumors as part of the National Cancer Institute-Molecular Analysis for Therapy Choice (NCI-MATCH) trial (subprotocol Z1I), aiming to elucidate its therapeutic potential in this genetically defined patient population.

PATIENTS AND METHODS

NCI MATCH Description

The NCI-MATCH trial, developed by the Eastern Cooperative Oncology Group (ECOG)-American College of Radiology Imaging Network (ACRIN) Cancer Research Group and the NCI, evaluates targeted treatments for actionable molecular alterations across tumor types. Based on promising activity in patients with solid tumors carrying BRCA1/2 PVs,¹³ we evaluated the clinical benefit of single-agent adavosertib treatment in this molecularly defined subset of patients, meeting the criteria for NCI-MATCH inclusion.

Study Design

Eligible patients met all criteria for the NCI-MATCH master protocol¹⁴ and provided tumor tissue and, in most cases, blood specimens for next-generation se uencing (NGS) to identify BRCA1/2 PVs; patients with germline or somatic PVs were accepted. Adavosertib was administered orally, 300 mg once daily, with 5 days on and 2 days off. Following this pattern, adavosertib was administered in 21-day cycles with 2 weeks on and 1 week off. Cycles were repeated until disease progression or unacceptable toxicity. This study was approved by the Central Institutional Review Board and followed the Declaration of Helsinki and the International Conference on Harmonisation Good Clinical Practice guidelines. All participants provided written informed consent before enrollment. The study was registered at ClinicalTrials.gov (NCT04439227).

Patient Selection

Adult patients with solid tumors carrying BRCA1/2 PVs were eligible if they had progressed on standard treatment or had no standard treatment available. Eligibility required adequate hematopoietic, liver, and kidney function; ECOG performance status (PS) ≤1; and submission of fresh biopsy or archival tissue collected <6 months before enrollment. Patients with ovarian cancer were required to have received prior poly (ADP-ribose) polymerase (PARP) inhibitor treatment. Patients must have been off prior therapies for at least 28 days to be eligible. Exclusions included prior WEE1 inhibitor use or active brain metastasis.

Tumor Profiling

Actionable mutations were assessed using an NGS panel of 143 genes, including single-nucleotide variants, insertions or deletions, amplifications, and selected fusions.^15,16 Patients were accepted if they had BRCA1/2 PVs identified by molecular profiling performed for clinical reasons at one of 26 CLIA-accredited laboratories approved to identify NCI-MATCH–eligible patients.

Assignment to the Subprotocol and Evaluation of Response

Patients were assigned to this NCI-MATCH subprotocol (EAY131-Z1I) using a prospectively defined NCI-designed informatics rules algorithm (MATCHBOX), with physician confirmation, as previously described.¹⁴ Briefly, if multiple actionable PVs were identified, the patient was assigned based on the variant with the highest level of evidence, followed by the variant with the highest allele frequency. If these were equivalent, subprotocol assignment was dictated by the protocol with the fewest patients. Response was evaluated by radiologic assessment every three cycles using RECIST v.1.1.¹⁷

Toxicity Evaluation

Toxicity was evaluated in all patients who started therapy (N = 33) using NCI Common Terminology Criteria for Adverse Events (AEs) version 4 (CTCAEv4, via Rave case report forms). Grade 3 and 4 toxicities had to resolve to grade ≤2 before starting the next cycle, with a maximum delay of 2 weeks beyond the actual cycle length of 21 days. Toxicities persistent beyond 2 weeks led to treatment discontinuation and follow-up for toxicity resolution. Patients were allowed up to two dose reductions (200 mg and 175 mg, each given once daily for 5 days on, 2 days off); a requirement for further reductions resulted in study removal.

Statistical Considerations

The primary efficacy analysis cohort included eligible, treated patients with solid tumors carrying BRCA1/2 PVs, with a goal of 35 patients to achieve 31 eligible patients. The safety cohort included patients who started protocol therapy regardless of eligibility. The primary objective was to evaluate the objective response (complete or partial response) rate (ORR) for adavosertib, with a response rate of 5 of 31 patients (16%) or more considered a signal of activity. The study had a 91.8% power to detect a 25% response rate at a 1.8% type I error rate (one sided), under the null response rate of 5%. If fewer than 31 patients were eligible, primary efficacy was assessed using 5% one-sided exact binomial tests of the null hypothesis that the response rate was ≤5%. Given that 30 patients were included in the primary efficacy analysis cohort, adavosertib would be considered worthy of further testing if five or more patients achieved partial response or complete response (CR). Secondary objectives were 6-month progression-free survival (PFS6), PFS, 6-month overall survival (OS6), OS, toxicity, and evaluation of predictive biomarkers. PFS and OS were estimated using the Kaplan-Meier method.

DNA Sequencing Workflow

Prealigned DNA sequencing and whole-exome sequencing (WES) data were stored in and retrieved from the Genomic Data Commons (GDC).¹⁸ All analyses were performed using the NIH HPC Biowulf supercomputing cluster.¹⁹ DNA sequencing libraries were prepared using the Nextera Rapid Capture Exome kit v1.2 and sequenced on an Illumina HiSeq platform (2 × 75 bp) at the Broad Institute of MIT and Harvard. Reads were aligned to human reference b37, an hg19/GRCh37 analog. Somatic short variants were identified using Mutect2 in tumor-normal mode for samples with available tumor and normal tissue. Mutect2 parameters included the panel of normal references, and common germline variant call format (VCF) retrieved from GATK MNP was turned off. VCFs were compiled to Mutation Annotation Format (MAF) using the vcf2maf tool, with standard functional annotations (ClinVar, PolyPhen, Sift, gnomAD, 1000G, etc). MAF files were used for downstream visualization and analysis (Tableau). Copy number variation was estimated using PureCN, with thresholds of ≥6 for focal amplifications, ≥7 for all other amplifications, and <1.5 for deletions. Shallow deletions could not be differentiated. For genomic analysis, we defined a responder as CR + PR + SD >6 months and a nonresponder as PD + SD <6 months.

Methods for Homologous Recombination Deficiency Assessment

BRCA1/2 PVs were called by Sentieon TNhaplotyper (GATK Mutect2 implementation²⁰ [Sentieon Inc, San Jose, CA])²¹ and annotated for oncogenicity using OncoKB.²² PV origin (somatic or germline) was determined by comparison with matched normal WES data. Allele-specific copy number data were derived from tumor/normal paired WES data using Sequenza.²³ Biallelic status was defined as the presence of BRCA1/2 PV with loss of one copy of the corresponding wild-type allele. Monoallelic status was defined as a BRCA1/2 PV without loss of heterozygosity (LOH) at the BRCA1/2 locus. Single Base Substitution Signatures (COSMIC version 2²⁴) were detected using the R package deconstructSigs²⁵ based on somatic mutations. Mutational Signature 3 (Sig3), associated with defective homologous recombination repair, was specifically assessed.

Genomic scar–based homologous recombination deficiency (HRD) scores were calculated using the scarHRD algorithm²⁶ using Sequenza data, incorporating HRD-LOH, telomericallelic imbalance, and large-scale state transitions. HRProfiler²⁷ was applied to the WES data to infer the probability of HRD using a machine learning approach. Here, the model pretrained on ovarian cancer was used for other types of cancer samples in the cohort.

RNA Sequencing Workflow

RNA sequencing libraries were prepared using the Illumina TruSeq RNA kit and sequenced on an Illumina HiSeq 2500 (Illumina, San Diego, CA) at the University of North Carolina. Harmonized RNA-seq BAM files were obtained from the GDC and aligned to the GRCh38d1.vd1 transcriptome. Analyses were performed using the NIH HPC Biowulf supercomputing cluster¹⁹ Transcript abundance was quantified using Salmon (GENCODE v22). Differential gene expression analysis was performed using DESeq2 (R), and gene set enrichment analysis was performed using GSEA and variance-stabilized expression data.

Genomic Data Availability

The genomic data generated in this study have been submitted to the GDC. The associated patient clinical and outcome data have been submitted to dbGaP with accession phs002058. The data are available through controlled access. Please refer to the dbGaP process for requesting data access.²⁸

RESULTS

Patient Population and Disposition

Between March and August 2017, 33 patients were enrolled with a data cutoff of May 29, 2020. Three patients were excluded from the primary efficacy analysis cohort (n = 30) because of an unresolved grade 2 baseline AE related to prior treatment, no measurable disease, an ECOG PS of 2, and a washout of <28 days (Data Supplement, Fig S1). The median age was 61.5 years (range, 36–79 years), 63.3% was female, 36.3% was male, and 67% (n = 20) had received >3 prior therapies (Table 1). Prior therapies included the PARP inhibitor olaparib, which was received by 71% and 22% of patients with gynecologic and breast cancer, respectively, or platinum therapies (cisplatin, carboplatin, oxaliplatin), which were received by 86%, 56%, and 64% of patients with gynecologic, breast, and other cancer, respectively. Solid tumor types are listed in Table 1. Of these 30 patients, 43.3% had BRCA1 PVs (n = 13), 56.7% had BRCA2 PVs (n = 17), and 46.7% had TP53 aberrations (n = 14).

TABLE 1.

Primary Efficacy Analysis of Patient Characteristics and Histologic Types

Patient Characteristics and Histologic Types	No. (% of total)a
Female	19 (63.3)
Age, years (min, 25%, med, 75%, max)	(36.0, 54.5, 61.5, 67.5, 79.0)
Race
White	23 (76.7)
Black	4 (13.3)
Asian	2 (6.7)
Multirace	1 (3.3)
Hispanic	2 (6.7)
PS 0	6 (20.0)
No. of prior therapies
1	4 (13.3)
2	2 (6.7)
3	3 (10)
>3	20 (66.7)
Missing	1 (3.3)
Weight loss in previous 6 months:
<5%	22 (73.3)
5% to <10%	6 (20.0)
10% to <20%	2 (6.7)
Breast	9 (30)
Ductal carcinoma of the breast	7 (23.3)
Neuroendocrine carcinoma of the breast versus adenocarcinoma with neuroendocrine differentiation	1 (3.3)
Infiltrating duct/inflammatory carcinoma of the breast	1 (3.3)
Serous adenocarcinoma of fallopian tube (HPV−)	1 (3.3)
Adenocarcinoma of the gallbladder	1 (3.3)
Gastrointestinal	4 (13.3)
Squamous cell carcinoma of the anus, HPV− associated (ISH+)	1 (3.3)
Adenocarcinoma of the rectum	1 (3.3)
Squamous cell carcinoma of the rectum, anus, anal canal, and male genital organs (p16 IHC+, HPV ISH−)	1 (3.3)
Adenocarcinoma of the rectosigmoid junction	1 (3.3)
Adenocarcinoma of the lung	1 (3.3)
Melanoma	1 (3.3)
Ovarian	5 (16.6)
Serous carcinoma of the female reproductive tract, ovary by historyb	1 (3.3)
Clear cell adenocarcinoma of ovary (small sample)	1 (3.3%)
Serous adenocarcinoma of ovary	3 (10)
Adenocarcinoma of pancreas	4 (13.3)
Epithelioid mesothelioma of anterior mediastinum	1 (3.3)
Adenoid cystic carcinoma of the parotid gland	1 (3.3)
Squamous cell carcinoma of esophagus (lower third)	1 (3.3)
Malignant solitary fibrous tumor of pharynx	1 (3.3)

NOTE. Bold values represent parent term.

Abbreviations: ISH, in situ hybridization; IHC, immunohistochemistry; HPV, human papillomavirus.

^an = 30.

^bNo pathology report.

The ORR in the primary efficacy analysis cohort was 3.3% (90% CI, 0.2 to 14.9), the estimated PFS6 rate was 23.4% (90% CI, 10.7 to 36.2; Data Supplement, Table S1; Figs 1A and 1B), and the estimated OS6 was 57.3% (90% CI: 41.9%-72.7%). One patient with fallopian tube serous carcinoma achieved a partial response (PR), and six patients achieved stable disease (SD; Figs 2A and 2B). Among eight evaluable TP53-mutant tumors, three experienced clinical benefit (37.5%; two SD, one PR); of the 14 evaluable TP53-wildtype tumors, four experienced clinical benefit (28.6%; four SD; Fig 2B). The patient achieving a PR, observed in a BRCA1-and TP53-mutated fallopian tube serous adenocarcinoma, experienced a 33% tumor reduction (Fig 3) and survived 11 months before dying of sepsis, an event possibly related to treatment.

FIG 1.

Kaplan-Meier plots of (A) PFS and (B) OS. Dotted lines indicate 90% CI. OS, overall survival; PFS, progression-free survival.

FIG 2.

(A) Best percentage change from baseline (20 patients included, excluded eight unevaluable patients and 2 PD patients without disease assessment, PD because of new lesions). *New lesions; ^nontarget lesions had unequivocal PD; ^interval progression of liver metastases and metastatic lymphadenopathy in the chest. (B) Variants by gene and patients for analyzable cases (n = 30). Each column represents a patient. The histograms at top and left give the number of variants for each patient (top) and proportion of patients with variants in each gene (left). Response data and histology for individual patients are listed at the bottom. NE, nonevaluable; PD, progressive disease; PR, partial response; SD, stable disease.

FIG 3.

Computed tomography scan results of the patient with serous adeno fallopian tube carcinoma. This patient had confirmed partial response with more than 30% reduction in target lesion: (A) Baseline, 189 mm and (B) 125 mm, cycle 15.

Adverse Events

AEs were evaluated in all patients who started treatment (N = 33). One patient did not experience any AEs; thus, data were recorded for 32 patients. The most frequently reported AEs were gastrointestinal and hematologic, the majority of which were grade 1 or 2 (Data Supplement, Tables S2–S4). The most common gastrointestinal AEs were diarrhea, nausea, and vomiting. The most common hematologic AEs were anemia and leukopenia. Fatigue was another common AE. Reported grade 4 AEs included anemia, neutropenia, thrombocytopenia, leukopenia, dehydration, and hyper-glycemia. There was one unexpected grade 5 AE of sepsis (Data Supplement, Table S3). AEs accounted for 13.3% (4 of 30) of patients who were taken off treatment; the majority (63%; 19 of 30) were taken off treatment because of progressive disease (Data Supplement, Table S5).

Genomic Profiling

Twenty-six patients had sufficient tumor material for DNA or RNA sequencing; 20 patient samples were evaluable for response. Matched tumor and blood DNA samples were analyzed by WES (18 evaluable), and tumor samples were analyzed by RNAseq for 20 patient samples. For genomic analysis, clinical benefit status was defined as CR + PR + SD >6 months (responder, n = 4) or PD + SD <6 months (nonresponder, n = 16).

To precisely define the genomic status of patient tumors, for those patients with matched tumor and blood samples (n = 22), we assessed the BRCA1/2 PV allelic status (monoallelic or biallelic), mutation origin (germline or somatic), and the presence of LOH at the BRCA1/2 loci and we established HRD-associated genomic scar metrics, including large-scale transitions (LST), telomeric allelic imbalance, and two composite scores, HRD-sum and HRProfiler (Table 2). Of tumors analyzed, 36.3% (n = 8) had monoallelic PVs, of which four were somatic mutations (Table 2). Generally, tumors harboring biallelic BRCA1/2 PVs demonstrated high HRD scores and a BRCA1/2 mutational signature (Sig3), which reflects defective HRD associated with widespread genomic instability that is often sensitive to DNA damage–targeting therapies, possibly including WEE1 inhibitors. Indeed, patients with biallelic BRCA1/2 PVs showed a trend for better probability of survival (P = .16; Data Supplement, Fig S2). Three of four responders (75%; CR + PR + SD >6 months) had biallelic BRCA1/2 PVs, a BRCA1/2 mutational signature, and high HRD scores, whereas patients who lacked these features failed to receive durable benefit, suggesting that HRD enriches for response to adavosertib. This suggests that HRD creates a vulnerability to WEE1 inhibition; however, WEE1 inhibition is insufficient in these tumors, suggesting that other compensatory pathways are likely activated.

TABLE 2.

Homologous Recombination Deficiency Assessment

Histology	BRCA1/2 PV	Biallelic/Monoallelicd	Mutational Signature	LOH	LST	TAI	HRD-Sum	HR Profiler	R/NRe
Miscellaneous neoplasm, NOS	BRCA1_p.K679*_0.4871a	Biallelic	Other	2	25	14	41	0.00	R
Invasive breast carcinoma	BRCA1_p.L1086*_0.3634a	Monoallelic	MMR	7	14	12	33	0.00	NE
Salivary gland cancer	BRCA1_p.L28Rfs*3_0.437a	Monoallelic	Other	2	23	12	37	0.00	NR
Breast cancer, NOS	BRCA1_p.Q563*_0.5217a	Monoallelic	BRCA1/2	1	19	12	32	0.00	NR
Adenocarcinoma—pancreas	BRCA2_p.K3326*_0.4615b	Monoallelic	Other	4	22	7	33	0.00	NE
Squamous cell carcinoma—anus	BRCA2_p.Q1925*_0.3719a	Monoallelicd	Other	9	11	14	34	0.00	NR
Invasive breast carcinoma	BRCA1_p.E670*_0.2992a	Monoallelicd	POLE	8	8	8	24	0.04	NR
Adenocarcinoma—colon	BRCA2_p.S2219*_0.1089a	Monoallelicd	Other	14	24	33	71	0.07	NR
Adenocarcinoma—pancreas	BRCA2_p.K3326*_0.785b	Biallelic	Other	9	25	16	50	0.12	NE
Invasive breast carcinoma	BRCA2_p.Q1507*_0.5593a	Monoallelicd	AID/APOBEC	14	23	11	48	0.32	NR
Melanoma	BRCA1_p.S713*_0.65a	Biallelic	Other	12	19	25	56	0.54	NR
Ovarian epithelial cancer	BRCA1_p.N1355Kfs*10_0.9649a	Biallelic	BRCA1/2	17	27	26	70	0.61	NR
Carcinoma, NOS	BRCA2_p.V1283Kfs*2_1a	Biallelic	UV	15	18	16	49	0.82	NE
Invasive breast carcinoma	BRCA2_p.W1692Mfs*3_0.5a	Biallelic	BRCA1/2	18	24	33	75	0.86	NR
Invasive breast carcinoma	BRCA1_p.E23Vfs*17_0.8144c	Biallelic	BRCA1/2	16	27	33	76	0.90	NR
Gall bladder carcinoma (adeno)	BRCA2_p.I2449Dfs*2_0.7339a	Biallelic	BRCA1/2	26	32	38	96	0.97	NR
Neuroendocrine cancer NOS	BRCA2_p.I1859Kfs*3_0.85a	Biallelic	BRCA1/2	22	31	30	83	0.97	NR
Ovarian epithelial cancer	BRCA2_p.R3128*_0.6218a	Biallelic	BRCA1/2	29	31	35	95	0.99	R
Invasive breast carcinoma	BRCA1_p.P1315Dfs*12_0.6986a	Biallelicd	BRCA1/2	23	31	33	87	0.99	NE
Fallopian tube carcinoma	BRCA1_p.Q1756Pfs*74_0.7093c	Biallelic	BRCA1/2	31	32	34	97	0.99	R (PR)
Adenocarcinoma—pancreas	BRCA2_p.N588Kfs*26_0.4783a	Biallelicd	BRCA1/2	33	37	33	103	1.00	NR
Female reproductive system cancer NOS	BRCA1_p.E23Vfs*17_0.6996b	Biallelic	BRCA1/2	19	30	33	82	1.00	R

Abbreviations: CR, complete response; HR, homologous recombination; HRD, homologous recombination deficiency; LOH, loss of heterozygosity; LST, large-scale transitions; NE, nonevaluable; NOS, not otherwise specified; NR, nonresponder; PD, progressive disease; PR, partial response; R, responder; SD, stable disease; TAI, telomeric allelic imbalance.

^aMutation likely oncogenic.

^bMutation likely neutral.

^cMutation oncogenic.

^dSomatic mutation (germline if not indicated).

^eClinical benefit status, defined as defined as R (CR + PR + SD >6 months) or NR (PD + SD <6 months).

Genomic correlates of response and resistance to adavosertib were evaluated using WES data. No mutated genes were found exclusively in patients achieving clinical benefit (responders), whereas mutations in five genes (CFLAR, BRF1, ADAMTS4, ADD2, CBX5) were only found in patients who did not achieve clinical benefit (nonresponders); however, none of these reached statistical significance (P >.05, Fisher’s exact test; Data Supplement, Figs S3 and S4). We also evaluated whether somatic alterations in genes in a WEE1 inhibitor gene set significantly affected time to progression; the difference was not significant (P = .8; Data Supplement, Table S6).

By RNAseq, we found several genes with significantly upregulated expression (P < .05 and at least a positive Log2 fold change) in responders compared with nonresponders (Fig 4A) and in nonresponders compared with responders (Fig 4B), classifying responders and nonresponders as defined in the genomic analysis. Using ENRICHR²⁹ to infer whether these genes are related to specific pathways or gene sets, among the genes significantly upregulated in nonresponders, we found genes associated with mTORC1 signaling, p53, and PI3K/AKT/mTOR signaling pathways (Fig 4C). The upregulation of mTOR pathways may be particularly relevant as mTOR pathways can participate in known mechanisms of resistance to WEE1 inhibition.³⁰ Several genes in these pathways were similar, including PPARG, TRIB3, and SFN, which were all downregulated in responders or upregulated in nonresponders.

FIG 4.

(A) Heatmap depicting the top responder-specific differentially expressed genes, overexpressed in responders compared with nonresponders. (B) Heatmap depicting the top nonresponder-specific differentially expressed genes, overexpressed in nonresponders compared with responders. In (A) and (B), heatmaps depict significant (raw P value with a cutoff of .05) genes which have a positive Log2 fold change (>2 or <–2). Samples are presented in dendrogram order according to complex agglomerative clustering with Euclidean distance. Samples are identified by best confirmed response (PD, PR, SD), by days to progression, and by clinical benefit status, defined as CR + PR + SD >6 months (responder, n = 4) or PD + SD <6 months (nonresponder, n = 16). (C) Gene set enrichment analysis (ENRICHR) of genes overexpressed in nonresponders, with significant gene set terms shown. The “Overlapping Genes” column indicates the number of genes in the Gene Set Term (denominator) and the number of those present in the genes overexpressed in nonresponders (numerator); overlapping genes are listed in the “Genes” column. CR, complete response; PD, progressive disease; PR, partial response; SD, stable disease.

DISCUSSION

We report the results of this multicenter, single-arm, phase II, single-agent study of the WEE1 kinase inhibitor adavosertib in patients with BRCA1/2 PV+ solid tumors. Overall, the PR rate was 3.3% (90% CI, 0.2 to 14.9) and the estimated PFS6 rate was 23.3% (90% CI, 10.7 to 36.2), with SD observed in patients with a solitary fibrous tumor, fallopian tube cancer, or ovarian cancer. This response rate was below what we expected based on results from our phase I trial¹³ and did not meet our primary end point, but it is consistent with four phase I/Ib trials of adavosertib conducted in heavily pretreated patients with advanced solid tumors that reported PR rates ranging from 3% to 14.3%.^13,31–33 Higher response rates have been reported in select phase II adavosertib monotherapy trials, including a study in recurrent uterine serous carcinoma (n = 34) that demonstrated an ORR of 29.4%, including one CR.^34,35 In our initial phase I trial, both responders were patients with ovarian cancer carrying germline BRCA1/2 PVs,¹³ suggesting increased adavosertib activity in these patients. In the current study, all clinical benefits were recorded among patients with germline BRCA1/2 PVs and biallelic somatic loss; however, patients with those features also experienced progressive disease. Supporting this, two other trials, a phase 1b study of patients with advanced solid tumors³³ and a phase II study of patients with small cell lung cancer,³⁶ also did not identify BRCA1/2 PVs as a biomarker for response to WEE1 inhibition. In addition, a recent preclinical study found that BRCA-mutant ovarian cancer cells were resistant to adavosertib, possibly because of enhanced nonhomologous end joining that managed to repair DNA damage.³⁷ These data suggest that DNA damage induced through WEE1 inhibition may not be repaired through BRCA1/2-dependent mechanisms. Previous work demonstrated that patients with non–BRCA-associated cancer types were less likely to respond to targeted therapies, including platinum agents and PARP inhibitors, than patients with BRCA-associated cancers because of heritable cancer risk (eg, breast, ovary, pancreatic, prostate).³⁸ Similarly, our finding that mutational signatures and HRD scores correlated with response to adavosertib may relate to tumor lineage.

Biomarkers related to WEE1 inhibition include members of the PTEN-PI3K/AKT/mTOR signaling pathway. Breast cancer cells with high PTEN recovered postadavosertib treatment, suggesting that PTEN might contribute to resistance against WEE1 inhibitors.³⁹ Overexpression of AKT/mTOR has been associated with resistance against adavosertib in small cell lung cancer and ovarian cancer models, and combining WEE1 and mTOR inhibition has shown synergistic effects.^30,40 In our study, adavosertib nonresponders, as defined in the genomic analysis, exhibited significantly increased RNA transcripts of PI3K/AKT/mTOR pathway genes, including SLC9A3R1, SLC7A5, SCD, TRIB3, and SFN, perhaps suggesting dysregulation of PIK3/AKT/mTOR signaling and activation of resistance against adavosertib. Notably, increased expression of SFN (14-3-3σ), which binds CDK1 to activate the G2/M checkpoint after DNA damage and blocks mitosis initiation, can cause treatment resistance and predict poor prognosis in various cancers.^41,42 Thus, increased SFN may compensate for WEE1 inhibition by enabling DNA repair and preventing mitotic catastrophe. Given the increase of SFN transcripts in nonresponding patients, increased expression of SFN may represent an mTOR-related resistance mechanism to WEE1 inhibitors.

CONTEXT.

Key Objective

This study reports findings from National Cancer Institute-Molecular Analysis for Therapy Choice (EAY131) subprotocol Z1I, which evaluated the therapeutic potential of WEE1 inhibition (adavosertib) in patients with solid tumors harboring BRCA1/2 alterations.

Knowledge Generated

Of 30 patients with heavily pretreated solid tumors, one achieved a partial response and six achieved stable disease. The overall response rate in the primary efficacy analysis cohort was 3.3% (90% CI, 0.2% to 14.9%). The estimated 6-month progression-free survival rate was 23.4% (90% CI, 10.7% to 36.2%).

Relevance

Clinical responses to adavosertib were variable in patients with BRCA1/2 pathogenic variants (PVs). While all clinical benefits were recorded among patients with biallelic, germline BRCA1/2 PVs, patients with those features also experienced progressive disease. These findings demonstrate that BRCA1/2 PVs alone are insufficient to predict sensitivity to WEE1 inhibition. Enrichment of responses in patients with molecularly defined homologous recombination deficiency (HRD) suggests HRD as a biomarker to evaluate in order to optimize WEE1-targeted therapies.

DATA SHARING STATEMENT

A data sharing statement provided by the authors is available with this article at DOI https://doi.org/10.1200/PO-25-00769.