Evaluation of Prexasertib, a Checkpoint Kinase 1 Inhibitor, in a Phase Ib Study of Patients with Squamous Cell Carcinoma

Checkpoint kinase 1 (CHK1) is a multifunctional protein kinase and regulator of DNA replication and the DNA damage response (DDR; ref. 1). In response to exogenous DNA damage, CHK1 mediates cell-cycle arrest to allow time for DNA repair or, if the damage is extensive, to trigger apoptosis. CHK1 is essential for homologous recombination-mediated repair of double-strand DNA breaks; and it affects the initiation of DNA replication origin firing, stabilization of replication forks, resolution of replication stress, and coordination of mitosis, even in the absence of exogenous DNA damage (2).

Recent advances in understanding how cancer cells lose control of the cell cycle and repair DNA damage have led to the development of a new generation of inhibitors targeting these mechanisms. Inhibitors of CHK1, ataxia telangiectasia mutated kinase (ATM), ataxia telangiectasia and Rad3-related kinase (ATR), and WEE1 inhibitors are in clinical development (3–5). These checkpoint kinases are crucial for maintaining genomic integrity by reducing replication stress and for mediating the responses for the repair of DNA damage. Although CHK1 inhibitors traditionally have been developed in combination with cytotoxic chemotherapy (6–9), newer CHK1 inhibitors, including prexasertib monomesylate monohydrate (hereafter referred to as prexasertib) (3), LY2880070 (10), and CCT245737 (11), are being evaluated in studies exploring the monotherapy effects of CHK1 inhibition. These agents seek to leverage the role of CHK1 not only in cell-cycle control and DDR, but also in facilitating normal cellular processes such as the coordination of DNA replication. As a result, tumors that have increased replication stress and/or defects in DNA damage repair pathways may be more sensitive to these agents (12).

Prexasertib is an inhibitor of CHK1 and, to a lesser extent, CHK2. Prexasertib disrupts DNA replication, induces DNA damage, and subsequently prevents repair, eventually leading to death by replication catastrophe (1). The phase I study reported here consisted of two parts: dose escalation and dose expansion. The dose-escalation phase in solid tumors determined the recommended phase II dose and schedule of prexasertib to be 105 mg/m² on day 1 every 14 days (3). The most common toxicity was grade 4 neutropenia; nonhematologic toxicity was predominantly grade 1 or 2. Objective clinical responses were observed in two patients with squamous cell carcinoma [SCC; SCC of the head and neck (SCCHN) and SCC of the anus]. These were the first objective responses reported with a CHK1 inhibitor as monotherapy and it may be notable that both SCCHN and SCC of the anus can be HPV-associated tumors, which are anticipated to be associated with higher levels of replication stress (13, 14). Because these initial responses were in patients with SCC, the study was amended to further evaluate prexasertib in phase Ib expansion cohorts in patients with SCC. Multiple cohorts were included in the dose-expansion portion of the study, but for ease of reporting, the data for patients with metastatic/advanced SCC of the anus, SCCHN, or squamous cell non–small cell lung cancer (sqNSCLC) were combined across cohorts and are the focus of this manuscript. Because patients with these tumors commonly have replication stress and defects in DDR-related genes, inhibition of CHK1 may be an attractive therapeutic strategy to improve outcomes for patients with these diseases (15–17).

This study included a dose-escalation cohort followed by expansion cohorts to gain preliminary information on efficacy in patients with SCC. Key inclusion criteria for all patients described in this report included an Eastern Cooperative Oncology Group performance status of 0 or 1; measurable disease by Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 (18); adequate hematologic, hepatic, and renal function; discontinuation and recovery from the acute effects of prior therapies before enrollment; and failure on available standard therapies.

Patients with metastatic/advanced SCC were included in five dose-expansion cohorts (Supplementary Table S1): two cohorts included patients with SCHHN (B1 and C1), one cohort included patients with sqNSCLC (C2), one cohort included SCC of the anus (C3), and one cohort included patients of any SCC histology (B2).

The specific criteria for each of these cohorts are as follows: Patients with SCCHN were included if they had histologically confirmed recurrent or metastatic SCC of the oropharynx, hypopharynx, or larynx not amenable to local therapy and had received one but not more than three prior systemic chemotherapies for recurrent or metastatic disease (cohort B1); or had histological diagnosis of recurrent or metastatic SCC of the oropharynx, hypopharynx, oral cavity, or larynx not amenable to curative local therapy and had at least two prior systemic regimens for recurrent or metastatic disease (cohort C1). Patients with histologically or cytologically diagnosed stage IV sqNSCLC who had received one but not more than two prior systemic regimens (including one platinum-containing regimen) for recurrent or metastatic disease were included (cohort C2; ref. 19). Patients with SCC of the anus were included if they were stage IIIB (N2 or N3) or stage IV (cohort C3) and the disease was considered not to be curable by local therapy (19). Patients with SCC of any histology were included if they had histologically confirmed metastatic or recurrent SCC (cohort B2).

Key exclusion criteria were symptomatic central nervous system malignancies, current hematologic malignancy, QTc interval >470 milliseconds on screening electrocardiogram, serious cardiac conditions, systolic blood pressure <90 mm Hg or recurrent symptomatic orthostatic hypotension, serotonin-secreting carcinoid tumor or prior history of drug-induced serotonin syndrome, family history of long-QT syndrome, and use of concurrent medication known to cause QTc prolongation or to induce Torsades de Pointes.

This phase I, multicenter, nonrandomized, open-label trial included a previously described dose-escalation phase (3), followed by dose-expansion cohorts (phase Ib) for patients with SCC. Prexasertib was administered as a 1-hour infusion at the recommended phase II dose and schedule of 105 mg/m² on day 1 of a 14-day cycle (3).

The primary objective of the expansion cohorts was to determine the safety, toxicity, and recommended phase II dose of prexasertib. In addition, the overall response rate according to RECIST version 1.1 for patients with specific types of SCC was estimated. The secondary objectives for parts B and C were to estimate clinical benefit rate (complete response + partial response + stable disease), progression-free survival (i.e., the time from study entry to confirmed disease progression or death), and duration of disease control (i.e., the time from the date of enrollment [first dose] to disease progression or death). As an exploratory analysis, biomarkers predictive for response were also measured in pretreatment tissue samples by next-generation sequencing.

This study was conducted in accordance with Good Clinical Practices, the Declaration of Helsinki, and approval by each institution's Ethical Review Board. Patients provided written informed consent.

Treatment-emergent adverse events (TEAE) were graded using the National Cancer Institute-Common Terminology Criteria for Adverse Events version 4.0 (20). Granulocyte colony-stimulating factor (G-CSF) and antibiotic use was permitted in accordance with American Society of Clinical Oncology guidelines (21). Prophylactic use of G-CSF was not permitted during cycle 1 of treatment.

Responses were evaluated by RECIST version 1.1 approximately every 6 weeks (18). Objective responses were confirmed at least 4 weeks later.

To identify genomic biomarkers associated with drug response, pretreatment biopsies or archived tissue were subjected to next-generation sequencing using the FoundationOne (Foundation Medicine, Inc.) gene panel (T5, 288 genes or T7, 393 genes), which is compliant with Clinical Laboratory Improvement Amendments. Human papillomavirus (HPV) status was inferred from DNA sequencing using HPV-specific capture probes. Per FoundationOne's guidance and based on their published work (22), more than 10 reads/million was considered HPV-positive. The percentage of tumor cells was based on hematoxylin and eosin staining.

Data were summarized by tumor type unless stated otherwise. Continuous variables were summarized using the number of patients, as well as means, medians, standard deviations, standard errors, minimums, and maximums. Categorical endpoints were summarized using the number of patients, as well as frequencies and percentages with standard errors. Progression-free survival was estimated using a Kaplan–Meier estimate of the median progression-free survival time [and 90% confidence interval (CI)]. All data were summarized and/or estimated using descriptive statistics and formal statistical testing was not performed, except for the analysis of progression-free survival by HPV status.

A total of 101 patients were included in the dose-expansion cohorts and treated with prexasertib at the recommended phase II dose of 105 mg/m² (Fig. 1). Results are reported by disease type across the dose-expansion cohorts (Supplementary Table S1). A total of 26 patients with SCC of the anus, 57 patients with SCCHN, and 16 patients with sqNSCLC were included. Two patients, one with SCC of the skin and one with vaginal SCC, were only included in the total dose-expansion population cohort. The baseline disease subclassifications for patients with SCCHN are shown in Supplementary Table S2.

Most patients across disease types were heavily pretreated, with 49% of patients across tumor types having received ≥3 prior systemic regimens. The median number of prior systemic regimens was 2 (range, 1–13) for patients with SCC of the anus, 3 (range, 1–5) for patients with SCCHN, and 2 for patients with sqNSCLC. Patients with SCCHN were the most heavily pretreated (Table 1). Only one patient (sqNSCLC) received prior immunotherapy.

Among all cohorts, the most common all-grade TEAEs related to study treatment were neutropenia, thrombocytopenia, anemia, leukopenia, and fatigue (Fig. 2). Serious adverse events (SAE) were seen in 35% of patients with SCC of the anus, 53% of patients with SCCHN, and 44% of patients with sqNSCLC. The TEAEs related to study treatment occurring in >10% of patients in any tumor type are shown in Supplementary Table S3.

The most frequently observed toxicity was neutropenia, which was predominantly grade 4 (71% across all tumor types). The nadir consistently occurred approximately 1 week after each dose. The extent of neutropenia was consistent across cohorts (Supplementary Table S3) and did not appear to attenuate over time. Although 12 patients experienced febrile neutropenia, there were no deaths or discontinuations attributed to febrile neutropenia. Patients with SCCHN had a higher rate of febrile neutropenia (18%) than that of patients with SCC of the anus (4%) or sqNSCLC (6%). Across tumor types, G-CSF was administered prophylactically to 40 (40%) patients and to treat low neutrophils in 50 (50%) patients; 29 (29%) patients required G-CSF treatment both prophylactically and to treat low neutrophils. Patients with SCC of the anus [17/26 (65%)] or SCCHN [39/57 (68%)] required a similar rate of G-CSF support; however, both of these were higher than the rate of G-CSF support required by patients with sqNSCLC [5/16 (31%)]. The only nonhematologic toxicities occurring in more than 10% of the total patient population were fatigue, nausea, and headache. Most nonhematologic toxicities were grade 1 or 2 (Supplementary Table S3).

Patients completed a median of three cycles (range, 1 to 44). Dose delays occurred in 61/101 (60%) patients across cohorts. Most commonly, delays across tumor cohorts were due to neutropenia: 9/26 (35%) patients with SCC of the anus, 10/57 (18%) patients with SCCHN, and 3/16 (19%) patients with sqNSCLC. Across all cohorts, dose delays for neutropenia were required in 22/101 (22%) patients, despite the relatively short 14-day cycle length, indicating the event resolved quickly. Dose reductions due to neutropenia happened in 10/101 (10%) patients of the total patient population. No patients discontinued due to an adverse event deemed related to study treatment.

A total of 101 patients in the dose-expansion cohorts were evaluable for efficacy evaluations (Table 2). One patient (4%) with SCC of the anus had a complete response that lasted 18 months. Three patients with SCC of the anus had partial responses that lasted 9.9, 10.1, and 14.0 months, respectively, resulting in a 15% overall response rate for patients with SCC of the anus. Three patients with SCCHN had partial responses that lasted 4.8, 7.0, and 12.4 months, respectively, resulting in a 5% overall response rate for patients with SCCHN. None of the patients in the sqNSCLC cohort had a complete or partial response. Across the dose-expansion cohorts, 45 (45%) patients had a best response of stable disease, with clinical benefit rates in each tumor type of 58% for SCC of the anus, 49% for SCCHN, and 56% for sqNSCLC. The clinical benefit rate at 3 months across tumors was 29% (23% SCC of the anus, 28% SCCHN, 44% sqNSCLC). The median duration of disease control was approximately 4 months for each tumor type: 4.2 months (90% CI, 2.8–6.2) for SCC of the anus, 4.2 months (90% CI, 3.1–5.2) for SCCHN, and 3.9 months (90% CI, 3.0–6.2) for sqNSCLC. Of the 45 patients with stable disease, 29 had stable disease for at least 3 months (Table 2). The clinical benefit rate for the total dose-expansion population was 51%, which was similar to rates across the individual tumor cohorts (Table 2).

Figure 3 shows the maximal percent change in tumor size from baseline by best overall response for the combined expansion cohorts as well as by tumor type. Heat maps below the waterfall plots for each patient by tumor type demonstrate notable genetic alterations and are discussed in the next section.

Kaplan–Meier curves of progression-free survival for patients with individual tumor types are shown in Figure 4. Median progression-free survival was 2.8 months (90% CI, 1.9–4.2) for patients with SCC of the anus, 1.6 months (90% CI, 1.4–2.8) for patients with SCCHN, and 3.0 months (90% CI, 1.4–3.9) for those with sqNSCLC.

Genomic data were generated from pretreatment tumor tissue (biopsy or archived tissue) for 34 patients with SCCHN and 14 patients with SCC of the anus. The analysis for genomic associations used both a hypothesis-centered approach focused on cell cycle, DDR, and phosphatidylinositol 3-kinase (PI3K) pathway alterations, as well as hypothesis-independent analysis. The latter approach did not reveal any notable associations. Samples from patients with sqNSCLC were not included due to the lower number of patients and the lack of efficacy observed in this cohort.

Best overall response and genetic data for 14 patients from the SCC of the anus cohort appear in Figure 3A. HPV positivity was 86% (12/14) and 100% mutually exclusive with TP53 mutations. No apparent association between clinical efficacy and HPV status was observed. As described below, genetic variants in patients with benefit included BRCA1/2, PARK2, FBXW7, and PIK3CA. Truncating polymorphic variants were observed in BRCA genes in two patients with partial response [A22 (BRCA2 K3326* and A23 (BRCA1 E23fs*17)); Fig. 3A and Supplementary Table S4]. The only patient with complete response in the trial (A24) displayed a likely deleterious splice mutation in the ubiquitin E3 ligase PARK2. Two patients with stable disease (A11 and A16) showed loss-of-function mutations in another E3 ligase, FBXW7. PIK3CA mutations were found in 43% of patients (6/14), with one being most likely subclonal. All PIK3CA mutations were present in HPV-positive patients and mapped to the helical domain of PIK3CA, with five of six present in patients with clinical benefit. Genetic variants found in patients with SCC of the anus with lower treatment benefit included cyclin D1 amplification (A1 and A6), alterations in tumor suppressors of the PI3K pathway including PTEN (loss in A2 and A7 and functional single nucleotide polymorphisms in A6 and A13), and a truncating mutation in STK11 (A6). Alterations affecting the open reading frame of two Fanconi genes were observed in a patient with progressive disease (A2).

Best overall response and genetic data for 34 patients from the SCCHN cohort appear in Figure 3B. HPV positivity was 47% (16/34) and 100% mutually exclusive with TP53 mutations. HPV positivity was observed at a higher frequency in patients with clinical benefit (68%, 13/19) than in patients with progressive disease (20%, 3/15). Similarly, greater clinical efficacy as measured by progression-free survival was observed in the HPV-positive versus HPV-negative cohort (median progression-free survival 4.5 months vs. 1.4 months, log-rank P = 0.0008; Supplementary Fig. S1). Genetic alterations in DDR pathway genes were observed in three patients with clinical benefit: BRCA1, BRCA2, and MRE11A were altered in an HPV-positive patient with stable disease (H38), truncating mutations in BRCA1 and ATR in a patient with stable disease (H43), and an MRE11A missense mutation with evidence of loss of heterozygosity in an HPV-positive patient with partial response (H45). The observed missense mutations correspond to rare polymorphisms that mapped to functional interfaces within the Mre11-Rad50-Nbs1 (MRN) complex (23). Similar to the SCC of the anus cohort, loss of function mutations in the E3 ubiquitin ligase FBXW7 were observed in three patients with SCCHN with stable disease (H32, H34, and H38). Genetic alterations observed in patients with lower treatment benefit in the SCCHN cohort included cell-cycle genes (cyclin D1, CDKN2A/B, RB1) as well as mutations in TP53, consistent with the expected features observed in HPV-negative SCCHN (17). Mutations in PIK3CA were observed at lower frequencies than reported in the Cancer Genome Atlas SCCHN study (12% vs. 37%; ref. 17). Several potentially activating alterations of the PI3K pathway appeared at a slightly higher frequency in patients with progressive disease (40%, 6/15) versus those with clinical benefit (26%, 5/19; Fig. 3B and Supplementary Table S4). Fanconi gene variants appeared at a higher frequency in patients with progressive disease (47%, 7/15) versus patients with clinical benefit (11%, 2/19).

This phase Ib trial was the first to evaluate the monotherapy clinical activity of a CHK1 inhibitor (3). After determining a recommended phase II dose of 105 mg/m² in patients with advanced or metastatic solid tumors, expansion cohorts in patients with SCC were initiated to confirm the safety and provide a preliminary characterization of the efficacy of prexasertib. As previously reported, a dose of 105 mg/m² administered intravenously once every 14 days results in exposure over the first 72 hours (area under the curve from 0 to 72 hours) that coincides with the exposure in mouse xenografts that resulted in maximal tumor responses (3). Although not reported here, pharmacokinetic data from the expansion cohort patients was consistent with the prior report and aligns with exposure predicted to correlate to clinical efficacy. Oral CHK1 inhibitors such as SRA737 (NCT02797964) and LY2880070 (NCT02632448), which may have different exposure and inhibition profiles than the intravenously administered prexasertib, are in phase I testing as single agents.

The safety profile of prexasertib was generally consistent across the 101 patients treated in the expansion cohorts, with grade 4 neutropenia being the most common treatment-related toxicity. G-CSF use was common. Because of the transient nature of neutropenia, using G-CSF to treat low neutrophils in the absence of other risk factors (e.g., fever or comorbidities) may not be of significant benefit. However, prophylactic use of G-CSF appeared to reduce the extent of neutropenia in a subset of patients. The most common nonhematologic toxicities deemed by the investigators to be related to prexasertib treatment were fatigue, nausea, headache, diarrhea, and anorexia. The majority of these events were grade 1 or 2 in severity. Notably, no patients experienced grade 3 or 4 vomiting, mucositis, constipation, or diarrhea. Other CHK1 inhibitors have been associated with cardiotoxicity, including myocardial infarction and significant QTc changes (7, 8). However, in this study, cardiac events related to study treatment were rare; only a single grade 1 study drug-related cardiac event (bradycardia) was observed. Thus, the dose of prexasertib at 105 mg/m² on day 1 of a 14-day cycle was confirmed as the phase II dose.

All three of the tumor types assessed in the expansion population lack effective treatment approaches for patients with advanced metastatic disease. Because of the rarity of metastatic SCC of the anus, most prior studies have been small retrospective assessments and have established 5-fluorouracil/cisplatin to be the standard of care (24). A notable exception is a recent phase II study in which 37 patients with SCC of the anal canal were treated with 3 mg/kg nivolumab every 2 weeks (25). Emerging data suggest that PD-1 or PD-L1 inhibitors, such as nivolumab, also may have benefit for these patients, and response rates of 17% and 24% have been reported (25, 26). Given the monotherapy activity of PD-1/PD-L1 inhibitors and prexasertib, combining these agents may be an intriguing therapeutic strategy (27, 28).

Similarly, the treatment options for patients with metastatic SCCHN are limited. Historically, response rates for patients with previously treated metastatic disease have been less than 15%, but PD-1 and PD-L1 inhibitors have recently demonstrated response rates of up to 18% (29, 30). In this study, patients with SCCHN were heavily pretreated (over 50% of the patients had three or more prior chemotherapy regimens). Although some objective activity was observed, the activity was modest with only 5% of patients achieving a partial response. This suggests that the activity of prexasertib in patients with SCCHN may best be leveraged in combination with other agents. An ongoing study is evaluating prexasertib in combination with chemoradiation in patients with untreated locally advanced SCCHN (clinicaltrials.gov identifier NCT02555644).

Although stable disease was observed in some patients with sqNSCLC, none of the patients had an objective response to prexasertib. The rationale for including patients with sqNSCLC was based on data indicating that sqNSCLC shared molecular similarities with SCCHN (17). However, recent data demonstrate that the activity of prexasertib is not limited to patients with squamous cell histology, and as demonstrated in this study, may be more influenced by tumors that have increases in replication stress and/or deficiencies in DDR pathways (31). As a result, it is not known whether the lack of objective response is due to the small sample size or whether the level of replication stress and/or defects in DDR pathways is insufficient to leverage the activity of monotherapy prexasertib in sqNSCLC.

To help understand why a subset of patients responded to prexasertib treatment, pretreatment tumor tissue was analyzed by targeted exome sequencing. A limitation of our analysis was that it was performed on a mix of archival tumor tissues and fresh biopsies (Supplementary Table S4); hence, genetic variation may have been induced by prior lines of therapy in a subset of tumors that we are not detecting. A recent genomic study in SCC of the anus reported similar genomic features before and after chemoradiotherapy (32).

Loss of function mutations in two classes of genes were found in patients with treatment benefit: DDR pathway genes (BRCA1, BRCA2, MRE11A, and ATR) and genes known to increase replication stress such as the E3 ubiquitin ligases known to target cyclin E1 (FBXW7 and PARK2; ref. 33). Potential loss of function mutations in these two classes of genes were not observed in patients lacking treatment benefit.

Prexasertib has been shown to induce replication catastrophe most likely by contributing to heightened replication stress (1). In addition, prexasertib is known to induce double-strand breaks that are repaired through DDR pathways such as homologous recombination repair (HRR), a pathway also dependent upon functional CHK1 (34). Consequently, the biomarkers identified as potentially being associated with response to prexasertib, such as BRCA1, BRCA2, MRE11A, ATR, PARK2, and FBXW7, are consistent with the role that CHK1 plays in facilitating HRR or reducing replication stress. In support of these biomarker observations, replication stress has emerged as a potential synthetic lethal mechanism for CHK1 inhibitors in various preclinical studies (5, 35). Loss of PARK2 and FBXW7 reduces proteolytic degradation of cyclin E1, leading to higher expression of cyclin E, which in turn results in the activation of CDK2 and increased licensing of late replication origins (33, 36). The role of CHK1 is to suppress origin firing and coincident replication stress by inhibiting CDK2. Consequently, cells with heightened replication stress due to enhanced activity of CDK2 are more susceptible to treatments such as CHK1 inhibition by prexasertib, which can further promote this stress to unstainable levels. The importance of either replication stress or HRR deficiencies in contributing to sensitivity to prexasertib is being explored further in an ongoing basket trial (NCT02873975) that focuses on patients whose tumors show alterations consistent with replication stress or defects in DDR.

In this study, a large subset of patients with SCC of the anus (86%) or SCCHN (47%) had HPV-driven tumors similar to those observed in other studies (13, 16, 37). However, the analysis is limited since HPV status was determined using a nonstandard approach (inferred from DNA sequencing using HPV-specific capture probes). This approach was used due to the limited amount of available tissue from the pretreatment biopsy, which precluded conducting both genomic profiling and assessing HPV by standard techniques such as in situ hybridization or p16 IHC. However, in a study assessing 40 SCCHN samples comparing HPV status as determined by p16 IHC and next generation sequencing, 100% concordance was observed (22). These data suggest that this nonconventional approach was a reasonable alternative given the limited quantity of sample. However, in future studies, it would be preferable to also have HPV determined by more standard techniques.

Patients with HPV-positive tumors have improved responses to chemotherapy and better outcomes compared to HPV-negative patients (37–40). Because approximately 90% of patients with SCC of the anus are HPV-positive and since the incidence of SCC of the anus is low (16, 41), it will be difficult to design studies that assess HPV-negative patients with SCC of the anus. This is in contrast to recurrent/metastatic SCCHN, for which the number of patients who are HPV-positive are comparable to the number of patients who are HPV-negative (13). In this study, HPV-positive SCCHN patients treated with prexasertib had a three-fold longer progression-free survival compared to patients who were HPV-negative. Because of the small sample size and nonrandomized nature of this assessment, it is unknown if this is a prognostic effect or influenced by the prexasertib mechanism of action. HPV-derived E6/E7 triggers replication stress via depletion of nucleotide pools (14). In addition, HPV replication triggers the expression of the cytosine deaminase APOBEC3B as part of the innate immune response (42). PIK3CA helical domain mutations (E542/E545 residues) are present at a higher frequency in HPV-positive versus HPV-negative SCCHN tumors, and evidence indicates that these mutations likely arise via an HPV-induced APOBEC-mediated mechanism (43). In fact, all seven PIK3CA mutations observed in our study in patients with HPV-positive SCCHN and SCC of the anus mapped to the helical domain, suggestive of APOBEC-induced mutagenesis as their source of origin. Previous studies suggest that cellular mechanisms responding to APOBEC3-induced DNA damage utilize the DNA replication checkpoint pathways (44). A knockdown of CHK1 in the context of APOBEC (A3A) expression has been shown to be detrimental to cellular viability (44). In our study, a nonstatistically significant trend toward increased benefit was observed in patients with SCC of the anus with a PIK3CA mutation compared to patients who did not have this mutation.

Finally, although the aforementioned points focused on markers of response, noted in our study was a higher frequency of PI3K pathway alterations associated with lower treatment benefit in the SCCHN cohort that could play a role as a prosurvival mechanism (45). Alterations in tumor suppressors of the PI3K pathway (as well as cyclin D1 amplifications) were also observed in the nonresponding patients with HPV-positive SCC of the anus. Also noted was the unexpected elevated frequency of Fanconi variants in patients with SCCHN who had a best response of progressive disease to prexasertib. A potential explanation for the latter observation lies in the role that Fanconi genetic variants play in the pathogenesis of non-HPV SCCHN (46). The role of Fanconi genetic variation as a determinant of response to CHK1 inhibitors needs further investigation.

In summary, the modest activity observed in this study suggests that further evaluations in these tumors should focus on combination therapy. A phase I evaluation of prexasertib in combination with cisplatin, pemetrexed, 5-fluorouracil, cetuximab, or a PI3K/mTOR inhibitor is currently ongoing (NCT02124148). In addition, the activity of prexasertib may not be driven by histology alone, but rather more by the underlying features of SCC tumors, such as replication stress (e.g., induced by HPV or genes upstream of cyclin E) and/or defects in DDR pathways (e.g., BRCA1 and BRCA2). Other pathways such as PI3K/mTOR and MAPK may be modifiers of prexasertib monotherapy. Recently, objective responses with prexasertib were reported in patients with high-grade serous ovarian cancer, another tumor type in which approximately 70% of patients are reported to have defects in HRR (e.g., BRCA1 and BRCA2) or increased replication stress (e.g., through cyclin E amplification; ref. 47). These data suggest the genetic attributes of the tumor, regardless of histology, may be more important drivers for guiding the future development of prexasertib.

K. Moore is a consultant/advisory board member for AstraZeneca, Clovis, Genentech/Roche, Immunogen, Janssen, and Tesaro. J. Infante is an employee of and has ownership interests (including patents) at Johnson & Johnson. S. Wijayawardana has ownership interests (including patents) at Eli Lilly. R. Beckmann has ownership interests (including patents) at Eli Lilly. A.B. Lin has ownership interests (including patents) at Eli Lilly. C. Eng reports receiving speakers bureau honoraria from Bayer, Merck, Roche and Sirtex, and reports receiving commercial research grants from Advaxis, Forty Seven, and Roche. No potential conflicts of interest were disclosed by the other authors.

Conception and design: D.S. Hong, H.A. Burris III, S. Jones, J. Infante, L. Golden, R. Martinez, A.B. Lin, C. Eng

Development of methodology: D.S. Hong, H.A. Burris III, S. Jones, J. Infante, L. Golden, R. Martinez, S. Wijayawardana

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D.S. Hong, K. Moore, M. Patel, S.C. Grant, H.A. Burris III, S. Jones, F. Meric-Bernstam, J. Infante, L. Golden, R. Martinez, J. Bendell

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D.S. Hong, K. Moore, M. Patel, H.A. Burris III, W.N. William Jr, J. Infante, L. Golden, W. Zhang, R. Martinez, S. Wijayawardana, A.B. Lin, J. Bendell

Writing, review, and/or revision of the manuscript: D.S. Hong, K. Moore, M. Patel, S.C. Grant, H.A. Burris III, W.N. William Jr, S. Jones, F. Meric-Bernstam, J. Infante, L. Golden, W. Zhang, R. Martinez, S. Wijayawardana, R. Beckmann, A.B. Lin, C. Eng, J. Bendell

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W.N. William Jr, L. Golden

Study supervision: D.S. Hong, K. Moore, M. Patel, J. Infante, L. Golden

The authors thank the patients who participated in this trial and the study coordinators, nurses, nurse practitioners, clinical research assistants, and doctors who assisted with the research. The authors thank Christine Martersteck for her input and contributions to the biomarker analyses. The authors also thank Elizabeth Kumm for statistical support and Eric Westin and Christopher Slapak for their contributions to the CHK1 program. Eli Lilly and Company contracted with Syneos Health for writing support provided by Andrea Humphries, PhD, and editorial support provided by Angela C. Lorio, ELS.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.