First-in-Human RNA Polymerase I Transcription Inhibitor CX-5461 in Patients with Advanced Hematologic Cancers: Results of a Phase I Dose-Escalation Study

Despite significant progress in the treatment of hematologic malignancies with chemotherapy, mAbs, and cellular therapies over the last 40 years, with corresponding improvements in survival outcomes, there remain many patients who are not cured with existing therapies, necessitating the investigation of agents with novel modes of action (1–4).

The availability of functional ribosomes is a fundamental requirement for growth and proliferation in mammalian cells. The uncontrolled growth of cancer cells correlates with elevated ribosome biogenesis and also morphologically abnormal nucleoli, the sites of ribosome biogenesis; in fact, increased nucleolar size and number has been used as a marker of aggressive malignancies for more than 100 years (5, 6). RNA polymerase I (Pol I) transcription of the ribosomal RNA (rRNA) genes (rDNA) is rate-limiting for ribosome biogenesis, a high-energy consumption process that is significantly elevated in rapidly dividing tumor cells (7, 8). As rDNA transcription underpins this process, it requires precise regulation; Pol I–mediated transcription is tightly controlled by oncogenes such as MYC, RAS, and PI3K, which when activated in cancer cells contribute to hyperactivation of rDNA transcription (9–13). Furthermore, perturbation of rDNA transcription is known to elicit a nucleolar stress response (NSR), which is heightened in cancer cells, and leads to activation of both p53-dependent and p53-independent stress-response pathways (10, 14–17). Thus, rDNA transcription represents a key hub of coordinated regulation by oncogenic and tumor suppressor signaling pathways and offers novel opportunities for therapeutic targeting to treat the broad range and large numbers of human malignancies, including those driven by these oncogenes. Furthermore, a ribosome biogenesis stress response elicited by the indirect and/or nonspecific targeting of rDNA transcription is associated with the efficacy of many standard chemotherapeutics, including actinomycin D and some platinum-based agents, thus supporting a rationale for advancing this clinically effective concept (18, 19).

CX-5461 is the first-in-class selective small-molecule inhibitor of Pol I–mediated transcription, which inhibits rDNA transcription in the low nanomolar range by preventing the association of the Pol I–specific transcription initiation selectivity factor SL1 with the rDNA promoter, exhibiting greater than 200-fold selectivity relative to the inhibition of Pol II–driven transcription (20, 21). Preclinical studies show that inhibition of Pol I transcription by CX-5461 leads to cell-cycle arrest and cell death by both a canonical p53-dependent NSR and a noncanonical p53-independent nucleolar-localized DNA damage response (DDR) that requires activation of the ATM and ATR kinase signaling pathways (10, 14, 15). Importantly, as a single agent, CX-5461 shows a robust survival benefit in murine models of a range of hematologic cancers including MYC-driven B-cell lymphoma (10, 22), acute myeloid leukemia (AML; ref. 16), and multiple myeloma (17, 23), leading to rapid tumor cell clearance and/or disease reduction with minimal toxicity.

On the basis of this encouraging preclinical data, we initiated a first-in-human dose-escalation study of CX-5461 in patients with relapsed and refractory hematologic malignancies (Australia and New Zealand Clinical Trials Registry, #12613001061729). The primary objective was to determine the safety and tolerability of CX-5461 when administered by intravenous infusion once every 3 weeks. The secondary objectives were to assess the pharmacokinetic and pharmacodynamic profile of CX-5461 and the preliminary antitumor activity, and to investigate the impact of TP53 mutational status as well as mutations in other potential CX-5461 response factors including ATM/ATR pathway members, as predictive biomarkers of efficacy. Dose escalations were planned in 7 cohorts (from 25 to 450 mg/m²), in an accelerated design, with change to a 3+3 design based on predefined toxicity criteria. We report here on the findings of this first-in-human, first-in-class study.

Between July 27, 2013, and May 4, 2016, 17 patients with advanced hematologic malignancies were recruited, of whom 16 received CX-5461 at the Peter MacCallum Cancer Centre (Melbourne, Australia). The demographic features and baseline characteristics of treated patients are summarized in Table 1. All patients had measurable progressive disease at the time of enrollment and were representative of a heavily pretreated population with a median number of 7 prior therapies (range, 1–14 therapies). The median patient age was 60 years (range, 21–79 years) with 50% (8/16 patients) being female. The majority of patients (15/16 patients, 94%) had an Eastern Cooperative Oncology Group (ECOG) performance status of 0–1 and the predominant tumor types were myeloma (6/16 patients, 38%) and diffuse large B-cell lymphoma (DLBCL; 4/16 patients, 25%). All subjects completed at least 1 cycle of therapy; one patient discontinued treatment due to dose-limiting toxicity (DLT) and one due to patient decision, whereas all others were withdrawn from the study following disease progression.

Sixteen patients were treated in sequential cohorts at dose levels of 25 mg/m² (n = 3), 50 mg/m² (n = 4), 100 mg/m² (n = 4), 170 mg/m² (n = 3), and 250 mg/m² (n = 2; Table 1; the duration on study for each patient is presented in Fig. 1). The initial protocol for accelerated dose escalation was changed to a standard 3+3 design during cohort 1, due to the observation of cutaneous adverse events in the first patient treated at 25 mg/m². Of the 16 subjects enrolled, the median treatment duration was 2 cycles, i.e., 6 weeks (range, 1–18 cycles). Early disease progression during cycle 1 resulted in the withdrawal of 2 patients from the study, both of whom were deemed ineligible for DLT assessment but were included in the pharmacokinetic and pharmacodynamic analysis. Fourteen patients were included in the main safety analysis for determination of the MTD. A DLT of palmar–plantar erythrodysesthesia (PPE) was observed in the first patient treated at a dose level of 250 mg/m², with a similar grade 2 adverse event noted in the second patient enrolled at this dose, although not fulfilling DLT criteria. The MTD was determined by the safety committee as 170 mg/m².

All treatment-emergent adverse events (AE) occurring as grade 3 are summarized in Table 2, with investigator-assessed treatment-related AEs, reported as either possibly, probably, or definitely related to CX-5461 treatment, indicated (all CX-5461 treatment-related AEs are shown in Supplementary Table S1; all additional treatment-emergent AEs are shown in Supplementary Table S2). A DLT of grade 3 PPE was observed in the first patient (PMC-16) enrolled into cohort 5 (250 mg/m²), with the same AE occurring as a grade 2 event in the second patient of this cohort (PMC-17). This toxicity was characterized by pain, swelling, paresthesia, and/or erythema in the palms and/or soles of the feet (Supplementary Fig. S1A). Patient PMC-17 recommenced at a reduced dose of 170 mg/m² and successfully received 17 more cycles at this level without any recurrence of symptoms.

Another significant but dose-independent toxicity noted was the development of grade 1–3 photosensitivity in 50% (8/16) of patients treated. This occurred within 48 hours of treatment and presented as a sunburn-like rash in sun-exposed areas (Supplementary Fig. S1B). After the observation of this AE in 2 patients in cohort 1 (25 mg/m²), subjects were subsequently asked at enrollment to adhere to strict photosensitivity precautions and were able to continue on therapy if sun protection was used for 72 hours following drug infusions. One patient (PMC-14) had a protocol-mandated dose reduction due to a grade 3 photosensitivity event and continued on treatment for 3 more cycles without recurrence.

All cutaneous toxicities resolved without any sequelae. The grade 3 PPE was treated with corticosteroids and resolved in 3 weeks. No other significant drug-related toxicity was seen. Overall, CX-5461 was well tolerated, with the longest treatment durations extending up to 48 and 54 weeks in patients PMC-08 and PMC-17, respectively, and no significant hematologic toxicity was observed (Supplementary Table S1).

Mean plasma concentration–time profiles following the first cycle of CX-5461 treatment and the resulting pharmacokinetic parameters from these analyses are displayed in Supplementary Fig. S2 and Table 3, respectively. In summary, following intravenous infusion, CX-5461 reached a maximum plasma concentration (C_max) within 60 minutes of drug administration in all dose cohorts (Supplementary Fig. S2A). The terminal half-life (t_1/2) ranged from 19.2 to 92.4 hours and showed a trend to increase with dose escalation, reaching the highest average of 83.3 hours in cohort 5 (250 mg/m²) with residual drug being detectable in patients in cohort 5 at day 15 postdose (Table 3). However, no residual drug was detectable at predose, cycle 2, day 1 (C2D1) in the 2 patients tested (PMC-15, PMC-17; data not shown), one each in cohorts 4 and 5, respectively, which is consistent with the longest observed terminal half-life of 83.3 hours in cohort 5 (Table 3). Pharmacokinetics were generally linear and dose-proportional in terms of both C_max (Supplementary Fig. S2B) and AUC (Supplementary Fig. S2C) exposure parameters. Moreover, the appearance of multiple secondary peaks in plasma concentration–time profiles and a flattened terminal slope resulting in a longer observed half-life, especially with increasing dose, suggested the presence of enterohepatic recirculation of the drug (Supplementary Fig. S2A; Table 3).

Of the 16 patients evaluated for efficacy (Fig. 1; Supplementary Table S3), the best response was a confirmed partial response (PR) sustained for more than 12 months. This patient (PMC-17), with anaplastic large cell lymphoma (ALCL), previously had very short durations of response to prior therapies including conventional chemotherapy, mAbs, and an autologous stem-cell transplant. Another patient (PMC-05), with cutaneous T-cell lymphoma (CTCL) with large cell transformation, had clinical and radiologic evidence of response in the anatomic area of transformed disease. One patient (PMC-08) with heavily pretreated DLBCL had a period of sustained stable disease, receiving 16 cycles of treatment over 10 months, whereas another patient (PMC-15) with DLBCL achieved stable disease for 4 cycles. The best response noted in patients with myeloma was stable disease in 50% (3/6) of patients. Although all three patients with myeloma had actively progressing disease at study entry, their disease stabilization was maintained for 4–6 cycles.

CX-5461 is a selective inhibitor of Pol I transcription of rDNA, functioning by occluding SL1, a complex crucial for the recruitment of transcription-competent Pol I to the rDNA promoter (20). To confirm on-target drug activity, we developed a highly sensitive assay to measure Pol I–mediated transcription rates via FISH (16). The 5′-external transcribed spacer (5′ETS) of rRNA lies at the 5′ end of the 47S transcript and is rapidly processed following rRNA synthesis; therefore, its fluorescent detection by FISH is used as a surrogate readout of rDNA transcription rate, and the accuracy of the assay was previously validated by comparison with direct metabolic labeling of newly synthesized rDNA (10, 16). This allowed us to quantitate the abundance of 47S pre-rRNA levels in peripheral blood mononuclear cells (PBMC; Fig. 2A and B) and tumor tissue (Fig. 3A–D) in sequential samples during cycle 1.

To assess the pharmacodynamic effects of CX-5461 therapy, PBMCs were collected from all patients before treatment and then at 1, 4, 8, and 24 hours postinfusion of their first cycle of treatment. Levels of rDNA transcription inhibition were analyzed as the median percentage change in 5′ETS signal intensity from baseline (Fig. 2A and B). A consistent and robust decrease in rDNA transcription rate was observed at 1 hour postinfusion in PBMCs (Fig. 2A and B). The average level of inhibition was 49% (range, 22.9%–69.9%), 51.1% (34.4%–64.4%), 19.6% (−72%–69.7%), 43.8% (37.1%–48%), and 38.6% (6.8%–70.5%) in cohorts 1–5, respectively (Fig. 2B). The exception to this was in PBMC samples taken from patient PMC-11, where the baseline detected levels of 5′ETS signal intensity were unusually low. However, PBMC samples collected at later time-points displayed a rebound in rDNA rate, typically achieving levels similar to baseline by 24 hours post-treatment (Fig. 2B).

A comprehensive assessment of the quantitative dose–response relationship between CX-5461 plasma levels and Pol I–mediated transcription activity (5′ETS signal intensity) in PBMC samples across all dose cohorts is shown in Supplementary Fig. S3. Blood sampling post-treatment revealed an inverse association, where the maximal inhibition of rDNA transcription observed at 1 hour postinfusion correlated with the initial peak in drug plasma concentration levels observed in each dose profile (Supplementary Fig. S3). Moreover, falling plasma levels of CX-5461 during the linear phase of drug clearance was associated with a rebound in 5′ETS signal intensity in each patient, consistent with the utility of this assay for monitoring on-target drug activity (Supplementary Fig. S3). We also noted that the extent of rDNA transcription inhibition did not correlate with increasing CX-5461 dose (Fig. 2B) despite dose-proportional pharmacokinetics (Supplementary Fig. S2B and S2C).

Paired tumor biopsy specimens were obtained from patients with accessible disease [11/16 patients; bone marrow trephine (BM), n = 4; lymph nodes, n = 5; liver, n = 1; skin lesions, n = 3]predose and 24 hours postinfusion for the determination of on-target activity in tumor cells following therapy (Fig. 3A and B). Inhibition of rDNA transcription was observed in the majority of tumor-infiltrated specimens (10/13); however, the level of inhibition in rDNA transcription was variable (median range: 4%–68.5%) with no correlation to increasing dose (Fig. 3B). Tumor cells isolated from bone marrow by magnetic activated cell sorting (MACS; 4 patients) displayed a decrease in rDNA transcription rate in all patients (Fig. 3C and D). The degree of inhibition varied but was consistently inhibited at 4 hours and, again, did not correlate with CX-5461 dose (Fig. 3C and D). These data confirm that 5′ETS signal intensity is a reproducible biomarker of on-target drug activity in patient tumor samples and is consistent with data in PBMC samples (Fig. 2) showing that the amount of inhibition of rDNA transcription is independent of CX-5461 dose and also reflecting the potential for a rebound in rDNA transcription rates by 24 hours post-treatment.

The best clinical response was detected in a 63-year-old patient (PMC-17) who had ALCL with hematologic and cutaneous involvement, where an objective PR was seen in both compartments after 6 months of CX-5461 treatment (Fig. 4A). Tumor tissue sampled from this patient for correlative studies also displayed a decrease in rDNA transcription rate at both the 4-hour (22.1%) and 24-hour (8.1%) time points post cycle 1 (Fig. 4B).

CX-5461 inhibition of rDNA transcription has been demonstrated to mediate its therapeutic response by inducing a nucleolar stress response leading to activation of p53, but also via p53-independent mechanisms in a range of cancer cell types (10, 14, 15, 16, 22, 23). To further investigate the role of these pathways of therapeutic response to CX-5461, we performed targeted sequencing of 79 genes that were curated on the basis of their previously described role in the CX-5461 response mechanism (i.e., TP53, ATM, ATR, CHK2, MYC), their regulation of rDNA transcription (MYC, PI3K/AKT/mTOR signaling), and/or nucleolar function (NPM1, ribosomal proteins) and including well-known components of DNA repair and DDR (BRCA1, BRCA2, RAD51; complete list in Supplementary Table S4) to determine their mutational status in tumors of enrolled patients as a potential biomarker of therapeutic response. DNA was extracted from available tumor samples (n = 13) and sequenced using hybridization-based next-generation sequencing (data summarized in Fig. 1; Supplementary Tables S3 and S5). Collectively, 5 patients in total were found to harbor mutations in TP53, with all but one being associated with an early progression of disease following treatment (Fig. 1). Notably, 1 patient with DLBCL that carried a TP53 mutation achieved the longest period of stable disease (PMC-08, 16 cycles), and another patient in whom the TP53 status was not available also achieved stable disease (PMC-01; Fig. 1). Of the confirmed TP53 wild-type patients (n = 8), 1 achieved a prolonged PR (PMC-17, 18 cycles), 1 achieved a clinical and radiologic response in an area of transformed disease (PMC-05, 2 cycles), and 3 achieved periods of stable disease (all, 4 cycles; Fig. 1). The sequencing also identified 2 patients harboring mutations in ATM, one of the major known effectors of the CX-5461 p53-independent DDR-like response (14–16). Importantly, these patients were mutually exclusive from those who had tumors with TP53 mutations and were represented by a best confirmed response of stable disease (PMC-03) as well as rapid disease progression (PMC-10; Fig. 1). Finally, patient tumor DNA was also assessed for copy-number variation (CNV) at the MYC and MDM2 loci; however, no CNVs were detected at either loci in any of the 13 patients (Supplementary Table S5).

To extend our interrogation of the role of p53 in the therapeutic response to CX5461, we expanded our biomarker analysis on the tumor sample from one TP53 wild-type patient, a 79-year-old patient (PMC-05) with CTCL who displayed a clinical and radiologic response in an area of high-grade transformation following 1 cycle of treatment (50 mg/m², Fig. 4C–F). As seen by ¹⁸F-FDG-PET and digital photography, a reduction in tumor metabolic activity and corresponding clinical improvement was observed in the focal aggressive scalp lesion (Fig. 4C). As observed with other patient tumor samples assayed (Figs. 3 and 4B), this tumor response was associated with a decrease in rDNA transcription rates when compared with baseline (8.7%, Fig. 4D). Given this patient's wild-type TP53 tumor status, we assayed total p53 protein levels by IHC in a punch biopsy that directly sampled the cutaneous lesion pretreatment and 24 hours post-treatment and observed elevated p53 expression in the CX-5461–treated sample when compared with baseline (Fig. 4E). Furthermore, Western blot analysis of tumor samples showed that along with stabilization of p53 protein levels, a corresponding increase in the p53-target protein p21 after 24-hour CX-5461 exposure was observed (Fig. 4F). These data demonstrate that in a setting where the tumor was TP53 wild-type and the patient had a clinical response, inhibition of rDNA transcription was associated with activation of p53.

Here we report the results of a first-in-human study assessing the tolerability, safety, and anticancer activity of the small-molecule RNA polymerase I inhibitor CX-5461 in patients with advanced hematologic malignancies. We have determined an MTD of 170 mg/m² when the drug is administered by intravenous infusion once every 3 weeks. A DLT of grade 3 PPE was observed at a dose of 250 mg/m². An additional AE of photosensitivity was noted in 50% of the patients treated, independent of dose level, and this was manageable with avoidance of sun exposure for 72 hours after drug dosing. Although these cutaneous AEs were not anticipated from the preclinical data, they resolved without any sequelae. No other significant hematologic or other AEs were noted. Moreover, the patients in the study were heavily pretreated, with a median of 7 prior lines of therapy and with 10 patients having prior high-dose therapy followed by autologous or allogeneic hematopoietic progenitor cell transplant. Despite this, one patient with ALCL had a prolonged partial response for more than 12 months, and 5 patients with either multiple myeloma or DLBCL achieved a period of stable disease. Interestingly, a patient with CTCL demonstrated clinical benefit and radiologic response in a site of transformed disease, suggesting that T-cell lymphoma may be a tumor type which warrants further specific investigation.

Here, analysis of the on-target effect of CX-5461 in humans has been demonstrated by a decrease in rDNA transcription rates in both PBMCs and tumor tissue by RNA-FISH, establishing the utility of this assay for monitoring on-target drug activity. Maximum inhibition occurs 1–4 hours postdose and correlates with peak drug levels, with a return to baseline levels by 24 hours postdose. This raises the possibility that more frequent dosing (i.e., once-weekly administration) may improve the efficacy profile of this agent, and therefore studies designed to investigate this dosing schedule in the same population are planned to commence in the near future. Furthermore, a phase I dose-escalation study with day 1 and day 8 dosing of CX-5461 in a 4-week cycle in patients with advanced solid cancers is also currently ongoing (Canadian Cancer Trials Group; ClinicalTrials.gov identifier: NCT02719977; ref. 24). Interestingly, the inhibition of Pol I–mediated transcription occurred independently of dose level, providing a strong rationale for trialing more frequent dosing strategies at or below the MTD determined in this study. The pharmacokinetic profile of CX-5461 was generally predictable, linear, and dose-proportional, with a mean plasma half-life of 45.5 hours at the MTD. Drug was detectable in the plasma for up to 2 weeks following infusion in the highest dose cohort. The suggestions of enterohepatic recirculation of the drug and the possibility of drug accumulation with repeated doses have been taken into account in the protocol design, incorporating more frequent albeit lower dosing strategies.

There is now extensive preclinical evidence for improved efficacy using combinations of CX-5461 with other agents in clinical use or trials, including everolimus (mTOR inhibitor) and AZD7762 (CHK1/2 inhibitor) in B-cell lymphoma (15, 22), VE-822 (ATR inhibitor) in acute lymphoblastic leukemia (14), and carfilzomib (proteasome inhibitor) and panobinostat (pan-HDAC inhibitor) in myeloma (23), as well as PIM kinase inhibitors in prostate cancer (25). These are diseases which have previously been most effectively treated by combination drug therapy, and the demonstration of on-target effects at low doses of CX-5461 is encouraging for the possibility of clinical synergy with low toxicity in combination therapies.

In this study, although 5 of 8 patients who were wild-type and 1 of 5 patients who were mutant for TP53 showed beneficial clinical responses and 4 patients with TP53 mutations showed early progression, the overall small number of patients allows only speculation as to the degree by which TP53 mutation status can be used as a predictor of CX-5461 efficacy. It has also been reported that CX-5461 therapeutic activity requires mechanisms which are independent of p53, including activation of an ATM/ATR-dependent DDR (14–16) and an NSR-driven downregulation of MYC (17). Furthermore, in addition to inhibiting rDNA transcription, a recent study reported that CX-5461 induces replication-dependent DNA damage through stabilization of G-quadruplex (G4) DNA structures (26). The breadth of activity demonstrated in these preclinical studies, all of which can be mediated through targeting of rDNA, illustrates that the extent of CX-5461 therapeutic efficacy may depend on cancer cell type and the presence or absence of key molecular pathways, which could serve as predictors of response. For example, elevated MYC is predicted to sensitize cancers to Pol I inhibition (10, 13) and was shown to confer sensitivity to CX-5461 in prostate cancer models (25), whereas homologous recombination–deficient cancer cells lacking p53 display an enhanced response to CX-5461 (26). These studies suggest therapeutic potential for CX-5461 in a broad range of tumor types, and importantly a phase I trial evaluating CX-5461 in advanced solid tumors is ongoing (24). The patient tumor sequencing performed here also revealed 2 patients harboring ATM mutations that did not co-occur with TP53 mutations, with 1 of these patients achieving stable disease (Fig. 1; Supplementary Tables S3 and S5), which may suggest that one of these key CX-5461 response pathways must be intact for drug efficacy. Moreover, although future studies will examine the extent to which these mechanisms of action and their downstream responses contribute to the therapeutic efficacy of CX-5461, the data in this study demonstrate that CX-5461 shows on-target rDNA transcription inhibition in parallel with drug plasma levels, and this on-target activity in tumor samples correlates with activation of p53 in a patient in whom a clinical response was demonstrable.

The observation of PPE and photosensitivity as the only significant toxicities in our study has important implications for the ongoing development of the drug. Both were noted within 48 hours of drug dosing, which provides a timeframe for maximum risk of the adverse event. Precautions requiring strict sun protection are essential, including sunscreens that block UVA, as one patient experienced photosensitivity after sitting behind glass, which absorbs up to 97% of UVB. Importantly, adherence to these measures for 72 hours after drug dosing prevented recurrence of these events in all patients and allowed continuing treatment for prolonged periods. Similar toxicities have also been seen with drugs such as the BRAF inhibitor vemurafenib, and it has been possible to continue their use with appropriate supportive care and without dose reduction (27), as noted in our study.

In summary, in this first-in-human, first-in-class study, we have determined an MTD for CX-5461 in patients with advanced hematologic cancers and demonstrated tolerability with extended periods of dosing. Clinical responses have also been noted, which have been sustained and beneficial in some cases. Importantly, cutaneous adverse events have been observed, which will need to be actively managed during future development of this agent. These data provide a basis for further studies in appropriate tumor groups to explore more frequent dosing and combination strategies. Taken together, this study demonstrates for the first time that Pol I–mediated transcription of rDNA can be selectively and safely targeted in humans and validates a previously unexplored targeted therapeutic approach.

Patients were eligible for participation in the study if they had any measurable, relapsed, or refractory advanced hematologic malignancy, without any standard therapeutic options available, aged ≥18 years, with adequate organ and bone marrow function, an ECOG performance status of 0–2 at screening, and life expectancy ≥3 months. Adequate organ and bone marrow function were defined by the following: organ, creatinine clearance greater than 50 mL/minute, a total bilirubin ≤ 2 times the upper limit of normal and hepatic transaminases ≤ 2 times the upper limit of normal; bone marrow, hemoglobin (Hb) ≥ 9 g/dL, absolute neutrophil count ≥ 1.0 × 10⁹/L without the use of granulocyte colony-stimulating factor for 7 days prior to day 1 and platelets ≥ 1 × 10¹¹/L (except in cases of bone marrow infiltration, where lower thresholds were allowed). Patients with other malignancies requiring concurrent anticancer therapy or known active central nervous system disease were excluded from the study. Other key exclusion criteria included patients with a QT interval greater than 450 msec or significant bacterial, viral, or fungal infection. All subjects provided written informed consent prior to trial enrolment. The trial protocol was approved by the Institutional Review Board and the trial was conducted in accordance with the Good Clinical Practice guidelines and the ethical principles outlined in the Declaration of Helsinki and the International Conference on Harmonisation.

This was a single-center, first-in-human, phase I, open-label, dose-escalation study designed to establish the safety, pharmacokinetic, and pharmacodynamic characteristics of CX-5461 in patients with advanced hematologic malignancies. CX-5461 was administered as a 1-hour intravenous infusion on day 1 of each 21-day cycle. Dose escalations were planned in 7 cohorts (25–450 mg/m²), initially in an accelerated design, with change to a 3+3 dose-escalation schema based on the predefined toxicity criteria and DLTs of CX-5461. Patients remained on trial until disease progression, significant toxicity, or a clinical observation satisfying another withdrawal criterion was evident. The primary objective of the study was to define the safety and tolerability of CX-5461, by determining the DLTs and the MTD. The secondary objectives were to assess the pharmacokinetic and pharmacodynamic profile of CX-5461, investigate any preliminary clinical effects on tumor response, and to identify predictive biomarkers of efficacy. The secondary endpoints were assessment of grade 3+ adverse events, overall response, and determination of the pharmacokinetic profile of CX-5461.

Serial heparinized blood samples were collected from a peripheral vein on the contralateral side of the body to the site of injection. These samples were acquired prior to dosing; at 15 and 30 minutes during infusion; immediately upon completion of infusion; and then at 15 minutes, 30 minutes, and 1, 2, 4, 6, and 8 hours after dose administration. Subsequently, all patients returned to the clinic on days 2, 3, and 4, for additional blood sample collections at 24, 48, and 72 hours after day 1 treatment. After collection, samples were shipped frozen to CPR Pharma Services Pty Ltd where analysis of all pharmacokinetic parameters was performed. Parameters calculated and reported on the basis of actual sampling times included: maximum observed plasma concentration (C_max), time of maximum observed plasma concentration (T_max), effective half-life (t_1/2), and area under the plasma concentration–time curve. The pharmacokinetic population consisted of patients who received at least one intravenous dose of CX-5461 and who had evaluable pharmacokinetic data from plasma.

To confirm on-target drug activity and identify predictive biomarkers of therapeutic response, all patients were asked to provide peripheral blood samples at baseline and then at 1, 4, 8, and 24 hours post day 1 infusion. In those patients who had biopsy-accessible disease, tumor tissue specimens (from bone marrow, lymph nodes, liver, skin lesions) were obtained prior to treatment and 24 hours post cycle 1. When available, aspirate samples from the bone marrow were also harvested by MACS and the tumor cells collected by negative selection (antibodies used for MACS sorting listed in Supplementary Table S6). Pol I transcription levels were measured indirectly in these samples via FISH targeting the 5′ETS region of 47S pre-ribosomal RNA (RNA-FISH). Detailed methods describing cell isolation and preparation as well as RNA-FISH and immunoblotting (antibodies used for Western blot analysis listed in Supplementary Table S7) are provided in the Supplementary Methods. A custom targeted hybridization-based next-generation sequencing panel was used to identify sequence variants in 79 genes after extraction of DNA from available tumor samples (see Supplementary Methods for extended methods). All correlative samples in this study were collected, deidentified, and processed according to a protocol-specified standard operating procedure (see Supplementary Methods for details).

All statistical analyses were performed in SAS Analytics Software (version 9.3; SAS Institute, Inc.). Demographics, baseline characteristics, pharmacokinetic parameters, and clinical laboratory evaluations were summarized with descriptive statistics. For analysis of pharmacodynamic response during drug treatment, levels of Pol I transcription inhibition were analyzed by calculating the median percentage change in FISH signal intensity from each patient's baseline measurement.

A. Khot is a consultant/advisory board member for Celgene, Janssen, and Amgen. N. Hein reports receiving a commercial research grant from Pimera, Inc. J. Lim is a vice president, clinical & regulatory affairs at Senhwa Biosciences, Inc. and has ownership interest (including stock, patents, etc.) in the same. J. Soong has ownership interest (including stock, patents, etc.) in Senhwa Biosciences Inc. G.A. McArthur is a consultant/advisory board member for Pimera, Inc. R.B. Pearson is a consultant/advisory board member for Pimera, Inc. R.D. Hannan is an executive group manager for the Centre for Health and Medical Research at ACT Health Directorate and a lab head at Peter MacCallum Cancer Centre, reports receiving a commercial research grant from Pimera Inc., and is a consultant/advisory board member for the National Breast Cancer Fund. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Khot, J. Lim, E. Link, G.A. McArthur, R.B. Pearson, R.D. Hannan, G. Poortinga, S.J. Harrison

Development of methodology: A. Khot, N. Brajanovski, D.P. Cameron, N. Hein, E. Sanij, J. Lim, E. Link, P. Blombery, G.A. McArthur, R.D. Hannan, G. Poortinga, S.J. Harrison

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Khot, N. Brajanovski, D.P. Cameron, J. Lim, P. Blombery, E.R. Thompson, A. Fellowes, S.J. Harrison

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Khot, N. Brajanovski, D.P. Cameron, E. Sanij, J. Lim, E. Link, P. Blombery, G.A. McArthur, R.B. Pearson, R.D. Hannan, G. Poortinga, S.J. Harrison

Writing, review, and/or revision of the manuscript: A. Khot, N. Brajanovski, K.H. Maclachlan, E. Sanij, J. Lim, J. Soong, E. Link, P. Blombery, A. Fellowes, K.E. Sheppard, G.A. McArthur, R.B. Pearson, R.D. Hannan, G. Poortinga, S.J. Harrison

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Soong, E. Link, R.D. Hannan, S.J. Harrison

Study supervision: G.A. McArthur, G. Poortinga, S.J. Harrison

We thank the patients and their families for their participation in the study. We also thank the Peter MacCallum Cancer Centre Research Nursing Team. We thank Dr. Megan Bywater and Dr. Stephen Lade for advising on assays used in this study. This work was supported by the National Health and Medical Research Council (NHMRC) of Australia Development grant (#1038852), Cancer Council Victoria (CCV) grants-in-aid (#1084545 and #1100892) and the Peter MacCallum Cancer Foundation. Senhwa Biosciences provided financial support with respect to drug supply and pharmacokinetic studies. Researchers were funded by NHMRC Fellowships (to G.A. McArthur, R.B. Pearson, R.D. Hannan) and the Snowdome Foundation fellowship funding (to A. Khot).