Pharmacokinetic Dose-Scheduling Study of Hedgehog Pathway Inhibitor Vismodegib (GDC-0449) in Patients with Locally Advanced or Metastatic Solid Tumors

Aberrant hedgehog pathway activation has been implicated in a number of cancers (1–9). Vismodegib (GDC-0449; 2-chloro-N-[4-chloro-3-(pyridin-2-yl)phenyl]- 4-[methylsulfonyl]benzamide) is a potent small molecule inhibitor of the hedgehog signaling pathway and is being developed for treatment of various cancers (10–13). Vismodegib binds to and inhibits smoothened, a 7-transmembrane hedgehog pathway signaling protein. Activity of vismodegib was first shown in vivo in preclinical models of medulloblastoma (14), colon, and pancreatic tumors (9). Data from these studies were used to established a target plasma concentration of vismodegib to meet, or exceed, the unbound IC₉₅ for Gli1 inhibition (0.042–0.068 μmol/L; ref. 15). In a phase I study for patients with advanced malignancies, vismodegib was well tolerated, with pharmacodynamic (PD) evidence of hedgehog pathway inhibition via GLI1 suppression, and tumor regressions in patients with basal cell carcinoma and medulloblastoma (10, 11, 13).

In general, the pharmacokinetics (PK) of vismodegib are similar between healthy volunteers and cancer patients, with minor differences attributed to differences in amount and variability of plasma proteins. Vismodegib showed an unexpected PK profile upon oral daily dosing in healthy volunteers, with high, sustained micromolar plasma levels, and an estimated terminal half-life of 10 to 14 days (16). In a phase I study with continuous once daily dosing, total and unbound steady-state plasma concentrations in patients were typically achieved within 7 to 14 days of dosing; increasing the daily dose from 150 to 270 or 540 mg did not result in higher plasma concentrations, suggestive of nonlinear PKs with regard to dose and time (13). Furthermore, vismodegib levels were strongly correlated with alpha 1-acid glycoprotein (AAG) levels, showing parallel fluctuations of AAG and total drug over time and consistently low unbound drug levels. This unique PK profile was shown to be due to high-affinity reversible binding to AAG and binding to albumin (17), in addition to nonlinear absorption and slow metabolic elimination (18). A mechanism-based conceptual PK model was developed to provide a quantitative description of the observed vismodegib phase I data (18).

Due to the nonlinearity in vismodegib PKs after daily dosing, it was of interest to determine whether less frequent dosing could result in similar steady-state levels of total and unbound vismodegib to those obtained with daily dosing. The objectives of this phase Ib study were to evaluate the safety, tolerability, and relative effect of different dosing schedules on the steady-state PKs of total and unbound vismodegib. A vismodegib dose of 150 mg was administered either once daily (QD), three times per week (TIW), or once weekly (QW) after an 11-day loading dose (150 mg QD) in patients with advanced solid tumors that were refractory to treatment or for whom no standard therapy existed. A secondary objective was to assess the PKs of unbound vismodegib for 72 hours after a single dose. This study was conducted to provide insight into the optimal dosage of vismodegib for ongoing and future clinical studies.

All patients were at least 18 years of age and had histologically documented, incurable, locally advanced, or metastatic solid malignancy that had progressed after first-line and second-line therapy or for which there was no appropriate standard therapy. Patients with basal cell carcinoma (BCC) were excluded from this study unless they did not qualify for another open vismodegib clinical trial. Patients had Eastern Cooperative Oncology Group performance status less than or equal to 1. Women of childbearing potential had a documented negative serum pregnancy test. Exclusion criteria included untreated central nervous system malignancies or treated brain metastases that were not radiographically stable for 3 or more months, active infection requiring i.v. antibiotics, inability to swallow pills, clinically important history of liver disease, any medical condition or diagnosis that would likely impair absorption of an orally administered drug, pregnancy or lactation, treatment with strong CYP450 inhibitors and inducers, and any other conditions that would contraindicate investigational drug use in the opinion of the investigator. The study was approved by a local Human Investigations Committee. All patients provided written informed consent according to Federal and Institutional Guidelines before study procedures began.

This phase Ib study was a 2-part, randomized, open-label, multicenter design in which all patients received a daily 150 mg loading dose for 11 days (loading phase), followed by a 150 mg maintenance dose administered QD, TIW (Monday, Wednesday, Friday), or QW (Monday). Considering the potential effect of AAG on the PKs of total and unbound vismodegib, patients were stratified according to baseline AAG concentration [low (≤ 100 mg/dL), mid-range (> 100 mg/dL to ≤ 150 mg/dL), or high (> 150 mg/dL)] and then randomized to 1 of 3 treatment arms to receive vismodegib QD, TIW, or QW (randomization ratio of 2:3:3).

Vismodegib was taken at approximately the same time each day on an empty stomach (nothing but water for 2 hours before and 1 hour after drug administration). Dose modification from assigned regimen or treatment interruption was not allowed before day 57. On study visit days, patients received vismodegib at the clinic. In part 1, all patients received a single dose of vismodegib on day 1. No drug was taken on days 2 or 3. Starting on day 4, daily doses of vismodegib were administered to all patients for 11 days to bring patients to steady state (loading phase). In part 2, patients received vismodegib QD, TIW, or QW as a maintenance dose for up to an additional 42 days. The study schema is shown in Figure 1.

Accrual to the TIW or QW arms could be closed before day 57 if available PK and/or efficacy data suggested that TIW or QW dosing was futile. If the QW arm was closed to enrollment, patients currently receiving QW dosing could cross over to receive QD dosing for the remainder of the study, and subsequent patients would be stratified based on baseline AAG levels and randomized to the 2 remaining study arms. If the TIW arm was closed to enrollment, the QW arm would also be closed; patients currently receiving TIW or QW dosing could cross over to receive QD dosing for the remainder of the study.

The safety and tolerability of vismodegib were assessed by collecting the incidence, nature, and severity of adverse events, including clinically significant changes in vital signs or abnormalities in safety-related laboratory parameters. Severity was determined based on the National Cancer Institute Common Terminology Criteria for Adverse Events, Version 3.0.

The same sampling times were used for blood sample collection for the assessment of vismodegib (total and unbound) and AAG concentrations. In part 1, blood samples for plasma PK analyses were collected on day 1 before dosing and at 1, 2, 4, 6, and 24 hours after dosing. Blood samples for plasma PK analyses were also collected on day 3 (48 hours), and before dosing on day 4 (72 hours), day 8, and day 10. In part 2, blood samples for PK analyses were collected weekly, before dosing on days 15, 22, 29, 36, 43, 50, and 57. In addition, on days 29 and 50, PK samples were drawn at 1, 2, 4, and 6 hours after dosing, and daily morning samples corresponding to 24, 48, 72, and 96 hours after the first dose of each week.

Total and unbound plasma levels of vismodegib were determined using liquid chromatography–tandem mass spectrometry (16). Unbound vismodegib was measured in ultrafiltrate from plasma samples that underwent equilibrium dialysis (18). AAG levels in serum were determined by immunonephelometry using a standard clinical laboratory procedure.

Individual patient vismodegib PK parameter values for total and unbound plasma concentration-time data were derived using noncompartmental methods (WinNonlin version 5.2.1, Pharsight Corp.). Trough concentration at steady state was estimated as the average of trough values on days 29 to 57. The relationship between total plasma or unbound vismodegib concentration and AAG level was explored graphically and by linear regression analysis for each dosing schedule using the average steady-state trough vismodegib concentration and average AAG concentration from day 29 through day 57. A previously developed mechanism-based PK model was used to prospectively simulate the impact of dosing schedule on the PKs of vismodegib during study conduct (prior to observation of actual PK results) and predict vismodegib PK profiles for lower daily doses (18).

Vismodegib was generally well tolerated in this study, and the overall rate and severity of adverse events was similar across dose groups. All but one patient experienced at least one adverse event while on study. The most common adverse event was nausea (30 patients; 47.6%). A total of 24 patients (38.1%) experienced grade 3 to 5 adverse events on study (Table 1), but only 1 event in 1 patient was considered related to vismodegib (i.e., grade 3 hyponatremia).

Twenty-two patients (34.9%) experienced serious adverse events, the most common of which was intestinal obstruction (3 patients; 4.8%). No serious adverse event was considered to be related to vismodegib. No pattern of clinically significant change was identified for any of the hematologic, chemistry, or urinalysis parameters measured.

The average maximum plasma concentration (C_max) of total and unbound vismodegib was observed at 48 hours and maintained through 72 hours after a single 150 mg vismodegib dose. Total and unbound plasma concentrations seemed to be at steady state within 11 days (i.e., by day 15) of continuous once daily dosing of 150 mg vismodegib (loading phase). During the 6-week multiple-dose period following the loading phase, total and unbound plasma vismodegib concentrations were maintained with the QD dosing schedule. Total plasma vismodegib concentrations declined in a dosing frequency-related fashion, with the greatest decline observed in the QW group (Fig. 2A). Patients in the TIW and QW groups seem to have attained a new steady state for total plasma vismodegib levels by day 29 (Fig. 2A). A similar pattern was observed for unbound vismodegib compared with total plasma vismodegib concentrations, although the magnitude of the decrease in the TIW and QW groups was more pronounced for unbound than for total drug concentrations (Fig. 2B). A new steady state for unbound plasma vismodegib concentrations seems to have been achieved between days 29 and 36 for the TIW and QW groups (Fig. 2B).

By day 57, the total steady-state plasma vismodegib concentration had declined by 24% and 46%, on average, compared with the concentration on day 15 for the TIW and QW groups, respectively. Compared with day 15 (after 11 days of continuous QD dosing), the total steady-state plasma vismodegib concentration on day 57 had decreased by more than 50% in almost half the patients in the QW group. The observed decline in unbound steady-state plasma vismodegib concentration from day 15 to 57 was even more pronounced, with an average decrease of 50% and 80%, respectively, for the TIW and QW groups. These differences were in accordance with the change in dose (57% and 86% lower weekly dose for TIW and QW compared with QD). Approximately half the patients in the TIW group and all patients in the QW group had a more than 50% decrease in unbound steady-state plasma vismodegib levels.

Within the respective dosing schedules, similar mean C_max and area under the concentration-time curve (AUC_29–36 and AUC_50–57) values were observed for the 2 extensive PK intervals (days 29–36 and days 50–57) for both total and unbound drug, consistent with attainment of new steady state by day 29 to 36 for all dosing schedules. For total vismodegib concentrations, mean (±SD) C_max and AUC values on days 50 to 57 were highest for the QD group (33.9 ± 12.2 μmol/L and 5,240 ± 2,010 μmol/L*h, respectively), with slightly lower values in the TIW group (28.6 ± 13.8 μmol/L and 4,070 ± 1,780 μmol/L*h, respectively) and even lower values in the QW group (18.2 ± 4.23 μmol/L and 2,660 ± 666 μmol/L*h, respectively). Average total vismodegib steady-state trough values for days 29 to 57 exhibited the same dosing schedule-related trends as observed for C_max and AUC. For unbound vismodegib, the mean C_max and AUC values for the TIW and QW groups were considerably lower (up to 53% for TIW and up to 74% for QW) than those in the QD group for both PK intervals days 29 to 36 and days 50 to 57. Average unbound steady-state trough concentrations over these periods were lower for the TIW and QW groups compared with the QD group, in accordance with the observed differences in C_max and AUC.

Mean total and unbound plasma concentration-time profiles were similar and fairly constant over the 24-hour period on day 29 and day 50 for all 3 dosing schedules, further showing attainment of steady state by day 29. Accordingly, mean total and unbound AUC_0–24 values were also similar between days 29 and 50. The mean vismodegib AUC_0–24 values were lower in the TIW and QW groups than in the QD group, with greater differences observed for unbound than for total drug.

Total vismodegib steady-state trough concentrations and AAG levels were strongly correlated for the QD and TIW groups and were somewhat less correlated for the QW group (Supplementary Fig. S1A in the Supplementary Appendix). There was no correlation between the analogous unbound vismodegib concentrations and AAG, as expected (Supplementary Fig. S1B in the Supplementary Appendix).

The fraction unbound of vismodegib increased markedly with QD dosing in almost all patients, with a 3-fold higher average fraction unbound at steady state (0.00653 ± 0.00286) than after a single dose (0.00253 ± 0.00144; Fig. 3). In the one subject for whom a significant drop in fraction unbound was observed, the AAG level had increased relative to baseline.

Steady-state unbound vismodegib concentrations for the 3 dosing regimens in this study were compared with the target IC₉₅ values for Gli1 inhibition (0.068 μmol/L for patient-derived colorectal cancer xenograft models and 0.042 μmol/L for a Ptch^+/− allograft model of medulloblastoma exhibiting mutational activation of the Hh pathway; ref. 15). As shown in Figure 4, only the QD regimen provided unbound steady-state vismodegib concentrations in excess of the target IC₉₅ values for GLI1 inhibition in all subjects. Although the TIW regimen had a mean unbound steady-state vismodegib concentration (0.088 μmol/L) which was greater than the target IC₉₅ values, almost half the subjects in this group had concentrations below the more conservative target (0.068 μmol/L). The mean unbound steady-state vismodegib concentration in the QW group (0.0489 μmol/L) was close to the lowest target IC₉₅ for Gli1 inhibition, with the majority of the subjects in this dosing regimen having unbound drug concentrations below the target.

A previously developed mechanistic PK model based on phase I data was used to simulate vismodegib PK profiles with the alternative dosing regimens in this study (Supplementary Fig. S2 in the Supplementary Appendix; ref. 18). This model incorporates 3 main hypotheses [saturable absorption, higher affinity plasma protein binding to AAG than human serum albumin (HSA), and slow intrinsic clearance] to explain the key properties of vismodegib PKs: nonlinearity, long half-life, and a strong correlation between total vismodegib and AAG concentrations. AAG and HSA were both incorporated with fast equilibrium binding to vismodegib. AAG binding is low capacity and high affinity, while HSA binding is high capacity and low affinity. As shown in Figure 5A and B, the previous PK model prospectively predicted the decrease in total vismodegib concentration with TIW and QW dosing and the larger decrease for unbound as compared with total concentrations with these alternative regimens, prior to observation of actual PK results from the current study. The decline in unbound steady-state vismodegib concentrations between day 57 of the maintenance phase and day 15 of the loading phase was greater than that in total concentrations, with an observed intrasubject decrease (mean ± SD) of 50% ± 20% and 24% ± 18%, respectively, for unbound and total concentrations with the TIW regimen and 80% ± 7% and 46% ± 7%, respectively, for unbound and total concentrations with the QW regimen.

The mechanistic PK model was extended to explore the effect of lower once daily vismodegib doses on total and unbound concentrations. As shown in Figure 6A, simulated total vismodegib concentrations decreased in a dose-related fashion from 150 to 25 mg. Unbound vismodegib levels were predicted to decrease more than the total concentrations at the analogous lower daily doses (Fig. 6B). These results are predicted by the saturable absorption and saturable AAG-binding components in the model. At doses lower than 150 mg, absorption is no longer saturated, but AAG binding is at least partially saturated at steady state (although less saturated than at 150 mg). In addition, due to saturated absorption and protein binding, no further increases in total or unbound vismodegib concentrations were predicted for doses greater than 150 mg, which is in agreement with observed phase I results (13).

Vismodegib was generally well tolerated in this phase Ib study with adverse events of similar incidence and severity among dose groups. Twenty-four patients (38.1%) experienced grade 3 to 5 adverse events, which were primarily due to the underlying disease and/or disease progression in this advanced cancer patient population; only 1 event (hyponatremia) in 1 patient was considered related to vismodegib. No reported serious adverse events were considered related to vismodegib. These findings confirm the favorable safety profile of vismodegib after 150 mg QD dosing as no dose-limiting toxicities are observed with this dosage regimen based on findings from the current and previous studies (13).

Due to the long single dose half-life and the nonlinearity in vismodegib PKs after daily dosing, this study was designed to determine whether less frequent dosing could result in similar steady-state levels of total and unbound vismodegib to those obtained with daily dosing. TIW and QW dosing schedules were chosen based upon the single dose half-life and because each represented a potentially suitable alternative to QD dosing from the perspective of patient compliance. The PK profiles of vismodegib in patients after a single 150 mg dose of vismodegib and after once daily dosing with 150 mg vismodegib were consistent with previous findings (13). Both total and unbound plasma vismodegib concentrations reached mean peak levels at 48 hours that were maintained through 72 hours after a single dose of vismodegib, and steady state was achieved within 11 days of continuous once daily dosing. When the dosing schedule was altered to TIW or QW, total plasma vismodegib levels decreased compared with QD dosing. The magnitude of the observed change in total concentrations was less than dose proportional, consistent with nonlinear PKs. The decrease in unbound plasma vismodegib concentrations with less frequent dosing was more pronounced than for total drug and was proportional to the decrease in weekly dose, suggestive of linear PK of unbound vismodegib.

Unbound steady-state vismodegib concentrations were 60% and 85% lower for the TIW and QW dose groups, respectively, relative to the QD dose group. Such decreases may be associated with loss of vismodegib activity based on findings from preclinical models. Integrated PK/PD modeling of vismodegib in xenograft models has revealed a steep relationship between pathway modulation (Gli1 inhibition) and antitumor effect, suggesting that even small reductions in exposure could lead to dramatic loss in vismodegib activity (15). Furthermore, the 150 mg QD regimen has shown activity in patients with advanced BCC (13). Integration of the preclinical concentration–response relationship, human efficacy data, and PK results of this dose scheduling study in cancer patients suggests that patients receiving 150 mg vismodegib on a TIW or QW regimen could be at risk of not achieving unbound drug concentrations associated with maximum clinical benefit.

There is no evidence that a dose greater than 150 mg daily would provide additional efficacy, especially considering that exposure does not increase with higher vismodegib doses (13). Our mechanistic PK model based on phase I data accurately describes the atypical PK of vismodegib (18). The predictability of this model was further supported by the prospective simulation of total and unbound plasma vismodegib PK profiles which were consistent with observations for different dosing schedules in the current study. The intermittent dosing schedules with 150 mg is somewhat representative of PK from lower dose levels with a daily schedule (less daily input and lower plasma concentration range), which provides confidence for using the mechanistic PK model to predict plasma vismodegib concentrations for once daily doses less than 150 mg. Overall, the simulation results indicate that, while total drug concentrations decrease relative to 150 mg QD at steady state, there is a more pronounced impact on unbound concentrations, similar to observations in the current study with less frequent dosing and therefore lower weekly doses. Unbound vismodegib concentrations change in proportion to dose at daily doses lower than 150 mg. For example, a daily dose of 75 mg QD would yield steady-state unbound plasma concentrations that are 47% lower than those achieved with 150 mg QD, consistent with linear PK for unbound drug.

In some cases, unwanted drug effects (e.g., toxicities) can be ameliorated by decreasing drug exposure. The most frequently reported adverse events that seem to be associated with vismodegib seem to be on-target toxicities (dysgeusia, alopecia, muscle spasms, fatigue, and nausea). As the adverse events and efficacy associated with vismodegib are probably a result of interaction with the same pathway, decreasing vismodegib exposure to avoid adverse events by lower or less frequent dose regimens may have a negative impact on efficacy.

Serum AAG concentrations remained relatively constant in all dose groups during the course of the study, thus maintaining the even stratification of AAG levels among dose groups and thereby minimizing any potential influence of AAG fluctuations on the extent of protein binding of vismodegib. As previously observed, total vismodegib and AAG concentrations were strongly correlated with QD dosing in this study (18). This strong correlation was maintained with the TIW dosing schedule. However, a weaker correlation between total vismodegib and AAG levels was observed for the QW dosing schedule, which may be due to less saturated AAG binding compared with more frequent dosing. The large decreases in unbound vismodegib concentrations after TIW and QW dosing are also consistent with a decreased saturation of AAG binding. The weaker correlation between total vismodegib and AAG with an intermittent dosing schedule (relative to QD), and the large decrease of unbound concentration were predicted by our mechanistic PK model after incorporation of saturable AAG binding. The apparent time dependent PK with continuous once daily vismodegib dosing is explained by the change in fraction unbound, thereby resulting in concentration-dependent (nonlinear) PK that is likely responsible for the increase in vismodegib clearance after repeated dosing. Therefore, the half-life of vismodegib is shorter after multiple dosing than after a single dose, and steady-state drug concentrations are attained earlier and are lower than expected. Although not necessarily similar mechanistically, this phenomenon has been observed for other compounds, particularly those with a long single-dose half-life, resulting in an effective or operational half-life that is considerably shorter than the measured half-life after a single dose (19).

Vismodegib fraction unbound is believed to be dependent on AAG saturation which occurs as the vismodegib concentration increases with repeated daily dosing, resulting in establishment of equilibrium with HSA (17). Fraction unbound remains low due to binding to HSA which serves as a low affinity, high capacity drug-binding protein relative to AAG due to its high level in plasma (Supplementary Fig. S3 in the Supplementary Appendix).

In conclusion, the results of this study provide important insights on the concentration-dependent changes in the PKs of vismodegib and validate a previously developed mechanistic PK model. The 150 mg vismodegib QD schedule has been generally well tolerated to date and shown to be active in patients with advanced BCC. Any reduction in vismodegib dose or frequency of dosing could put patients at risk of suboptimal exposure of unbound drug. Therefore, these findings support the use of a 150 mg once daily vismodegib regimen for future clinical trials.

No potential conflicts of interest were disclosed.

We thank the patients for enrolling in this study and taking valuable time away from their personal lives to advance science. We also thank Drs. Bert Lum, Josina Reddy, and Howard Mackey for their important contributions to the study design, Dr. Marie Borin for technical contributions and assistance with the preparation of this manuscript, Dr. Alan Deng for his bioanalytical contributions, Ms. Eileen Nash and Ms. Lisa Nelson Beckel for their work in clinical operations, and the clinical site staff who all worked so diligently to ensure quality conduct of this study. Their support is gratefully acknowledged.

Vismodegib was discovered by Genentech and was jointly validated through a series of preclinical studies performed under a collaborative agreement between Genentech, Inc. (South San Francisco, CA) and Curis, Inc. (Lexington, MA).

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.