Phase 2 randomized study of enzastaurin (LY317615) for lung cancer prevention in former smokers




Chemoprevention for lung cancer with nutraceutical or anti-inflammatory agents has had mixed clinical benefit. Novel targeted agents hold the promise of greater efficacy and selectivity. The authors of this report evaluated enzastaurin, a selective protein kinase C-β (PKC-β) inhibitor with antiproliferative and proapoptotic properties, in former smokers.


The primary objective of this study was to compare the average fraction of Ki-67–stained cells (the Ki-67 labeling index [LI]) in bronchial biopsy specimens that were collected before and after treatment. Participants were randomized (2:1) to receive either 6 months of daily oral enzastaurin (500 mg) or placebo. Stratification was based on morphology, history of lung cancer, and airway obstruction.


In pretrial investigations, the rationale for PKC-β inhibition and pathway interrogation was established in premalignant lesions and early stage lung cancer. In an intent-to-treat analysis, of 40 randomized participants, there was no significant difference in the pretreatment/post-treatment change in the Ki-67 LI between the enzastaurin group and the placebo group (P = .53). Six participants discontinued enzastaurin, including 4 participants who had adverse events, including abdominal distension, deep vein thrombosis, hyponatremia, and rash, and 2 participants who decided to discontinue. One participant in the placebo group was discontinued on the study because of noncompliance. Two participants had ≥1 serious adverse event (bradycardia, deep vein thrombosis, and hypotension).


To the authors' knowledge, this represents the first chemoprevention trial with a non-US Food and Drug Administration-approved, oral, small-molecule–targeted agent. Although the primary endpoint was not met, enzastaurin was tolerable for 6 months by 75% of participants, and there was a suggestion of response in a subset analysis that was restricted to those who had metaplastic or dysplastic lesions. Cancer 2013. © 2012 American Cancer Society.


Lung cancer accounted for 1.4 million deaths world wide in 2008.1 Because of low cure rates (approximately 15%), there is an urgent need to develop compounds that can reduce the disease incidence. The most powerful step in lung cancer prevention is avoidance of carcinogen exposure through smoking cessation. However, individuals who have smoked in the past and have successfully quit still have a higher risk of developing lung cancer than those who are lifetime never smokers.2

Chemoprevention, defined as the use of specific agents to reverse, suppress, or prevent the development of cancer and a means of reducing cancer incidence and mortality, has been successful for colorectal cancer, breast cancer, and head and neck cancer.3,4 In lung cancer, several compounds have been tested, but trials have been either negative or, in the case of those evaluating β-carotene in smokers, deleterious.5–10 Because of the large number of required participants and long interventions and follow-up times, trials that use lung cancer incidence as a primary endpoint pose a formidable challenge from a logistic and financial perspective. To our knowledge, no prior lung cancer chemoprevention studies have investigated a non-US Food and Drug Administration (FDA)-approved, small-molecule–targeted agent.

To allow for testing of an increased number of promising agents over a short period, the use of surrogate endpoint biomarkers has been explored as an alternative to cancer incidence. These include premalignancy evaluated by morphologic criteria and proliferative markers evaluated by immunohistochemistry (IHC). The best described proliferation marker is Ki-67, which can be identified in normal and premalignant lesions of the lung. A high fraction of tumor cells that are stained positive for Ki-67 correlates with decreased survival in patients with nonsmall cell lung cancer (NSCLC).11,12 Enzastaurin hydrochloride is a well tolerated, minimally toxic, oral serine/threonine kinase inhibitor. Its main biologic effect is on inhibition of the phosphorylation of proteins involved in cell proliferation and survival. Specifically, it inhibits activation of protein kinase C-β2 (PKC-β2) in vivo at doses that are achieved with single, oral daily dosing.13–15 The role of PKC in tumorigenesis, through its activation by tumor-promoting phorbol esters, was first described 30 years ago16–18and, in mouse models, studies have linked PKC to carcinogenesis.19–21 In lung cancer cells, enzastaurin demonstrated inhibitory activity and modulation of intracellular signaling proteins.13,21,22 In cell line and nude mouse xenograft studies, enzastaurin treatment suppressed glycogen synthase kinase 3β (GSK3β) phosphorylation and reduced subcutaneous tumor formation. Because of its molecular mechanism of action and low side-effect profile, this drug is an excellent candidate for chemoprevention in high-risk individuals.

We first assessed the expression of phosphorylated PKC-β2 (pPKC-β2) and GSK3β (pGSK3β) in a small cohort of individuals who had bronchial abnormalities at a 3-month interval and in a large panel of early stage lung cancers, and we observed that expression levels remained stable over time and were higher in smaller tumors. Then, we initiated a phase 2 biomarker-endpoint trial of enzastaurin versus placebo in former smokers who had metaplasia and dysplasia of the bronchoepithelium to test the efficacy of enzastaurin as a chemopreventive agent. This trial is registered as a national clinical trial (clinicaltrials. gov identifier: NCT00414960.


Biomarker Evaluation in Bronchial Abnormalities and Early Stage Lung Cancer

Metaplastic (N = 18) and dysplastic (N = 6) bronchoepithelial specimens were collected from a cohort of 14 individuals who were at risk for lung cancer with repeat sampling after approximately 3 months of observation in 7 individuals, which has been described elsewhere.23 We had also constructed a tissue microarray from 237 surgical resection specimens from unique patients with NSCLC that included 85 specimens from patients with stage IA completely resected NSCLC and 102 specimens from patients with stage IB completely resected NSCLC who did not receive perioperative chemotherapy or radiation.24 Standard IHC and automated quantitative in situ fluorescence-based analysis (AQUA), which allows for the accurate and quantitative measurement of proteins in routinely processed, formalin-fixed, paraffin-embedded (FFPE) specimens, of pPKC-β2, total PKC-β2, pGSK3β, Ki-67, and cleaved caspase 3, and digoxigenin-tagged DNA fragments as markers of apoptosis were performed as previously described24,25 using rabbit polyclonal anti-pPKC-β2, anti-pGSK3β, and anticleaved caspase 3 (catalog no. 9371, 9336, and 9661; Cell Signaling Technology, Beverly, Mass), rabbit polyclonal anti-PKC-β2 (catalog no. SC-210; Santa Cruz Biotechnology, Santa Cruz, Calif), mouse monoclonal Mib1 anti-Ki-67 (catalog no. M7240; Dako, Carpenteria, Calif), and the Apoptag assay (catalog no. S7100; Chemicon, Temecula, Calif). IHC scoring included the percentage of positively stained tumor cells (0%-100%), with staining intensity scores that ranged from 1 to 3 (weak, moderate, strong) and a composite score that ranged from 0 to 300 (reflecting the percentage of positively stained cells multiplied by the intensity score). AQUA scores ranged from 0 to 255.

Chemoprevention Study Population

This was a single-institution, phase 2, randomized, double-blind, placebo-controlled study comparing the chemopreventive effect of enzastaurin in individuals at risk for lung cancer with a 6-month intervention and a Ki-67 biomarker endpoint (NCT00414960). The Institutional Review Board approved the protocol, all amendments, and the informed consent document before participant enrollment; and the study complied with the Declaration of Helsinki. Participants were recruited from lung cancer databases at the Moffitt Cancer Center. The following key inclusion criteria were used: aged ≥45 years, a ≥30 pack-year history of smoking, quit smoking ≥1 year before study entry, metaplasia or dysplasia on at least 1 bronchoscopy specimen, an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1, negative cotinine status by urine dipstick, and adequate organ function, including bone marrow reserve.

Key exclusion criteria included a prior history of malignancy in the past 5 years (except nonmelanoma skin cancer, localized prostate cancer with definitive therapy but no history of hormone therapy, cervical carcinoma in situ, stage I NSCLC ≥12 months postsurgery without evidence of recurrence), current evidence of lung cancer, any prior chemotherapy or hormone therapy for the purpose of cancer treatment, previous radiation to the chest in the past 5 years, significant cardiopulmonary comorbidity, and a history of asthma that required oxygen, inhaled steroids, or bronchodilators. The protocol and all subsequent amendments were approved by the Institutional Review Board.

Study Design and Treatment Plan

After informed consent was obtained, eligible participants underwent pulmonary function tests, a white light and autofluorescence bronchoscopy with brushing at 1 site (main carina) and biopsy of at least 3 prespecified locations (main carina, left upper lobe, and right upper lobe). These sites where chosen because bronchial abnormalities are most frequently discovered in these locations. Additional sites were biopsied based on their visual appearance. Individuals with metaplasia or dysplasia on at least 1 of these bronchial biopsy specimens were considered eligible. Stratification was performed according to the following criteria: history of stage I NSCLC and >12 months disease free (present vs absent), evidence of airway obstruction (a forced expiratory volume in 1 second/forced vital capacity ratio of <0.7 vs >0.7), and pretreatment bronchial biopsy histology (presence vs absence of dysplasia). Randomization in a 2:1 ratio of enzastaurin versus placebo was done through a telephone registration system.

The study medication was given once daily as 4 oral tablets (each active tablet contained 125 mg enzastaurin, for a total of 500 mg daily) or identical-appearing placebo. During treatment, individuals were followed bimonthly by visits to the treatment center along with complete blood counts, metabolic panels, and 12-lead electrocardiograms. After a 6-month treatment, a bronchoscopy with brushings and biopsies, a focused history and physical examination, a study medication assessment, and blood collection were performed. Adverse events (AE) were collected using Common Terminology Criteria for Adverse Events version 3.0.

All biopsies were fixed in formalin, embedded in paraffin, and stained sequentially for hematoxylin and eosin and Ki-67 (mouse monoclonal antibody Mib-1 at 1:100 dilution; Dako). The biopsies were graded according to the World Health Organization classification by a board-certified cytopathologist who was blinded to the treatment assignments. The number of bronchoepithelial cells that were stained positive for Ki-67 were enumerated and divided by the total number of cells in each specimen, referred to as the labeling index (LI). The average Ki-67 LI for all histologic specimens collected was then calculated for each participant before and after treatment as prespecified in the protocol. A positive Ki-67 LI prespecified response was defined as a relative decrease >50% in the Ki-67 LI in all metaplastic/dysplastic lesions from 1 participant post-treatment versus pretreatment. The prespecified secondary biomarker endpoints included apoptosis assessment by detection of cleaved caspase 3 using IHC (rabbit polyclonal antiserum 9661; 1:50 dilution; Cell Signaling Technology) and downstream target inhibition by detection of pGSK3β (rabbit polyclonal antiserum 9336; 1:25 dilution; Cell Signaling Technology) using AQUA.24,26

Statistical Methods

The primary objective was to compare the difference between the average Ki-67 LI in all bronchial biopsy specimens collected before and after treatment per participant between the enzastaurin group and the placebo group. The secondary objectives were to compare the treatment-emergent AEs (TEAEs) between the enzastaurin and placebo arms in this patient population. The prespecified exploratory objectives included assessment of target inhibition, induction of apoptosis, and measurement of metaplasia and dysplasia by light microscopy.

The intent of the study was to enroll a total of 186 participants, which, expecting an attrition rate of approximately 25%, would allow us to evaluate 138 participants, including 92 in the enzastaurin arm and 46 in the placebo arm. We assumed an effect size of at least 0.51 standard deviations for a mean change in Ki-67. With a total of 138 evaluable participants, there was 80% power for a test at a 2-sided level of .05. An interim safety analysis was planned after randomizing 40 individuals. A second interim analysis was planned at 50% of the information period to allow for potential sample size adjustments based on interim results. A subsequent amendment also allowed for an efficacy analysis after randomizing 40 individuals, which would allow for study continuation only if profound positive results were observed in the 40 individuals.

Spearman correlations were used to assess associations between average pre-treatment and post-treatment scores among markers. Wilcoxon signed-rank tests were used to compare scores before and after treatment between the groups. A t test for paired data was used to evaluate change between scores before and after treatment. All P values < .05 were considered statistically significant. The analyses were performed using the SAS statistical software package (version 9.2; SAS Institute Inc., Cary, NC).


Pretrial Biomarker Evaluation in Bronchial Abnormalities and Early Stage Lung Cancer

In premalignant bronchoepithelial specimens, pPKC-β2 was expressed in 17 of 24 (71%) specimens in 5% to 100% of cells, and the staining pattern was cytoplasmic and homogeneous. An analysis of 7 serial specimens that were collected approximately 3 months apart revealed no significant change in the levels of expression (t test for paired data). Because premalignant bronchoepithelial specimens are a scarce resource, we used early stage lung cancer specimens to investigate associations among expression levels of PKC-β2, GSK3β, Ki-67, and apoptosis and relevant clinical characteristics.

We assessed PKC-β2 (Fig. 1), pPKC-β2, pGSK3β, Ki-67, cleaved caspase 3, and digoxigenin-labeled DNA fragment expression in tumor specimens from 187 patients who had completely resected stage I NSCLC using IHC and AQUA. The expression levels of all 6 biomarkers (Table 1) were not associated with participants' age, sex, smoking status, performance status, or tumor histology. The median overall survival for patients who had high pPKC-β2 expression (N = 83; events, 27; censored, 56) versus low pPKC-β2 expression (N = 83; events, 48; censored, 35), analyzed by AQUA and dichotomized using a median of 15.3, was >120 months versus 55.0 months, respectively (95% confidence interval, 34.3-75.6 months; log-rank P = .003). All other biomarkers had no significant association with overall survival. PKC-β2, pPKC-β2, and pGSK3β expression levels were inversely associated with tumor size; ie, small tumors were more likely to express high levels than large tumors (Spearman rank order correlation coefficient: −0.16 [P = .03] for PKC-β2; −0.23 [P = .002] for pPKC-β2; −0.22 [P = .004] for pGSK3β). PKC-β2 was not associated significantly with pPKC-β2 or pGSK3β expression, whereas pPKC-β2 and pGSK3β levels were significantly correlated (Spearman rank order correlation coefficient, 0.29; P ≤ .001).

Figure 1.

Protein kinase C-β2 (PKC-β2) staining and immunohistochemical (IHC) quantification are illustrated. (A, B) These are photomicrographs of specimens from patients who had completely resected stage I nonsmall cell lung cancer with (A) low and (B) high IHC scores.

Table 1. Molecular Marker Expression Characteristics in 187 Stage I Nonsmall Cell Lung Cancers
MarkerPKC-β2pPKC-β2pGSK3βKi-67Cleaved Caspase 3Apoptag
  1. Abbreviations: AQUA, automated quantitative analysis; IHC, immunohistochemistry; pGSK3β, phosphorylated glycogen synthase kinase 3β; PKC-β2, protein kinase C 2 β; pPKC-β2, phosphorylated protein kinase C 2 β.

Evaluable specimens168166186/178185167184
Minimum level00.40/23.8000
Maximum level300206.9285/250.12401820
Median level815.320/58.11010.2

Participant Recruitment

Between December 2007 and March 2010, 1740 potential participants (of 3094 screened) were invited to participate with an introductory letter and at least 3 attempted telephone contacts (Fig. 2). Telephone contact was established with 1184 participants (68%). Of these, 532 individuals (45%) declined, 551 (47%) were ineligible, and 101 (9%) were interested in trial participation. The main plausible reasons for declining were transportation limitations (N = 104), not wanting to ingest a non-FDA-approved medication (N = 45), and not wanting to undergo a bronchoscopy (N = 16). The study closed after completion of the interim analysis, because no major effects were observed. In total, 40 participants were randomized to receive 6-month treatment with either enzastaurin (N = 25) or placebo (N = 15). One participant in the enzastaurin group did not receive treatment after randomization because of a pretreatment cerebrovascular accident.

Figure 2.

This is a Consolidated Standards of Reporting Trials (CONSORT) diagram of the current study.

Participant Characteristics and Drug Exposure

Of the 40 participants who were randomized, most were Caucasian (97%), men (65%), and had a median age of 68 years (range, 52–77 years) (Table 2). Two participants had stage I NSCLC, 3 had dysplasia, and 16 had airway obstruction. Eighteen participants (72%) in the enzastaurin group completed the study; 1 participant never started, and 6 discontinued treatment, 4 because of AEs (abdominal distension, deep vein thrombosis [DVT], hyponatremia, and rash) and 2 because of participant decision. Fourteen participants (93%) in the placebo group completed the study; 1 participant discontinued because of noncompliance.

Table 2. Baseline Participant Demographic and Illness Characteristics: Intent-To-Treat Population
CharacteristicPlacebo Group, N = 15Enzastaurin Group, N = 25Total, N = 40P
  1. Abbreviations: ECOG: Eastern Cooperative Oncology Group; NSCLC: nonsmall cell lung cancer; SD: standard deviation.

Median age [range], y69 [60–74]68 [52–77]68 [52–77].94
Sex   .09
Men7 (47)19 (76)26 (65) 
Women8 (53)6 (24)14 (35) 
Ethnicity   > .99
Caucasian15 (100)24 (96)39 (97) 
Hispanic0 (0)1 (4)1 (2) 
Airway obstruction   .32
Absence11 (73)13 (52)24 (60) 
Presence4 (27)12 (48)16 (40) 
ECOG performance status   > .99
012 (80)19 (76)31 (77) 
13 (20)6 (24)9 (22) 
History of stage I NSCLC1 (7)1 (4)2 (5)>.99
Dysplasia in pretreatment bronchial biopsy0 (0)3 (12)3 (7).28
Smoking pack-year history: Mean ± SD, y58 ± 21.350 ± 21.453 ± 21.4.26

Of 39 participants who received treatment, the mean duration of drug exposure was 165 days (range, 21–199 days), and the mean dose was 476 mg daily (range, 331–554 mg daily) (Table 3). All 39 participants who received treatment used at least 50% of the prescriptions and were considered treatment compliant. Three participants in the enzastaurin group had dose reductions because of lower extremity edema, diarrhea, or QT prolongation on an electrocardiogram.

Table 3. Drug Exposure: Safety Population
VariablePlacebo Group, N = 15Enzastaurin Group, N = 24Total, N = 39
  • a

    This participant discontinued/withdrew from the study because of noncompliance.

  • b

    These participants discontinued/withdrew from the study because of adverse events (n = 4) and participant decision (n = 2).

Mean duration of exposure, d170162165
Mean dose, mg/d486470476
No. of participants with dose adjustment (%)0 (0)3 (12)3 (8)
No. of withdrawals/ discontinuations (%)1 (7)a6 (25)b7(18)

Morphologic Assessment

Thirty-four participants underwent 2 bronchoscopies, and 6 underwent 1 bronchoscopy with biopsies of at least 3 prespecified sites. Of the total 254 specimens collected, normal histology was revealed in 52 specimens, 153 presented metaplasia, 6 presented dysplasia, and 43 were insufficient for morphologic assessment. All 3 participants who had dysplasia before therapy (in 1 of 3 biopsy specimens) were randomized to the enzastaurin arm and had metaplasia identified on the second biopsy. One participant in the placebo group and 2 participants in the enzastaurin group had dysplasia (in 1 of 3 biopsy specimens) at the end of the study.

Ki-67 Analysis

The primary analysis was performed on 2 populations: the intent-to-treat population (Fig. 3A) and the efficacy population (Ki-67 LI measurements from biopsy specimens before and after treatment). For the efficacy population (N = 34), both arms had an increase in the mean Ki-67 LI at the end of the study (pretreatment to post-treatment Ki-67 LI: enzastaurin, 6.7 to 10.4, respectively [P = .16]; placebo, 4.0 to 10.1, respectively [P = .07]), and the changes between both groups were statistically insignificant. In the enzastaurin group, 6 of 21 participants had a decreased Ki-67 LI after treatment (greatest mean change, −27.7%); in the placebo group, 4 of 13 participants had a decreased Ki-67 LI after treatment (greatest mean change, −4.5%). The mean Ki-67 LI change did not differ significantly between the 2 arms (mean difference, −2.4; 95% confidence interval, −10.6 to 5.8; P = .56). Similar results were observed for the intent-to-treat population, and the differences were insignificant (P = .53) between both groups.

Figure 3.

Changes in the Ki-67 labeling index are illustrated. (A) In the intent-to-treat population (N = 40), the study failed to meet the primary endpoint. There was no significant difference in the percentage change in the Ki-67 labeling index between the enzastaurin group and the placebo group. A within-group comparison of the pretreatment and post-treatment samples did not indicate a statistically significant difference. (B) Metaplastic or dysplastic samples from the enzastaurin-treated efficacy population (N = 21) are illustrated. The Ki-67 response rate was 19% (4 of 21 individuals). Of the 3 individuals who had dysplasia at baseline, all had metaplasia post-treatment. (C) Metaplastic or dysplastic samples from the placebo-treated efficacy population (N = 13) are illustrated. The Ki-67 response rate was 0%. One of 12 individuals had a reduction in Ki-67, but this did not meet response criteria, ie, a relative change ≥50%. One individual who had metaplasia at baseline failed Ki-67 staining and, thus, was excluded from the analysis of abnormal specimens.

In analyses that were conducted using only the abnormal specimens (metaplasia or dysplasia; N = 159 [ie, specimens with normal histology were ignored in the analysis]), we observed a similar pattern of results (Fig. 3B, C). Both arms had numerically increased mean Ki-67 values after treatment, and the enzastaurin arm had a numerically smaller increase than the placebo arm. No statistically significant results were observed between or within treatment comparisons. In the enzastaurin group, 4 of 21 participants (19%) had a ≥50% relative decrease in Ki-67 LI in all metaplasia or dysplasia lesions after treatment compared with 0 of 12 participants in the placebo group. However, this difference in the response rate was not statistically significant (P = .27).

Adverse Events

One participant in the enzastaurin group, who had a prior history of DVT and was receiving conjugated estrogen, had 2 DVT events (grade 3) possibly related to study drug and had a full recovery. Another participant who was receiving enzastaurin had 2 nonstudy drug-related AEs—bradycardia (grade 3) and hypotension (grade 3)—after accidentally ingesting the spouse's hypertension medication. Severe AEs were not reported in the placebo group. Twenty-four participants (100%) who were receiving enzastaurin had ≥1 TEAE (grade 1, n = 12; grade 2, n = 7, grade 3, n = 5), whereas 10 participants (66.7%) who were receiving placebo had ≥1 TEAE (grade 1, n = 8; grade 2, n = 2). In the enzastaurin group, a significantly greater percentage of participants had ≥1 TEAE compared with the placebo group (P < .01), the enzastaurin group had a significantly larger percentage of participants with metabolism and nutrition disorders (P = .05). The most frequent TEAEs in the enzastaurin group were hyperkalemia (33%), decreased hemoglobin (25%), hyponatremia (25%), and hyperglycemia (21%) (Table 4). All 32 participants who completed the study had negative cotinine test results at the end of the study.

Table 4. Common Adverse Events (>10% of Participants)
 No. of Participants (%)a
Adverse Event: Maximum GradePlacebo Group, N = 15Enzastaurin Group, N = 24
  • a

    Note that 2 participants in the enzastaurin group (8%) had ≥ 1 serious adverse event (SAE). Two SAEs presented by 1 of these participants were considered possibly related to the study drug. No participant in the placebo group experienced an SAE.

Grade 11 (7)7 (29)
Grade 20 (0)1 (4)
Decreased hemoglobin  
Grade 10 (0)6 (25)
Grade 11 (7)5 (21)
Grade 20 (0)0 (0)
Grade 30 (0)1 (4)
Grade 11 (7)5 (21)
Grade 12 (13)3 (13)
Grade 20 (0)1 (4)
Abnormal urine color  
Grade 10 (0)4 (17)
Prolonged QT  
Grade 11 (7)2 (8)
Grade 20 (0)1 (4)

Apoptosis and Phosphorylated Glycogen Synthase Kinase 3β

Although there were some increases in cleaved caspase 3 staining noted in some specimens, no significant differences before and after treatment were noted in apoptosis (Wilcoxon signed-rank test; P = .68) between the enzastaurin and placebo groups. We observed no significant correlations (all P values were > .05; Spearman correlation) among pre-treatment and post-treatment average caspase and Ki-67 staining scores. Wilcoxon signed-rank tests revealed no significant changes in caspase levels for either the enzastaurin group (P = .61; N = 17) or the placebo group (P = .95; N = 9).

A wide range of pGSK3β staining scores according to AQUA (version 2.1) was observed among the samples. The extreme ranges are highlighted in Figure 4A, B. There was no significant difference in pGSK3β levels between the enzastaurin and placebo groups before and after treatment (Fig. 4C, D), and no significant correlation was observed between pGSK3β and Ki-67.

Figure 4.

Phosphorylated glycogen synthase kinase 3β (pGSK3β) staining and automated quantitative analysis (AQUA) are illustrated. (A, B) These photomicrographs are examples of (A) high AQUA scores and (B) low AQUA scores. (C, D) Pretreatment and post-treatment scatter plots at the log level illustrate staining for pGSK3β among (C) 17 individuals in the enzastaurin group and (D) 9 individuals in the placebo group. Blue indicates nucleus; red, cytokeratin; green, pGSK3β.


To our knowledge, this study represents the first lung cancer chemoprevention trial with a non-FDA-approved, oral, small-molecule–targeted agent. In the intent-to-treat population, the study failed to meet the primary endpoint. There was no significant difference in the change of the Ki-67 LI between the enzastaurin group and the placebo group. A within-group comparison of the pre-treatment and post-treatment samples did not reveal a statistically significant difference. However, in an analysis that was restricted to metaplastic and dysplastic samples only, 19% of participants who were receiving enzastaurin had a reduction >50% in Ki-67 response compared with 0% of those who were receiving placebo.

Studies by Graff et al have demonstrated an up to 90% reduction of pGSK3β in nude mouse xenografts for up to 8 hours after oral enzastaurin dosing and an effect loss after 12 hours.22 At clinically achievable plasma levels, enzastaurin directly interferes with the phosphorylation of several molecules of key importance in carcinogenesis and tumor progression, including protein kinase B (Akt), GSK3β, ribosomal protein S6K, and the transcription factor CREB (cAMP response element-binding); and it has been demonstrated that enzastaurin inhibits migration, metastases, and invasion of NSCLC cells in vivo.27 In addition, our pretrial investigations of PKC-β2, pPKC-β2, and pGSK3β suggested that active signal transduction through PKC-β2 may be more frequent in early stage lung cancer, thus also suggesting its biologic importance in early tumorigenesis. Given the crucial position of PKC-β2 in the regulation of cell growth and survival, the frequent expression of a seemingly intact PKC-β2/GSK3 signal transduction in early stage NSCLC, and the vanishingly low toxicity of the oral PKC inhibitor enzastaurin,28 a phase 2 chemoprevention study in former smokers was initiated.

To assess for target inhibition, pre-treatment and post-treatment tissue specimens were analyzed, and the expected reduction in GSK3β phosphorylation and increase in cleaved caspase 3 were not apparent. This lack of a pharmacodynamic response indicates that apoptosis and antiproliferative effects were not achieved uniformly, thus likely explaining the improved, yet nonsignificant, responses in Ki-67 LI in a limited population—in the abnormal samples only and not across all samples. Drug concentrations may have played a role is these findings. Still, previous data suggest that the prescribed dose of enzastaurin is effective in reducing tissue levels of pGSK3β, and our results also may be explained by the time elapsed from last dosing to tissue collection, which was 24 hours or greater in all but 2 participants. It is noteworthy that the largest reduction in GSK3β phosphorylation was observed in a participant who received enzastaurin within 8 hours of biopsy. This participant also had the largest relative reduction in Ki-67 LI (76%) and also had a 50% increase in the cleaved caspase 3 score. The study design did not specify the precise temporal relation between last dosing and post-treatment biopsy, which future studies should keep in mind, because it is possible that a biologic rebound may obscure a true pharmacodynamic effect. Although the timing of post-treatment biopsies will have to vary from agent to agent, it is imperative to consider the pharmacokinetic characteristics of the specific agent as well as the stability of the specific biomarker to have a true pharmacodynamic assessment.

Another factor that may have played a role in the lack of significant findings includes the possibility that drug concentrations in the current study may not have been high enough to lead to antiproliferative effects in tissues with lower proliferation rates. Our observed reduction in Ki-67 in morphologically abnormal specimens in the enzastaurin group compared with the placebo group suggests that tissues with low proliferation may not be suitable for efficacy evaluation of antiproliferative agents. This was observed previously in a similar trial that compared celecoxib with placebo in former smokers. Those investigators reported that specimens with high pretreatment proliferation rates (Ki-67 LI: ≥2 vs < 2) were more likely to demonstrate reductions post-treatment than those with low pretreatment proliferation rates.29 It is noteworthy that, in our intent-to-treat population, the Ki-67 LI increased in the enzastaurin and placebo groups, although not at a statistically significant rate. These findings, although not previously described in the recently reported trials of celecoxib or iloprost,10,30 may be because of outside events that can influence levels of Ki-67.30, 31 Furthermore, Ki-67 alone may not be sufficiently specific or sensitive for predicting future lung cancer development.31,32

In this study, we were able to establish a recruitment algorithm and to quantify the participant attrition (nearly 99%) for future chemoprevention trials by focusing on a population at risk that had prior familiarity with clinical investigations. We demonstrated that once-daily oral enzastaurin was generally well tolerated and that serial Ki-67 assessments in former smokers were feasible. The biomarker assessments for target inhibition performed in the treatment and placebo groups allowed us to document the natural history of bronchial airway changes in this patient population.

Although our results are clearly affected by the small sample size and do not afford the opportunity to evaluate the potential effects of enzastaurin on the prevention of peripheral lung carcinomas, our accrual rates and histology results in former smokers are comparable to previously published chemoprevention trials. In 2010, Kim et al30 published a celecoxib lung cancer chemoprevention study, which accrued a total of 53 former smokers and 162 current smokers who had normal, metaplastic, or dysplastic changes in their bronchial epithelium during a recruitment period just short of 5 years. In a population in which no minimum bronchial airway changes were required, the metaplasia or dysplasia rate was 5.7% (14 of 248 samples) in material obtained from former smokers. We were able to accrue 40 former smokers during a recruitment period of 2.3 years with a minimum requirement of metaplasia, resulting in a higher metaplasia/dysplasia rate of 62% (157 of 254 samples). In 2011, Keith et al published the results from their 6-year phase 2 study of iloprost versus placebo which accrued 71 former smokers and 81 current smokers who met 1 of following criteria: 1) sputum atypia (30% of former smokers) or 2) bronchial dysplasia. Compared with those studies, in the current study, we used stricter inclusion criteria, such as a ≥30 pack-year history of smoking and a minimum requirement of metaplasia on bronchial samples, to allow the inclusion of only those who had histologic evidence of airway changes. Despite these additional restrictions, participants were successfully accrued. Although cross-trial comparisons carry inherent limitations, it is noteworthy that many similarities in accrual achievements can be observed among these trials. A clear recruitment plan, infrastructure, resources, and experience, and dedication of the investigative team are paramount for a successful accrual to lung cancer chemoprevention trials.

The current findings argue against the further investigation of enzastaurin as a lung cancer chemopreventive agent. Although smoking cessation clearly can reduce lung cancer risk, former smokers represent a large population at risk of developing lung cancer. Hence, there exists the potential for added benefit from a proven chemoprevention agent. For progress to be made in lung cancer chemoprevention, advanced preclinical risk models that better delineate the mechanisms of actions are encouraged. Future trials must include biomarker assessments to afford pharmacodynamic and mechanistic insight into agent activity. Our results highlight the required detail and logistics necessary to optimize the informative value of chosen biomarkers. We conclude that further studies with enzastaurin or similar agents must collect post-treatment specimens while accounting for agent-specific pharmacokinetic and biomarker dynamics and documenting the time interval between last dosing and specimen collection. We further conclude that Ki-67 or other biomarkers of proliferation should be used only in tissues with high baseline levels, such as metaplastic/dysplastic lesions, if the inhibitory activity of an agent is being investigated. Further studies with enzastaurin or similar agents that take these considerations into account are warranted to advance cancer prevention with targeted agents.


We thank Tim Estrella for trial coordination, Bhupendra Rawal for data analysis, and Farah Khalil, MD, for pathology support.


This work was supported by grants from the National Cancer Institute to Dr. Bepler (grants U01-CA10122 and P50-CA119997) and by Eli Lilly and Company.


This trial was partially funded by Eli Lilly and Company.