Clinical and Molecular Responses in Lung Cancer Patients Receiving Romidepsin

Purpose: Our preclinical experiments indicated that Romidepsin (Depsipeptide FK228; DP) mediates growth arrest and apoptosis in cultured lung cancer cells. A phase II trial was done to examine clinical and molecular responses mediated by this histone deacetylase inhibitor in lung cancer patients.

Experimental Design: Nineteen patients with neoplasms refractory to standard therapy received 4-h DP infusions (17.8 mg/m²) on days 1 and 7 of a 21-day cycle. Each full course of therapy consisted of two identical 21-day cycles. Plasma DP levels were evaluated by liquid chromatography–mass spectrometry techniques. A variety of molecular end points were assessed in tumor biopsies via immunohistochemistry techniques. Long oligo arrays were used to examine gene expression profiles in laser-captured tumor cells before and after DP exposure, relative to lung cancer cells and adjacent normal bronchial epithelia from patients undergoing pulmonary resections.

Results: Nineteen patients were evaluable for toxicity assessment; 18 were evaluable for treatment response. Myelosuppression was dose limiting in one individual. No significant cardiac toxicities were observed. Maximum steady-state plasma DP concentrations ranged from 384 to 1,114 ng/mL. No objective responses were observed. Transient stabilization of disease was noted in nine patients. DP enhanced acetylation of histone H4, increased p21 expression in lung cancer cells, and seemed to shift global gene expression profiles in these cells toward those detected in normal bronchial epithelia.

Conclusion: Although exhibiting minimal clinical efficacy at this dose and schedule, DP mediates biological effects that may warrant further evaluation of this histone deacetylase inhibitor in combination with novel-targeted agents in lung cancer patients.

Dynamic patterns of gene expression during embryonic development and cellular homeostasis, as well as stable repression of gene expression associated with X chromosome inactivation and imprinting, are mediated via DNA methylation in conjunction with acetylation, methylation, and phosphorylation of core histone proteins (H2a, H2b, H3, and H4; refs. 1, 2). Perturbation of these tightly linked epigenetic processes during multistep pulmonary carcinogenesis results in global DNA demethylation, derepression of endogenous retroviruses and germ cell–restricted genes, such as BORIS, NY-ESO-1, and MAGE-3, and paradoxical silencing of a variety of tumor suppressor genes, including p16, p14/ARF, TFPI-2, and RASSF1A (3–5).

Whereas considerable literature pertains to mechanisms contributing to aberrant DNA methylation in cancer cells (6), less information is available regarding specific histone modifications that arise during malignant transformation. However, recent studies indicate that unique changes in the histone code occur early during multistep carcinogenesis and contribute significantly to altered chromatin structure and gene expression in cancer cells (7). For example, leukemia and breast cancer cells exhibit loss of monoacetylation at lysine 16, in conjunction with trimethylation of lysine 20 of histone H4 within repetitive DNA sequences (rather than CpG islands of promoter regions) in association with global DNA demethylation (8, 9). Additional studies indicate that expression profiles of histone-modifying genes may correlate with tumor histology and prognosis in cancer patients. For instance, HDAC-4, EZH-2, and CREBBP are significantly overexpressed in breast cancers, but not bladder or colorectal carcinomas, whereas SUV39H1, histone deacetylase-1 (HDAC-1), HDAC-5, HDAC-7A, and SIRT-1 are overexpressed in colorectal carcinomas, but not bladder or breast cancers (10). Although a signature of histone-modifying genes in lung cancer cells has not been described, overexpression of HDAC-1 and decreased expression of HDAC-5 and HDAC-10 seem to correlate with advanced stage of disease and adverse outcome in lung cancer patients (11, 12).

The recent delineation of the role of histone acetylation in mediating DNA methylation, chromatin structure, and gene expression provides the rationale for the use of HDAC inhibitors for lung cancer therapy (13). Our preclinical experiments have shown that Romidepsin (Depsipeptide FK228; DP) markedly enhances 5-aza-2′ deoxycytidine–mediated gene induction and apoptosis in cultured lung cancer cells, but not in normal bronchial epithelia (4, 14, 15). Furthermore, we have reported that DP-mediated growth arrest and apoptosis coincides with acetylation of heat shock protein 90 and subsequent inhibition of survival signaling in lung cancer cells (16, 17). Recently, we reported results of phase I evaluation of gene induction mediated by 5-aza-2′ deoxycytidine infusion in patients with primary thoracic malignancies (18). This phase II study was undertaken to assess clinical and molecular responses in lung cancer patients receiving DP therapy as a prelude to a sequential 5-aza-2′ deoxycytidine/DP protocol. Herein, we present data, which show that DP can modulate chromatin structure and gene expression in primary lung cancers. These findings, together with our published laboratory data (17, 19), support further evaluation of HDAC inhibitors in combination with agents targeting survival signaling and/or death receptors in lung cancer patients.

Trial design. The trial was originally designed as a two-group study with separate enrollment based on small cell versus non–small cell histology. In each group, a Simon two-stage optimal design (20) was to be used to rule out a 5% null response rate (P₀ = 0.05) and to target a 30% response rate (P₁ = 0.30). With α = 0.05 (the probability of accepting a poor agent) and β = 0.10 (the probability of rejecting a good agent), nine patients were to be initially enrolled in the first stage (20). If no patients exhibited a clinical response or disease stabilization, then accrual would stop at that point for that arm. If at least one patient experienced stable disease or a true clinical response, then accrual would continue until a total of 17 patients were enrolled with that disease type. If at least three responses in 17 evaluable patients were noted, the treatment would be considered worthy of further investigation based on clinical outcome. Due to extremely limited accrual of small cell lung cancer patients meeting eligibility requirements, the protocol was amended to a single-arm trial, again using a Simon two-stage design for phase II studies with the same characteristics and a total requirement of 17 patients.

Treatment regimen. DP was supplied by Gloucester Pharmaceuticals through the Cancer Therapy Evaluation Program at the National Cancer Institute (NCI). Lyophilized drug was reconstituted in 2 mL of 20% ethanol in propylene glycol, and then further diluted in 0.9% normal saline to a final concentration of 0.04 mg/mL. DP was given at a dose of 17.8 mg/m² by continuous central i.v. infusion over 4 h on days 1 and 7 of each treatment cycle, which lasted ∼21 days. Treatment evaluation was done after two cycles.

Eligibility and response criteria. Patients (ages 18 years or older) with histologically or cytologically proved primary lung cancer were eligible for evaluation. All patients had Eastern Cooperative Oncology Group performance status of 0 to 2 and had not received chemotherapy, biological therapy, or radiation to target lesions within 30 days of commencing DP treatment. All patients had measurable disease; those individuals who had received radiation therapy had a measurable lesion outside radiation portals or had evidence of active disease in the radiation field. Additional eligibility criteria included the following: FEV1 and DLCO > 30% predicted value; pCO₂ < 50 mm Hg, pO₂ > 60 mm Hg on room air ABG; no evidence of decompensated coronary artery disease; platelet count > 100,000; Hgb > 10 g/dL; WBC > 3500/μL; normal prothrombin time; total bilirubin < 1.5 times upper limits of normal; and serum creatinine < 1.6 mg/mL.

Individuals with carcinomas that could not be readily biopsied by bronchoscopic or percutaneous fine needle aspirates (FNA) techniques, as well as patients with untreated limited stage of small cell lung cancer or operable non–small cell lung cancer, active intracranial or leptomeningeal metastases, or those requiring anticonvulsant medications were excluded from the study. In addition, patients with prior DP or mitomycin C exposure, as well as individuals who had received more than two prior cytotoxic chemotherapy regimens, were excluded from the study. Individuals with unstable angina (LVEF < 40%) or uncontrolled cardiac dysrhythmias, as well as patients with pulmonary embolism within 6 months of protocol enrollment or deep venous thrombosis requiring anticoagulation, were excluded from the study. Patients with life expectancy of <6 months were also excluded from the study. RECIST criteria (21) were used to determine clinical response after two cycles of therapy. Patients who exhibited treatment response evidenced by either stabilization or regression of disease after two cycles of therapy were eligible to receive two additional cycles of DP. Treatment continued in this manner until off-study criteria had been met.

Toxicity, treatment modification, and off-study criteria. Grade 4 or greater hematologic toxicities exceeding 5-day duration, or grade 3 or greater nonhematologic toxicities, excluding alopecia, as assessed by Common Toxicity Criteria for Adverse Events version 2 during cycle 1 of therapy were considered dose limiting. Patients who exhibited response to treatment yet experienced reversible dose-limiting toxicity were eligible for additional treatment after dose reduction to 15 mg/m². All patients who exhibited disease progression or experienced either irreversible or unacceptable dose-limiting toxicities despite dose modification were removed from the study. Similarly, individuals who withdrew voluntarily for any reason were removed from the study.

Pharmacokinetic studies. Plasma DP levels were analyzed 0, 2, 4, 4.5, 6, 9, and 28 h after the start of the first infusion during the first two cycles in all patients. For all subsequent infusions, two samples were obtained: one before and one 4 h after commencing DP infusion. Blood samples (5 mL each) were collected in heparinized tubes, immediately transferred to the laboratory on ice, and centrifuged at 2000×g for 15 min; plasma samples were aliquoted and stored at −80°C. DP concentrations in plasma were quantitated by liquid chromatography with single-quadrupole mass spectrometric detection over the concentration range of 0.5 to 100 ng/mL according to a validated, published procedure (22, 23). The values for precision and accuracy were ≤7.88% and <3.33% relative error, respectively. Estimates of pharmacokinetic variables for DP were derived from individual concentration-time data sets using independent model methods, as implemented in the computer software program WinNonlin v4.0 (Pharsight Corporation). The a priori level of significance was set at 0.05. The interpatient variability in clearance was assessed as the ratio of the mean and SD and expressed as a percentage. The intrapatient variability in clearance was assessed by evaluating DP pharmacokinetics on days 1 and 7 by calculating the ratio of drug clearance and the difference of this value relative to a hypothetical value of 1, using the statistical software program NCSS (version 2001).

Tissue acquisition and molecular analyses. Biopsies of target lesions were obtained before commencing DP therapy and on day 8 (∼24 h after completion of the second DP infusion) during the first treatment cycle using endoscopic or CT-directed percutaneous FNA techniques. For each patient, the same lesion was biopsied before and after DP infusion. Considerable efforts were made to biopsy the same region of each tumor to minimize sampling artifact. Immediate cytopathologic review confirmed the presence of tumor cells in all biopsy specimens. Immunohistochemistry methods were used to evaluate expression of NY-ESO-1, MAGE-3, p16, acetylated histone H4, p21, Ki67, phosphorylated extracellular signal-regulated kinase (phospho-ERK), and cleaved caspase-3 protein levels in formalin-fixed, paraffin-embedded biopsy tissues. Techniques for analysis of NY-ESO-1, MAGE-3, p16, acetylated histone H4, and p21 expression in biopsy tissues have been previously described (18, 24, 25). For analysis of Ki-67, phospho-ERK, and cleaved caspase-3, tissue sections were incubated in citrate buffer (pH 6.0; DAKO) for 1 h in a water bath at 95°C for antigen retrieval. Sections were washed with TBS (pH 7.6) and then rinsed in TBS for 5 min at room temperature. Samples were preincubated with blocking solution (1% bovine serum albumin in TBS) for 30 min at room temperature. Thereafter, sections were incubated for 1 h in a humid chamber at room temperature with the antibodies recognizing Ki-67 (DAKO), phospho-ERK (Cell Signaling, Boston, MA), or cleaved caspase-3 (Cell Signaling). After washing in TBS, samples were incubated with labeled polymer horseradish peroxidase (DAKO) for 30 min at room temperature, rinsed in TBS, and incubated in 3,3′-diaminobenzidine substrate buffer (DAKO) for 5 to 10 min at room temperature. Specimens were rinsed in H₂O and then counterstained with hematoxylin. The slides were then mounted in aqueous mounting media (Aquatex, Merck). Nonneoplastic lung tissues from nine different subjects were used as internal controls. Negative controls for the antisera consisted of sections stained with an isotype-matched nonimmune antibody at the same protein concentration (DAKO).

All antibodies were titrated on appropriate cell line controls with expression levels of targets confirmed by quantitative reverse transcription–PCR, Western blot, and ApoBrdUrd techniques. Pre- and posttreatment samples were analyzed by a pathologist (M.R.F.) in a blinded manner using the following criteria for scoring slides based on the percentage of positive cells: 0, all tumor cells are negative for the protein in question; 1+, 1% to 24% of tumor cells are positive; 2+, 25% to 49% of tumor cells are positive; 3+, 50% to 74% of tumor cells are positive; and 4+, 75% or more tumor cells are positive. In addition, the intensity of the staining was scored as follows: −1, decreased intensity; 1+, equal intensity; and 2+, increased intensity of posttreatment relative to pretreatment staining. The percent positive and intensity of staining scores were multiplied to give a final number. Similar criteria have been used for evaluation of molecular end points in biopsy materials and have been shown to yield results comparable with those obtained by computerized image analysis techniques (26, 27).

Laser capture microdissection, RNA amplification, and microarray analysis. Five sets of pre- and posttreatment tumor biopsies from four patients were processed for laser capture microdissection and comprehensive gene expression profiling using microarray techniques. Briefly, FNAs were immediately frozen in ornithine carbamyl transferase compound on dry ice. Primary tumor tissues and histologically adjacent normal lung parenchyma from eight lung cancer patients undergoing definitive resections were processed in a similar manner. Serial 8-μm thick, frozen sections were prepared from each biopsy specimen. Laser capture microdissection was done using the PixCell IIe laser capture microdissection system (Acturus) according to the manufacturer's protocol with modifications, as recently described (18).

Total RNA was amplified into antisense RNA via the amino allyl MessageAMP antisense RNA kit. Test and reference antisense RNAs were labeled with Cy5 and Cy3 and cohybridized to custom-made 23K long oligo microarrays, printed at the NCI with the Human Operon Version 2.0 genome oligo set. Hybridizations were done at 60°C overnight (10-16 h). Thereafter, the slides were washed for 2 min in 2× SSC with 0.1% SDS, 1× SSC, and 0.2× SSC, respectively, and spun dry at 100×g for 10 min. Fluorescence images were captured using a Genepix 4000B (Axon Instruments) scanner. Fluorescence intensities were normalized at 50% median ratio value and filtered at criteria of ≥200 (based on a scale of 1-68,000 units) and spot sizes of ≥50 μm.

Data acquisition and analysis. Acquisition and initial quantification of array images were done using the GenePix Pro 6.0 (Axon Instruments). Subsequent data analysis was done using GeneSpring GX 7.3.1 (Silicon Genetics). Data were filtered initially by removing genes with signal intensity of lower or equal to the “median intensity of randomized negative controls + 1 SD” across all samples. The filtered data were then entered into GeneSpring GX 7.3.1 and normalized to its median intensity per sample. This step converts the raw expression value into a relative value within each sample and, thus, made the samples comparable. The data for each gene relative to its median intensity were normalized across samples, which optimized data visualization; these procedures traditionally have been called “per chip” and “per gene” normalizations. The thresholds for selecting potentially significant genes were set at a relative difference of >2-fold and/or statistical difference at P < 0.05 (parametric test was done assuming unequal variances). Those genes, which met these thresholds, were considered potentially significant.

Patient accrual, toxicities, and clinical response. The primary end points of this phase II trial included examination of clinical responses and toxicities in lung cancer patients mediated by DP, given at a dose of 17.8 mg/m² by 4-h central venous infusions on days 1 and 7 every 3 weeks. This dose and schedule of DP administration were chosen for the study based on the mean therapeutic dose defined during phase I evaluation of DP given by 4-h infusion on days 1 and 5 of a 21-day schedule, and Cancer Therapy Evaluation Program (CTEP) requests to use a 4-h infusion weekly thrice every 4 weeks for phase II evaluation of DP in patients with solid tumors. Because we were intending to combine this regimen in a sequential manner with Decitabine given via continuous 72-h infusion, CTEP allowed us to modify the recommended phase II schema so that we would not encounter overlapping hematologic toxicities secondary to Decitabine and DP as we developed our gene induction protocols. Secondary end points included evaluation of NY-ESO-1 and p16 induction, epidermal growth factor receptor–mediated signaling and apoptosis, and acetylated histone H4 and p21 expression levels in primary lung cancer cells exposed to DP. Additional end points included refinement of laser capture microdissection and microarray techniques for global gene expression profiling of lung cancer cells exposed to chromatin remodeling agents. Although, based on our preclinical data, it was not expected that DP would mediate robust induction of cancer-testis antigens or tumor suppressor genes that were repressed in lung cancer cells, the analyses in this protocol were intended to establish molecular responses to DP as a prelude to a sequential Decitabine/DP trial.

Nineteen patients were enrolled on this phase II trial. One had an equivocal lung cancer diagnosis after receiving DP and was not evaluable. With permission from CTEP, an additional patient beyond the 17 allowable individuals was accrued to trial due to uncertainty that a previously treated patient would be evaluable for treatment response. At the time of protocol entry, 16 patients were diagnosed with primary non–small cell lung cancer, whereas three individuals had small cell lung cancer. All but one patient had a history of tobacco abuse. Sixteen patients had received prior chemotherapy, three of whom had also received a treatment regimen using only molecular targeted agents. Approximately, 50% of the individuals had also received radiation to primary or metastatic sites of disease (Table 1). The median age (50.5 years; range, 41-77 years) and genders of the patients (11 males and 8 females), as well as the number of prior therapies, reflected the histologies and stages of lung cancers targeted in this study and the experimental nature of the protocol.

No objective responses were observed in 18 (15 non–small cell lung cancer and 3 small cell lung cancer) evaluable patients. Stabilization of disease after two cycles of DP was observed in 10 individuals, eight of whom continued therapy. One patient withdrew from the study after the third cycle of DP. Three patients experienced disease progression after two additional cycles of treatment. One patient with stable disease after four cycles of DP was removed from the study due to incapacitating pain from a preexisting lumbar spine compression fracture. Three patients received six cycles of DP before exhibiting progression of disease. The median number of cycles received was four. Data pertaining to clinical responses are summarized in Table 1.

All 19 patients enrolled on this study were evaluable for toxicities (summarized in Table 2). Dose-limiting myelosuppression was observed in one small cell lung cancer patient who had received two prior cytotoxic chemotherapy regimens and mediastinal radiation. Grade 3 anemia was observed in four individuals, grade 3/4 neutropenia was seen in four patients, and grade 4 (non–dose limiting) thrombocytopenia was observed in one patient. Grade 3 hypoxia was observed in three patients, which seemed attributable primarily to disease progression (Table 2A). During a later cycle, an additional patient developed transient right upper lobe pneumonitis shortly after DP infusion, suggestive of radiation recall. Grade 3 tumor pain was observed in one patient exhibiting stable disease. One patient developed an axillary vein thrombosis after administration of DP via PIC line, which resolved after direct administration of 12-O-tetradecanoylphorbol-13-acetate. Two additional patients developed cellulitis at their PIC line insertion sites (not associated with myelosuppression), requiring removal of their lines (Table 2B).

Cardiac events associated with DP infusion. Because preclinical data suggested that DP might cause myocardial injury (28), considerable monitoring was done for all patients accrued to this trial. Participants underwent initial evaluation with EKG; individuals with abnormal baseline EKGs underwent ECHO or radionuclide scans to verify absence of cardiac dysrhythmia or ischemia and to confirm that EF is >40%, as well as normal wall motion. All patients were monitored in the intensive care unit for 24 h during DP infusions in the first treatment cycle. Additional EKGs were obtained on days 2, 7, 8, and 21 of each treatment cycle; at the request of the CTEP, more frequent EKGs were obtained in nearly half of the patients accrued during the latter phase of the study. In all, ∼460 EKGs were obtained in association with 120 DP infusions during the course of the trial. Correlative CPK-MB and troponin levels were measured at baseline and in association with subsequent abnormal EKGs.

Results of cardiac monitoring are summarized in Table 2C. Nearly all of the individuals exhibited asymptomatic nonspecific ST-T wave changes at some point after DP infusions, which were not associated with increased CPK-MB or troponin levels. No dose-limiting increases in QTc intervals were noted at any time during the trial. Although several patients exhibited sinus tachycardia or mild hypotension during or shortly after DP infusion, none required pharmacologic intervention. Furthermore, whereas premature atrial or ventricular complexes were occasionally observed on monitor strips or EKGs, no patient experienced atrial or ventricular dysrhythmias that required intervention. No on-study deaths were observed in this trial.

Pharmacokinetics of DP infusion. A total of 222 plasma samples were analyzed for DP levels, including 125 from 19 patients during the first administration (day 1), 89 from 19 patients during the second administration (day 7), and 8 from 4 patients during the third administration (day 22). Mean pharmacokinetic variables for DP are listed in Table 3. The mean and SD values for DP clearance during the first administration were 7.71 (+1/−2.90) L/h/m², which were within the range of values previously observed in T-cell lymphoma patients treated with DP at doses of 18.0 mg/m² (clearance, 8.25 ± 1.85 L/h/m²)⁷

or patients with solid tumors receiving DP at a dose of 17.8 mg/m² (clearance, 10.5 ± 6.4 L/h/m²) after noncompartmental analysis (22).

DP concentration versus time profiles, which were obtained using data from all 19 patients during the first infusion (17.8 mg/m²), are provided in Fig. 1. The data confirm previous observations that DP rapidly accumulates in plasma during the infusion, reaching an apparent steady-state before the end of infusion, and rapidly declines after cessation of administration (22). The mean (±SD) estimate for the terminal half-life was 3.18 (±1.81) h, which was slightly shorter than that estimated previously in T-cell lymphoma patients [7.18 (±2.07) h]⁷ treated at the NCI. Because the contribution of the terminal phase to the overall AUC was relatively small, this difference in half-life did not substantially affect clearance estimates.

The interindividual variability in absolute drug clearance was relatively extensive (38.3%); this variability was only marginally reduced after correction for individual difference in body surface area (37.6%), suggesting that body-size measures are not contributing significantly to interindividual pharmacokinetic variability. In contrast, the intraindividual variability in absolute drug clearance was relatively minor in seven patients with complete paired available data, with a mean (±SD) ratio for clearance on the second and first administration of 1.12 (±0.59). The ratio of observed peak concentration on the second and third administration was 1.21 (±0.48). Full data are available upon request. These observations suggest that pharmacokinetic profiles of DP in lung cancer patients are essentially unchanged during consecutive administrations within a short time period.

Analysis of molecular end points. A major goal of the present study was evaluation of target gene expression in tumor tissues, relative to plasma DP concentrations in lung cancer patients. In general, all patients accrued to the trial underwent pre- and posttreatment biopsies during cycle 1, without experiencing clinically significant sequelae; essentially all of these biopsies were obtained by CT-guided FNA techniques. Unfortunately, despite real-time cytopathology confirmation of malignant cells in all aspirates, extensive necrosis, inflammation, and limited number of tumor cells precluded comprehensive analysis of molecular end points in nearly one third of patients accrued to this trial. Furthermore, because of scant amounts of cancer cells in the biopsies, not all molecular end points could be simultaneously evaluated in each patient with potentially acceptable tumor aspirates. Nevertheless, immunohistochemistry analysis revealed enhanced acetylation of histone H4 in posttreatment biopsies from four of eight patients; increased p21 expression in cancer cells was also observed in four of eight patients after DP therapy. As expected based on our preclinical studies, no patients exhibited reproducible induction of NY-ESO-1 or p16. Interestingly, one patient exhibited enhanced MAGE-3 expression in a tumor aspirate after DP infusion; this patient had previously exhibited induction of this cancer-testis antigen after Decitabine infusion (18). No obvious changes in Ki-67, phospho-ERK, or cleaved caspase-3 levels were detected in five of five patients after DP infusions. Representative immunohistochemistry results are depicted in Fig. 2.

Five pairs of pre- and posttreatment biopsies from four patients with sufficient specimens were processed for laser capture microdissection and long oligo array analysis of global gene expression profiles. Pretreatment RNA was used as the reference for each respective posttreatment array. Consistent with our observations regarding DP treatment of lung cancer cells in vitro⁸

and similar to our findings pertaining to Decitabine infusions in lung cancer patients (18), considerable heterogeneity was detected in baseline and posttreatment gene expression profiles. The limited number of arrays from DP-treated patients precluded rigorous statistical analysis of gene modulation in these individuals; as such, any genes, which on average were induced or repressed ≥2-fold in posttreatment relative to pretreatment samples, were considered potentially relevant. Surprisingly, only 16 genes were induced ≥2-fold in one or more patients after DP treatment. In contrast, >1,000 genes were repressed ≥2-fold in one or more patients after DP infusion (Table 4). Results of these arrays were compared with a data set pertaining to gene expression profiles in laser-captured lung cancer cells and adjacent histologically normal bronchial epithelia from eight patients undergoing potentially curative resections; the size of these data set enabled formal statistical analysis using a threshold of 1.2-fold induction/repression with P < 0.05 (two-tailed t test) to define potentially important lung cancer–associated changes in gene expression within extremely heterogeneous tumor specimens. Interestingly, those genes which were induced or repressed ≥2-fold by DP seemed to be down-regulated or overexpressed, respectively, in the resected primary lung cancers relative to adjacent, histologically normal bronchial epithelial cells (Table 4). Although limited and exploratory in nature, these data confirm our previous observations that comprehensive gene expression profiling is feasible using RNA amplified from laser-captured tumor cells derived from FNAs of primary pulmonary carcinomas, and raise the possibility that DP can shift global gene expression profiles in lung cancer cells toward those observed in histologically normal bronchial epithelia from lung cancer patients.

The emerging relationships between cell cycle regulation, epigenetics, and malignant transformation provide the rationale for the use of chromatin remodeling agents for lung cancer therapy (13). Of particular interest in this regard are HDAC inhibitors, which induce growth arrest, differentiation, and apoptosis of transformed cells in vitro and in vivo, while exhibiting relatively little toxicity in normal cells (29, 30). HDAC inhibitors of diverse structural classes, such as sodium butyrate, Trichostatin A, Depsipeptide FK228, and SAHA mediate hyperacetylation of core histones, facilitating chromatin relaxation, which enhances transcription factor binding to cognate recognition sequences within promoter regions (4, 5, 31). Interestingly, whereas HDAC inhibitors are known to alter expression of a variety of genes regulating cell cycle progression and apoptosis, <15% of all genes are modulated by these agents (29, 30, 32). Furthermore, numerous recent studies indicate that the antitumor effects of HDAC inhibitors may not be primarily attributable to alterations in chromatin structure, but instead may be due to the ability of these agents to modulate oncoprotein signaling, and apoptotic thresholds in cancer cells (16, 17, 33, 34).

In the present study, we sought to investigate clinical and molecular responses in lung cancer patients receiving DP at a dose of 17.8 mg/m² using a schedule not previously assessed. In general, DP was well-tolerated in these individuals. No dose-limiting nausea or vomiting was observed when DP was given in conjunction with a standardized antiemetic regimen. More importantly, no significant cardiac toxicities were observed during 120 DP infusions in this trial. These findings are consistent with results from phase I evaluation of DP, as well as recently published data pertaining to DP given at doses ranging from 14 to 18 mg/m² in patients with T-cell lymphomas at the NCI (22, 28, 35). Shah et al. (36) observed several serious cardiac events in patients with metastatic neuroendocrine tumors treated with DP at a dose of 14 mg/m² on days 1, 8, and 15 of a 28-day cycle (a treatment regimen identical to that used in T-cell lymphoma patients at the NCI). The propensity of patients with metastatic neuroendocrine tumors to exhibit cardiac toxicities which seemed attributable to DP, including one grade 5 event and two nonsustained runs of ventricular tachycardia, may have been related to underlying ventricular hypertrophy and myocardial conduction abnormalities secondary to carcinoid syndrome (37).

Several recent trials have used gene induction or histone acetylation changes in peripheral blood mononuclear cells as surrogate markers of treatment response in cancer patients receiving HDAC inhibitors (38, 39). In our trial, we focused our efforts on examination of end points in tumor tissues, as this provided the most relevant analysis of molecular response to DP. Unfortunately, scant amounts of tumor cells in FNAs precluded comprehensive evaluation of molecular end points in this study. Nevertheless, despite the limitations of the molecular analysis, data from this trial, as well as our published preclinical experiments, support further evaluation of DP in lung cancer patients. The fact that hyperacetylation of histone H4 and enhanced p21 expression were observed in posttreatment biopsies confirms that, at least in some individuals, intratumoral drug concentrations were sufficient to mediate chromatin remodeling and alter gene expression. Furthermore, the microarray results suggest that DP can shift global gene expression profiles toward those observed in “normal” bronchial epithelia from lung cancer patients. Additional, comprehensive studies are required to confirm these observations, and to ascertain if the apparent molecular responses are attributable to inhibition of specific HDACs, or are secondary to time-dependent alterations in signal transduction and microRNA expression (40).

Our published data clearly indicate that DP-mediated alterations in chromatin structure and oncoprotein signaling are primarily time-dependent, rather than concentration-dependent, under doses achievable in clinical settings (4, 16, 17). These data could explain the apparent lack of changes regarding phospho-ERK and Ki67 levels in tumor tissues harvested 18 to 24 h after completion of DP infusions. Indeed, DP may be much more effective for cancer therapy when given by prolonged infusion. Although concerns regarding potential cardiac toxicities may prevent evaluation of prolonged infusions of DP, there are presently several potential strategies that could be used to enhance the antitumor effects of this HDAC inhibitor in lung cancer patients. For instance, we have shown that abrogation of mitogen-activated protein kinase, phosphatidylinositol 3-kinase/AKT, or cyclooxygenase-2 signaling dramatically enhances DP-mediated cytotoxicity in lung cancer cells (17, 41); these findings clearly support the use of novel agents targeting these survival pathways as a means to augment DP responses in lung cancer patients. Furthermore, we have shown that p21 protects cancer cells from DP-mediated cytotoxicity (42). Indeed, DP-mediated induction of p21 may have accounted (at least in part) for the lack of objective responses observed in this trial. Under clinically achievable exposure conditions, Flavopiridol abrogates DP-mediated induction of p21, markedly potentiating apoptosis mediated by this HDAC inhibitor in cancer cells (42). These observations have provided the preclinical rationale for an ongoing phase I evaluation of sequential DP/Flavopiridol infusion in patients with thoracic malignancies at the NCI. Lastly, our observations that DP enhances cancer-testis antigen expression (4, 14) and sensitizes tumor cells to T cell–mediated, as well as natural killer cell–mediated, cytotoxicity (14, 19) support evaluation of this HDAC inhibitor in conjunction with a variety of immunotherapy regimens in lung cancer patients.