Cyclin-Dependent Kinase Inhibitors and the Treatment of Gastrointestinal Cancers

Abstract

The cell cycle is a highly conserved and tightly regulated biological system that controls cellular proliferation and differentiation. The cell cycle regulatory proteins, which include the cyclins, the cyclin-dependent kinases (CDKs), and the CDK inhibitors, are critical for the proper temporal and spatial regulation of cellular proliferation. Conversely, alterations in cell cycle regulatory proteins, leading to the loss of normal cell-cycle control, are a hallmark of many cancers, including gastrointestinal cancers. Accordingly, overexpression of CDKs and cyclins and by contrast loss of CDK inhibitors, are all linked to gastrointestinal cancers and are often associated with less favorable prognoses and outcomes. Because of the importance that the cell cycle regulatory proteins play in tumorigenesis, currently there is a broad spectrum of cell-cycle inhibitors under development that, as a group, hold promise as effective cancer treatments. In support of this approach to cancer treatment, the growing availability of molecular diagnostics techniques may help in identifying patients who have driving abnormalities in the cell-cycle machinery and are thus more likely to respond to cell-cycle inhibitors. In this review, we discuss the prevalence of cell-cycle abnormalities in patients with gastrointestinal cancers and provide a preclinical and clinical overview of new agents that target cell-cycle abnormalities with a special emphasis on gastrointestinal cancers.

Unrestrained proliferation is a hallmark feature of gastrointestinal (GI) cancers.¹ The molecular pathogenesis of GI cancers is linked to oncogene activation such as RAS, dysfunction of tumor suppressor genes such as adenomatous polyposis coli, alternations in DNA repair pathways such as mismatch repair gene abnormalities, and cell-cycle dysregulation such as cyclin-dependent kinase (CDK) 4 overexpression.1, 2, 3 Other events that play integral roles in the development of GI cancers include inflammation and immune dysregulation, and the interaction of these causative factors was recently reviewed.³ Given the prevalence of cell-cycle abnormalities in GI cancers, there is growing interest in developing and testing inhibitors that target the cell cycle.¹

Overview of the Mammalian Cell Cycle

The cell cycle is a highly structured and regulated system, composing of multiple regulatory, catalytic, and inhibitory proteins that act to direct normal mammalian cell proliferation and differentiation. It is not surprising, therefore, that the mechanisms that control normal cell division are frequently altered in many diseases, and aberrant cell-cycle control is a hallmark of most cancers.⁴ Cell division is divided into two distinct stages, mitosis (M), in which the cell prepares for and undergoes cell division,⁵ and interphase, which is further divided into three subphases, G₁, S, and G₂ (Figure 1). All phases of the cell cycle are controlled primarily through the cyclic expression of the regulatory cyclins and their catalytic partners, the CDKs, and inhibited by the CDK inhibitors (CDKis).4, 6 At least nine CDKs are described, although only five of them have defined roles in the cell cycle⁶ (Table 1). CDK4/6 associates with the D-type cyclins (D1, D2, and D3) to regulate cell-cycle progression in the G₁ phase.4, 6 Similarly, the cyclin E/CDK2 complex regulates the late G₁ phase and the induction of DNA synthesis in early S phase. CDK2 also associates with cyclin A to control proper DNA replication and synthesis in the S phase.⁷ As the cell-cycle progresses, cyclin A then associates with CDK1 to promote cell entry into the M phase. This function is also aided by the activity of CDK7 and cyclin H. CDKis include several families that differ on the basis of their structure and target.⁸ Finally, CDK1 and cyclin B function as the key mediators of mitotic entry.⁸

Figure 1.

Key regulators of the mammalian cell cycle. The green plus signs represent positive regulators of cell cycle progression, whereas the red minus sign are cell cycle inhibitory proteins. The yellow P represent phosphorylation events on the Rb. Also shown are three CDK4/6 inhibitors that are currently in various stages of clinical development: PD-033299 (Pfizer, New York, NY), LY2835219 (Eli Lilly, Indianapolis, IN), and LEE011 (Novartis, Basel, Switzerland). CDK, cyclin-dependent kinase; Rb, retinoblastoma protein.

Table 1.

Cyclins, CDKs, and Their Inhibitors

Cell-cycle phase	Cyclins and CDKs	CDK inhibitors (manufacturer)
G1	Cyclin D1, D2, D3 + CDK4,6	Flavopiridol∗ (Sanofi-Aventis, Bridgewater, NJ) PD-0332991∗ (Pfizer, New York, NY) P276-00∗ (Piramal Life Sciences Limited, Mumbai, India) PHA-848125∗ (Nerviano Medical Sciences, Nerviano, Italy) LY2835219∗ (Eli Lilly, Indianapolis, IN) LEE011∗ (Novartis, Basel, Switzerland) Fascaplysin (Sigma-Aldrich, St. Louis, MO) Cynnamaldehydes (Sigma-Aldrich) AZD5438∗ (Astra Zeneca, London, England) BAY 100394∗ (Bayer, Barmen, Germany) P1446A-05∗ (Piramal Life Sciences Limited) PD183812 (Pfizer) Pyrrolo-carbazoles, indolocarbazoles, tryaminopyrimidine, dioxobenzothiazoles
S	Cyclin E + CDK2 Cyclin A + CDK2	Flavopiridol∗ SNS-032∗ (Sunesis Pharmaceuticals, South San Francisco, CA) Bryostatin-1∗ (Tocris Bioscience, Ellisville, MO) Roscovitin∗(Cyclacel Pharmaceuticals, Short Hills, NJ) Dinaciclib∗ (SCH727965) (Merck, Whitehouse Station, NJ) P276-00∗ PHA-848125∗ UCN-01∗ (Sigma-Aldrich) BAY 100394∗ Olomucine (Sigma-Aldrich) Purvalanol A (Tocris Bioscience) Aloisines (Enzo Life Sciences, Farmingdale, NY) Indirubins (Enzo Life Sciences) Hymenialdisine (Enzo Life Sciences) SU 9516 (Tocris Bioscience) AZD5438∗ (Astra Zeneca), CVT-313 (Enzo Life Sciences), butyrolactone I (Sigma-Aldrich), pyrazolo-pyridine (Sigma-Aldrich), pyrazolo-quinazolines (Astra Zeneca), indenopyrazoles (Bristol-Myers Squibb, New York City, NY), nitroso-pyrimidines (Sigma-Aldrich)
G2	Cyclin A + CDK1	Flavopiridol∗ AZD 5438∗ UCN-01∗ SNS-032∗ Bryostatin-1∗ Roscovitin∗ Dinaciclib∗ (SCH727965) BAY 100394∗ Olomucine Purvalanol A Aloisines Indirubins SU 9516 CVT-313, butyrolactone, hymenialdisine, PHA-848125, pyrazolo-pyridine, pyrazolo-quinazolines, indenopyrazoles, nitroso-pyrimidines
M	Cyclin B + CDK1

CDK, cyclin-dependent kinase.

^∗Compounds that have entered clinical development.

The Cell Cycle

The G₁ or Gap 1 phase was originally described as the period of time that occurred before the onset of DNA synthesis (the S phase).⁹ Normal cells deprived of the proper growth conditions arrest in a resting or G₀ part of G₁.⁷ On stimulation by mitogenic signals, these quiescent cells begin to progress toward a major G₁ checkpoint termed the restriction (R) point in G₁. The R point is regulated by the retinoblastoma 1 (Rb1) tumor suppressor gene. In its hypophosphorylated state Rb binds to members of the E2F family of transcription factors, most notably E2F1.⁷ This Rb/E2F interaction suppresses transcription of critical E2F-regulated cell-cycle genes through the recruitment of chromatin remodeling enzymes such as histone deacetylases.¹⁰

Early G₁ progression is controlled by the D-family of cyclins and their catalytic protein partners, CDK4 or CDK6. Cyclin D (CCND1) gene expression and protein amounts are low in quiescent cells and are rapidly induced on stimulation by growth-supportive conditions or through the activity of many oncogenes.⁷ Cyclin D then dimerizes with CDK4 or CDK6 to form a catalytically active protein complex. One of the major substrates for the cyclin D/CDK4 or CDK6 complex is Rb. The phosphorylation of Rb by the cyclin D/CDK complex begins the stochastic inactivation of Rb and allows the cell cycle to progress toward the R point of the cell cycle. Once past the R point, cells become committed to entering the S phase. The inactivation of Rb relieves its inhibitory action on the transcription factor E2F, which thereby supports further progression through the cell cycle.⁷ E2F directs the synthesis of cyclin E and CDK2 which further inactivates Rb. It is now evident that E2F regulates the expression of a variety of genes that mediate DNA replication, nucleotide biosynthesis, and DNA repair activities such as DNA polymerase α, thymidine kinase, thymidylate synthase, ribonucleotide reductase, and RAD51.⁹

To maintain proper early G₁ regulation, the activity of the cyclin D1/CDK4 or CDK6 complex is antagonized by members of the INK4 family of CDKis. The INK4 family is composed of p16^INK4a, p15^INK4b, p18^INK4c, and p19^INK4d, which specifically inhibit the catalytic subunits of CDK4 and CDK6.⁹ Similarly, the CDK interacting protein/kinase inhibitory protein (CIP/KIP) family of CDKis (p21^Cip1, p27^Kip1, and p57^Kip) are potent inhibitors of the E- and A-type cyclins and their catalytic partners, CDK2 and CDK1, and to a lesser extent of the CDK1/cyclin B complex in G₂.3, 8 Progression through late G₁ and the induction of DNA synthesis in S phase are induced by the increased cyclin E protein amounts, and the resultant association with CDK2, the primary catalytic partner of cyclin E. The activity of the cyclin E/CDK2 complex is potentiated by Cdc6,¹¹ and this complex phosphorylates p27^Kip1, further facilitating the entry of cells into S phase.⁹ It is thought that removal of p27^Kip from the cyclin E/CDK2 complex is essential for entry of cells into the S phase.⁷

CDK2 also forms a functional complex with cyclin A, a major E2F target gene. The cyclin A/CDK2 complex is necessary for proper DNA replication and synthesis in the S phase.¹² In late G₂ and early M phase, cyclin A also associates with CDK1 to promote cell entry into the M phase.¹³ This occurs in part through the activity of CDK7 and cyclin H which form a regulatory complex referred to as the CDK-activating kinase.⁷ Finally, cell-cycle progression through G₂ into the M phase is controlled by the cyclin B/CDK1 complex, a tightly regulated and nonredundant cyclin/CDK complex that is required to promote the proper traversing of the cell into M phase. Ubiquitin-mediated degradation of cyclin B1 is required for the cells to properly progress through mitosis.⁷

As mentioned earlier, the dysfunction of the cell cycle occurs in most human cancers as a result of aberrant cyclin and CDK function.¹⁴ Therefore, targeting CDK is a potentially effective strategy in developing cancer therapeutics (Table 1).

Cell-Cycle Abnormalities in GI Tumors

Colon Cancer

To get a better understanding of the mechanisms of tumorigenesis in GI cancers and to design better, more-effective therapies, numerous studies have investigated the alterations in cell-cycle regulation that occur in GI tumors and the concomitant effects that these changes have on tumor aggressiveness, drug and radiation sensitivity, and overall patient outcome (Table 2). Perhaps, not surprisingly, the data underline the complexity of cell-cycle regulation and the importance of knowing which pathways are altered, on a patient-by-patient level, when designing therapeutic regimens. For example, Cole et al³² found that cyclin D2 and CDK4/6 are overexpressed after APC loss in the intestinal epithelium which suggests that deregulation of CDK4/6 is required for enterocyte proliferation and adenoma formation. In addition, Zhao et al³³ evaluated paraffin sections of 74 cases of colorectal carcinoma and found stronger immunostaining of p16^INK4a and CDK4 in the cytoplasm of carcinomas than in adenomas and adjacent normal tissue. Of the 74 specimens examined, 73 stained positive for p16^INK4a of which 53 showed a strong expression pattern. Prognosis was substantially better for tumors with strong expression of p16^INK4a. The expression of p16^INK4a and CDK4 was scored by multiplying the extent of positivity and its intensity and by grading it on a scale of 0 to 12 where strong staining was defined as a score of 9 to 12. All 74 specimens showed CDK4 expression, but only 33 specimens showed strong expression. Stronger immunostaining for CDK4 was predictive of a worse prognosis (P < 0.001). Conversely, Ogino et al³⁴ conducted an intriguing study to evaluate the prognostic relevance of cyclin D1, independent of other confounding variables such as p53, p21^Cip1, p27^Kip1, KRAS (alias Ki-ras), BRAF mutation, microsatellite instability, the CpG island methylator phenotype, and long-interspersed nuclear element-1 hypomethylation. Their cohort study of 602 patients with colon cancer found that cyclin D1 was overexpressed in 55% of tumors. Surprisingly, cyclin D1 overexpression was associated with low cancer-specific mortality on multivariate regression analysis [hazard ratio (HR), 0.57; 95% CI, 0.39–0.84; P = 0.0048). A similar favorable trend was observed for overall mortality (HR, 0.74; 95% CI, 0.57–0.98; P = 0.036). These results indicate that cyclin D1 expression in colon cancer tumors may be associated with a favorable prognosis, a finding that was previously reported³⁵ but contradicts the common belief about the poor prognosis of cyclin D1 and the observations by Zhao et al³⁶ that reported the association of overexpression of the cyclin D1 heterodimeric partner CDK4 with worse prognosis.

Table 2.

Cell-Cycle Abnormalities in Patients with Gastrointestinal Cancers

Malignancy	Abnormality	Prevalence	Significance
Colon	CDK4/6 overexpression15 Cyclin D2 overexpression16		Associated with APC loss and intestinal proliferation
	CDK4 overexpression15	74/74 specimens; 33/74 yielded strong expression	Strong expression-associated with worse prognosis
	CDK2 overexpression17	86% in carcinoma cells; 28% in adenomas
	CDK1 overexpression18		Higher risk of distant metastasis (HR, 6.2)
	Cyclin D1 overexpression19, 20	55% of tumors	Better overall mortality (HR, 0.74)
	p16 overexpression15	73/74 specimens	Better prognosis
	Loss of p2721, 22, 23	10% tumors	Worse prognosis; risk ratio of death, 2.9
	p27 overexpression24	30% of tumors
Esophagus	Chromosomal region for CDK4 amplified25, 26	10%	Worse overall survival
	CDK4+6 amplification25, 26	4/116 tumors	Poor survival
	CDK1 overexpression25, 26	54%	Poor prognosis
	CDK2 overexpression26	56%
	Cyclin D1 overexpression26	41%	Poor prognosis
	p27 overexpression26		Poor survival; higher grade tumors
Stomach	CDK4 overexpression27	48% of 260 specimens
	Cyclin D1 overexpression27	34% of specimens
	Cyclin D2 overexpression27	30% of specimens	Disease progression
	Cyclin E overexpression27	44% of specimens
	Low p27 expression27, 28	62%	Poor survival (RR, 2.64)
	Loss p2129		Distant metastasis
Pancreas	CDK4 overexpression19	50%–75%
	Cyclin D1 overexpression16	68%	Poor survival
	p21 overexpression30, 31	85%
	p16 Loss19	83%	Increased risk of early metastasis and poor survival
HCC	CDK4 overexpression24	73%	Poor survival
	Cyclin D1 overexpression24	33%	Poor prognosis
	Cyclin E overexpression24	36%
	p16 expression24	90%
Biliary cancers	Cyclin D122	62%
	p16 loss22	31%–40%	Poor prognosis

APC, adenomatous polyposis coli; CDK, cyclin-dependent kinase; HCC, hepatocellular carcinoma; HR, hazard ratio; RR, relative risk.

Yamamoto et al¹² found that CDK2 was overexpressed in a higher percentage of colon cancer tumor cells than of adenoma cells (86% versus 28%), and Zeestraten et al³⁷ found that elevated expression of CDK1 in tumor specimens from 254 patients with stage II colon cancer correlated with higher risk of distant metastasis (HR, 6.2; 95% CI, 1.44–26.9; P = 0.012). Moreover, CDK1 amounts were substantially elevated in microsatellite-stable tumors. This finding may be a result of the rigorous multivariate regression analysis conducted by the investigators or could be a statistical aberration. Loda et al³⁸ examined a series of 149 primary human colorectal cancer (CRC) specimens and determined that absence of the CDKi p27^Kip1 resulted in a risk ratio for death of 2.9 (P = 0.003). p27^Kip1 expression was absent in 10% of tumors, low (≤50% of cells staining positive) in 60% of tumors, and high (≥50% of cells staining positive) in 30% of specimens. Median survival was 69 months in tumors that lacked p27^Kip1 expression versus 151 months in p27^Kip1-positive tumors. These results indicated that the lack of p27^Kip1 expression conferred worse prognosis in patients with CRC. Subsequent studies found that loss of p27^Kip1 expression (both nuclear and cytoplasmic) correlated with microsatellite instability in patients with CRC,¹⁷ highlighting an association between loss of CDKi and tumorigenesis. The same group also described a statistically significant association between cyclin D1 overexpression and microsatellite instability–high status (P ≤ 0.02).³⁹ In addition, others have observed a positive association between p21^Cip1 overexpression and microsatellite instability–high status.⁴⁰

Gastroesophageal Cancer

In esophageal cancer a study conducted by Ismail et al⁴¹ evaluated a cohort of 116 patients and found that the chromosomal region that contained CDK4 was amplified in 10% of tumors and was associated with worse overall survival (P = 0.019). Only 4 of 116 tumors showed amplification of both CDK4 and CDK6, and this was associated with poor survival (HR, 2.1; P = 0.0008). Similarly, loss of p27^Kip1 expression correlated with higher grade (P < 0.001) and poor survival (P = 0.05).⁴² Takano et al,²⁷ found that CDK4 overexpression was detected by immunohistochemistry (IHC) in 48% of 260 gastric cancer cases. Cyclin D1, D2, and E were overexpressed in 34%, 30%, and 44% of cases, respectively. Overexpression of CDK4 and cyclin D2 and loss of p27 substantially correlated with tumor progression on univariate analysis.⁴³ On multivariate analysis, only cyclin D2 and p27^Kip1 changes correlated with progression. In gastric cancer, Mori et al²⁸ analyzed p27^Kip1 expression in 138 patients. Low p27^Kip1 expression was present in 62% of tumors and correlated significantly with increased tumor size, invasion outside the gastric wall, the presence of lymph node metastasis, and higher stage. Loss of p27^Kip1 staining was independent of prognostic markers for survival on multivariate analysis (relative risk, 2.64; P < 0.01). Similar results were observed with p21^Cip1. The loss of p21^Cip1 predicted increasing histologic grade, depth of invasion, and distant metastasis in patients with gastric cancer.²⁹

Pancreatic Cancer

Cell-cycle abnormalities are frequently observed in pancreatic cancer. Cyclin D1 overexpression was found in 68% of pancreatic cancer specimens with the use of IHC,²⁸ consistent with the known induction of cyclin D1 by oncogenic ras,¹⁵ a driver oncogene in pancreatic cancer. In addition, Southern blot analyses revealed amplification of the cyclin D1 coding region in 25% of the pancreatic cancer specimens, and mRNA was overexpressed with RT-PCR in 82% of the examined tissue.¹⁶ Investigators have also reported that cyclin D1 overexpression correlated with poor survival (median survival, 18.1 months versus 10.5 months for normal expression versus cycling D1 overexpression; P < 0.01). Overexpression of CDK4 and CDK2 was noted in approximately 10% of pancreatic intraepethelial neoplasia (PanIN) 1B lesions. Although there was a progressive increase in higher grades of PanIN and invasive cancer, the trend was not statistically significant. The median percentage of expression was 60% to 75% in carcinoma cells.¹⁹ In pancreatic cancer, p21^Cip1 overexpression is also common and appears to be an early event in pancreatic neoplasia. An analysis by Biankin et al⁴⁴ found that p21^Cip1 overexpression was present in 9% of normal pancreatic ducts, 16% of PanIN1A lesions, 32% of PanIN1B lesions, 56% of PanIN2 lesions, and 85% of patients with pancreatic adenocarcinoma (P < 0.01). Similarly, p16^INK4a inactivation was observed in the majority (83%) of pancreatic cancer tumors in a study conducted by Rozenblum et al.²⁰ It was also thought to be an early event in pancreatic tumorigenesis.¹⁸ Sasaki et al¹⁸ observed that partial loss of p16^INK4a expression was observed in 9 of 16 and 12 of 13 pancreatic adenoma and carcinoma samples, respectively. In addition, low amounts of p16^INK4a expression were associated with larger tumors, risk of early metastasis, and poor survival.¹⁸

Hepatocellular Carcinoma

Several studies have evaluated the expression of cell-cycle proteins in hepatocellular carcinoma (HCC) cells. A study by Ito et al²⁴ found that p21^Kip1 was expressed when evaluated by IHC in 54 of 104 specimens. p21^Kip1 expression was substantially higher in cases of intrahepatic metastasis, but no other correlation could be established between p21^Kip1 and the tumors' clinicopathologic features. 16^INK4a, however, was expressed in 94 of 104 cases. The labeling index of 16^INK4a in HCC cells was lower in stages III or IV than in stages I and II (36.5 ± 26.8 versus 51.0 ± 28.2; P = 0.121). The labeling index of p27^Kip1 was overexpressed in HCC cells and was substantially decreased in cases with portal invasion, poor differentiation, and larger tumor size. Cyclin D1 was overexpressed in 34 of 104 HCC specimens. Overexpression of cyclin D1 was associated with higher Ki-67 and poor differentiation. Similarly, cyclin E was overexpressed in 37 of 104 specimens, and its overexpression was associated with higher Ki-67 and higher stage.²⁴ A study by Lu et al²¹ found that CDK4 was overexpressed by IHC in 73% of 59 specimens. Overexpression of CDK4 was an independent prognostic factor for poor survival.

Biliary Cancers

Cell-cycle abnormalities are also common in biliary neoplasm. Cyclin D1 overexpression was observed in 62% of patients with intrahepatic cholangiocarcinoma. In addition, loss of p16^INK4a and p27^Kip1 was detected in 31% and 12% of tumors, respectively.²² Similarly, another study found that p16^INK4a expression was lost in 40% of extrahepatic biliary neoplasms and was associated with poor prognosis.⁴⁵ In gallbladder cancer, cyclin D1 overexpression was detected in 41% of the examined specimens. Cyclin D1 overexpression was significantly associated with decreased overall survival (P < 0.05) in those patients.²³

Genetic Alternations Affecting the Cell-Cycle Components in GI Tumors

The association between genetic mutations and CDK overexpression has been the focus of several studies. Grady et al⁴⁶ found that CDK4 overexpression was found in colorectal tumors that carry type II transforming growth factor-β receptor mutations, a mutation that results in tumor microsatellite instability. These results are important because approximately 15% of human colon cancers have microsatellite instability caused by this mutation⁴⁶ and provided evidence that deregulation of CDK4 can be a consequence of transforming growth factor-β receptor mutations. Similarly, inherited mutation of CDKN2A (the gene that encodes p16^INK4a)²⁵ is responsible for familial cutanenous melanoma and was associated with an increased risk of pancreatic cancer in families that harbor the mutation. The CDKN2A gene is located on chromosome 9p, the most frequently inactivated gene in pancreatic cancer.⁴⁷ Loss of p16^INK4a in pancreatic cancer can also occur through homozygous deletion (40%), single allelic loss associated with a mutation in the second allele (40%), and promoter hypermethylation (15%).⁴⁷ The CDKN2A gene was also found to be inactivated in 40% of colon cancers.⁴⁸ The inactivation of CDKN2A was thought to be secondary to de novo methylation of 5′ CpG island of the gene.³⁰ Cyclin D1 protein overexpression is rather common in CRC.⁴⁹ The mechanism of cyclin D1 overexpression is thought to be secondary to transcription activation of the CCND1, the gene coding for cyclin D1. A study by Bondi et al,³¹ however, found that extra gene copies of cyclin D1 were seen in 50% of 219 colon adenocarcinoma tumors. Another study suggested copy number changes, predominately gains (7.6%) and rarely amplifications (2.5%), may be associated with the increase in the cyclin D1 expression.³¹ These results suggest that several mechanisms may contribute to the overexpression of cyclin D1 in patients with CRC. Taken together, these studies suggest cell-cycle abnormalities occur frequently in GI tumors. It appears that overexpression of CDK and cyclins and by contrast loss of CDKis such as p21^Cip1 and p27^Kip1 are associated with less favorable tumor phenotypes and poor prognosis. Therefore, targeting the cell cycle represents a promising treatment modality and perhaps identifying patients who have abnormalities in the cell-cycle machinery, either through IHC or genetic sequencing, may enrich the population of patients most likely to respond to treatment with agents that target the cell-cycle abnormalities. Currently, many cell-cycle–targeting agents are in various stages of clinical development (Table 1).

Preclinical Data for Cell-Cycle Inhibitors in GI Tumors

Colon Cancer

Preclinical models have evaluated the use of pharmacologic CKDis in solid tumors and GI malignancies.50, 51 PD-0332991 (palbociclib; Pfizer, New York, NY) is an orally administered, highly specific inhibitor of CDK4 and CDK6. In mice bearing Colo-205 human colon carcinoma, PD-0332991 produced marked tumor regression.⁵⁰ With 14 days of therapy at a dose of 150 mg/kg, there was a tumor growth delay of 50 days and greater than on log of tumor-cell kill. Retreatment of the tumors with PD-0332991 was not associated with development of resistance. As expected from a CDK4/6 inhibitor, PD-0332991 administration eliminated phosphorylated Rb, down-regulated genes regulated by E2F, and caused a G₁ cell-cycle arrest. Conversely, PD-0332991 was not active against Rb-negative tumors. Similarly, Yamamoto et al¹² evaluated the use of butyrolactone I, a specific CDK2 inhibitor, in four colon cancer cells lines (HCT116, LoVo, HT29, Colo 320DM). Butyrolactone I inhibited proliferation of the four colon cancer cell lines and induced apoptosis in the LoVo cell line with induction of p53. Another CDK2 inhibitor, SU 9516 (Tocris Bioscience, Ellisville, MO), was investigated by Lane et al⁵² in human colon cancer cell lines (RKO, SW480, and Colo250). Administration of SU 9516 resulted in selective inhibition of CDK2 and either a G₀ to G₁ or a G₂ to M arrest. A novel pan-CDK inhibitor RGB 286199 was also investigated in human colon carcinoma cell line HCT-116.⁵³ Cell lines treated with RGB 286199 showed loss of Rb phosphorylation, accompanied by loss of cyclin A protein. In vivo, RGB 286199 results in cell-cycle arrest and loss of viability. CDKi combination with DNA-damaging agents was also investigated in preclinical models. Takagi et al⁵⁴ found that the CDKi SU 9516 reduced the expression of thymidylate synthase in human CRC DLD-1 cells. This effect enhanced the sensitivity of colon cancer cells to 5-fluorouracil (5-FU) and suggested that a CDKi/5-FU combination may be a promising future treatment option. Correspondingly, Pishvaian et al⁴² found that the combination of PD-0332991 and oral 5-FU (capecitabine) resulted in synergistic anticancer activity in mice bearing HCT-116 colon cancer xenografts. Similarly Ziemke et al⁵⁵ found that the combination of PD-0332991 and multiple mitogen-activated protein extracellular signal-related kinase inhibitors was associated with higher frequency of regression of colon cancer cell lines compared with either agent alone. These data suggest that combination therapy, including cell-cycle inhibitors, may be a promising strategy in treating patients with CRCs.

Hepatocellular Carcinoma

PD-0332991 was also evaluated in human hepatocellular carcinoma cells Huh7, HepG2, and Hep3B and xenograft models that harbor Rb knockdown and mice with liver-specific Rb deletion.⁵¹ PD-0332991 resulted in arrest of cell-cycle progression in hepatoma cells, irrespective of Rb status. The model suggested that cell-cycle arrest can be achieved in Rb-deficient tumors exposed to CDK4/6 inhibitors. Pishvaian et al⁵⁶ found that most hepatoma cell mouse models exhibit increased expression of cyclin D1 and CDK4. Moreover, in human HCC cell lines, PD-0332991 inhibited cell growth by 30%. This was associated with a decrease in cyclin E expression. Finally, PD-0332991 antitumor effects were additive to doxorubicin in HepG2 xenografts. These results suggest that CDK4/6 inhibition may be a feasible treatment for HCC.

Gastroesophageal Carcinoma

Investigators have combined flavopiridol (Sanofi-Aventis, Bridgewater, NJ) with radiation therapy in a human esophageal adenocarcinoma (SEG-1) cell model.⁵⁷ This model showed that flavopiridol given before or after radiation therapy exhibits a radiosensitizing effect. The cells were arrested at the G₁ phase of the cell cycle. Moreover, flavopiridol enhanced radiation-induced apoptosis and inhibited transcription activity as evidence by reduction in RNA polymerase II. These results suggested that CDKis can potentially be combined with radiation therapy as radiosensitizers. Moreover, PD-0332991 was shown to have potent antiproliferative activity at low nanomolar range in multiple gastric cancer cell lines. Of note, cyclin D1- and HER2-amplified cells exhibited greater sensitivity to this agent. In addition, combination therapy of trastuzumab and PD-03329991 indicated substantial synergy in HER2-amplified gastric cancer models.⁵⁸

Pancreatic Cancer

Studies in pancreatic cancer cell lines suggested that the CDKi flavopiridol (alvocidib) has synergistic activity when combined with gemcitabine, a commonly used chemotherapeutic agent in pancreatic cancer.⁵⁹ The observed synergy is mediated through up-regulation of RR-M2, a DNA enzyme involved in DNA synthesis and gemcitabine resistance that is regulated by E2F. Similarly, another study found that treatment of pancreatic cancer cell lines with E2F-1 virus and roscovitine (Seliciclib; CYC202; Cyclacel Pharmaceuticals, Short Hills, NJ), a CDK2, CDK7, and CDK9 inhibitor, results in an additive effect on cell growth inhibition and induction of apoptosis.⁶⁰ In addition, investigators have reported that roscovitine can effectively block the proliferation of human pancreatic cancer cells, regardless of their mutational status of KRAS, p53, or p16 genes.⁶¹

Clinical Data for Cell-Cycle Inhibitors in GI Tumors

Flavopiridol remains the most extensively studied CDKi in solid tumors and specifically in GI cancers. Flavopiridol inhibits CDK1, CDK2, CDK4, CDK7, and CDK9 by their respective ATP-binding sites.⁶² Flavopiridol also has the ability to enhance apoptosis induced by chemotherapy.⁶² It is thought that the proapoptotic effect of flavopiridol is mediated through the inhibition of antiapoptotic genes at the transcription level.⁶³ This effect is often sequence dependent. As an example, tumor cell killing is enhanced when flavopiridol is administered after exposure to taxanes.⁶⁴ Flavopiridol was therefore investigated in combination with chemotherapy agents. For example, flavopiridol was tested in combination with irinotecan by Shah et al⁶⁵ in a phase 1 trial that involved 45 patients with advanced malignancies. Irinotecan was administered first, followed 7 hours later by escalating doses of flavopiridol administered over 1 hour weekly 4 weeks on/2 weeks off. The recommended phase 2 dose was irinotecan 100 mg/m²/flavopiridol 60 mg/m² or irinotecan 125 mg/m²/flavopiridol 50 m/m². At irinotecan dose of 125 mg/m²/flavopiridol 60 mg/m², dose-limiting hyperbilirubinemia, fatigue, and myelosuppression were observed. Three patients had a partial response (PR), and 36% of patients had prolonged stable disease (SD) (>6 months). Of the 27 patients with CRC, 14 (52%) had SD, 1 had PR, and 12 patients (44%) had progressive disease. The median duration of disease control (PR + SD) was 6.8 months. Flavopiridol was also evaluated in a phase 1 trial of 48 patients with advanced solid tumors in combination with FOLFOX (5-FU, leucovorin, and oxaliplatin).⁶⁶ Patients were treated with a biweekly sequential regimen of flavopiridol at a starting dose of 40 mg/m² over 1 hour, concomitant oxaliplatin at a starting dose of 60 mg/m² and leucovorin, followed by a bolus of 5-FU at a fixed dose of 400 mg/m² and continuous 5-FU at a starting dose of 1800 m/m². The maximum tolerable doses (MTDs) were flavopiridol 70 mg/m², oxaliplatin 85 mg/m², and 5-FU 1800 mg/m² continuous infusion. Encouraging clinical activity was noted in platinum-resistant germ cell tumors. Responses were also observed in pancreatic and gastric cancers. A similar regimen was evaluated by Meng et al⁶⁷ in a phase 1 trial that involved 19 patients with advanced solid tumors. The regimen included a fixed 40-mg/m² dose of flavopiridol administered over 1 hour concurrently with escalating doses of oxaliplatin given as part of a modified FOLFOX6 regimen given at standard doses every 2 weeks. Grade 3 hyponatremia and syncope were encountered at an oxaliplatin dose of 85 mg/m² and 5-FU dose of 1200 mg/m² per day. This regimen was clinically active with one PR in a patient with pancreatic cancer and SD in gastric cancer, anal cancer, and CRC. Flavopiridol was also evaluated in a phase 1 trial in combination with standard dose of 5-FU and irinotecan (FOLFIRI) every 2 weeks.⁶⁸ Two assessments of the MTD were planned. MTD₁ was evaluated with flavopiridol administered over 1 hour, and MTD₂ was evaluated with flavopiridol as a 30-minute bolus followed by a 4-hour infusion. Seventy-four patients were treated, 63 were evaluable for toxicity and 56 for response. MTD₁ of flavopiridol was 80 mg/m² and the dose-limiting toxicities (DLTs) were diarrhea, fatigue, neutropenia, and neuropathy. MTD₂ was a bolus of 35 mg/m² and 35 mg/m² over 4 hours. DLTs were diarrhea, neutropenia, and fatigue. Of 25 patients with CRC, 11 had SD lasting >3 months (median, 6 months; range, 4.2 to 15.4 months), despite prior progression on ≥1 irinotecan-containing regimen. Six of those patients had significant reduction (36% to 78%) in their carcinoembryonic antigen amounts. One patient with small bowel cancer had a PR that lasted 10.3 months.

Phase 2 trials of flavopiridol include a single-agent study with flavopiridol in patients with previously untreated advanced CRC.⁶⁹ Twenty patients were treated with flavopiridol at a dose of 50 mg/m² per day as a 72-hour continuous infusion every 2 weeks. No objective responses were observed, and five patients had SD that lasted a median of 7 weeks, the median survival was 65 weeks, and the median time to progression was 8 weeks. The most commonly occurring toxicities were diarrhea (21%), fatigue (11%), and hyperglycemia (11%). A similar regimen of flavopiridol was investigated in 16 patients with advanced gastric cancer. There were no objective responses. Grade 3 or 4 fatigue and diarrhea occurred in 27% and 20% of patients, respectively. An unexpectedly high number of patients (5 of 14 evaluable patients) developed central line–associated venous thrombosis. This study added more evidence that flavopiridol has minimal activity as a single agent. Attention was therefore shifted to the role of flavopiridol in combination with other chemotherapeutic agents. Flavopiridol was evaluated in phase 2 trials in combination with docetaxel in 10 patients with refractory metastatic pancreatic cancer.⁷⁰ The regimen included docetaxel 35 mg/m² followed by flavopiridol 80 mg/m² on days 1, 8, and 15 of a 28-day cycle. Three patients had SD, and median survival was 4.2 months. Adverse events (AEs) included transaminitis (11%), grade 4 neutropenia, grade 3 fatigue, and grade 3 diarrhea. This regimen was thought to have minimal activity in patients with pancreatic cancer. Although preclinical studies of flavopiridol sparked optimism about the role of flavopiridol as a new option for cancer treatment, clinical studies to date have not found that it possesses any meaningful activity as single agent or combination in GI tumors, likely because of the substantial toxicity that may be a result of pan-CDK inhibition and off-target effects.

PD-0332991 has gained interest because it is a selective CDK4/6 inhibitor. It was investigated in two phase 1 clinical trials. PD-0332991 was administered for 14 days, followed by 7 days off treatment to 33 patients with Rb-positive advanced solid cancers.71, 72 Six patients had colon cancer. The MTD was 200 mg daily. Six patients (18%) had DLTs. Treatment-related AEs were encountered in 29 patients (88%). These AEs were generally mild to moderate. The most common nonhematologic AEs were fatigue, nausea, diarrhea, constipation, epistaxis, and rash. Grade 3 and 4 hematologic toxicity included neutropenia (24%), leukopenia (21%), thrombocytopenia (9%), and anemia (3%). No responses were reported in patients with GI malignancies. A second phase 1 dose escalation trial was conducted with the use of a 21 days on, 7 days off schedule. Forty-one patients were enrolled, including three with CRC, in six-dose escalation cohorts in a standard 3 + 3 fashion. The MTD was 125 mg. Five patients (12%) experienced DLTs. Neutropenia was the only DLT. After cycle 1, neutropenia, leukopenia, and anemia occurred in five patients (12%), one patient (2%), and three patients (7%), respectively. The most common nonhematologic AEs were fatigue, nausea, and diarrhea. Of the patients with GI cancers, SD was observed in one patient with appendiceal carcinoma treated at the 100-mg dose until the time of data cutoff (39 cycles).

PD-0332991 was investigated in a phase 2 trial in patients with HCC. Patients with refractory HCC received PD-0332991 at a dose of 125 mg daily for 3 weeks with a 1-week break. Preliminary results from 10 patients indicated that the most common toxicities were neutropenia and thrombocytopenia. Three patients developed grade 3 neutropenia that required treatment delay. Preliminary efficacy results were encouraging with four patients remaining on the trial with the best progression-free survival being 8 months.⁷³ Of note, other CDK4/6 inhibitors, LY2835219 (Eli Lilly, Indianapolis, IN) and LEE011 (Novartis, Basel, Switzerland),74, 75, 76 have also entered clinical development and have shown promising activity in breast cancer, lung cancer, and other solid tumors. In a preliminary report, LEE-01 has demonstrated encouraging activity and safety profiles in a phase 1 study in patients with advanced solid tumors.⁷⁵ Furthermore, a preliminary report of LY2835219 suggested that it had a favorable safety profile as monotherapy in metastatic breast cancer and in combination with fulvestrant in hormone-receptor–positive breast cancer.⁷⁶ As monotherapy, LY2835219 had promising clinical activity [eight confirmed and three unconfirmed PR (n = 47)]. Clinical activity of the combination was not reported, however, because of the preliminary nature of the report. Similarly, a phase 1 study of LY2835219 in 49 patients with advanced non–small cell lung cancer indicated an encouraging safety profile. Disease control rate was 51% with one confirmed PR.⁷⁴ To date, this is the most encouraging data to indicate that cell-cycle inhibitors may have promising antitumor activity and an acceptable safety profile.

Emerging evidence suggests that certain aberrations in cell-cycle proteins could predict clinical outcome of patients treated with CDK4/6 inhibitors58, 77, 78, 79 (Table 3). These data, although preliminary, may allow for selection of subgroups of patients that may preferentially benefit from treatment with CDKis.

Table 3.

Predictive Biomarkers for CDK4/6 Inhibitors

Author	Agent	Setting (n)	Biomarker	Outcome
Wainberg58	PD-0332991	Gastric cancer cell lines (17)	Cyclin D1 amplification	↑Sensitivity
			Cyclin E	↑Resistance
		Colon cancer cell lines (27)	p16 loss	↑Resistance
			p21 gain	↑Resistance
Konecny78	PD-0332991	Ovarian cancer cell lines (40)	p16 low expression	↑Sensitivity
Finn79	PD-0332991	Breast cancer cell line (47)	p16 decrease	↑Sensitivity
			Cyclin D1 increase	↑Sensitivity
			pRb	↑Sensitivity
von Euw77	PD-0332991	Melanoma cell lines	Hedgehog pathway activation	↑Resistance

CDK, cyclin-dependent kinase; pRb, phosphorylated retinoblastoma.

BAY 1000394 (Bayer, Barmen, Germany) is also a pan-CDKi that targets CDK1, CDK2, CDK4, CDK7, and CDK9 but is available in tablet form and as an oral solution. A recent report of a phase 1 trial of BAY 1000394 as an oral solution suggested limited tolerability with a regimen of 28 days on/14 days off. The main DLTs were hyponatremia and hypokalemia. Four of 10 patients, however, had SD, including a patient with esophageal cancer who had SD for 2.5 to 3 months.⁸⁰ Interestingly, when the same agent was administered in tablet form with the regimen of 3 days on/4 days off, it showed acceptable tolerability.⁸¹ Main grade 3 AEs were asthenia, nausea, and vomiting. SD was observed in 9 of 34 patients, including 1 patient with cholangiocarcinoma who had SD that lasted for 5 months.

UCN-01 (Sigma-Aldrich, St. Louis, MO) is another cell-cycle inhibitor that has gained interest recently.⁸² p53-deficient cells depend on activation of Chk1 pathway for G₂ cell-cycle arrest and DNA repair after DNA damage.⁸³ This pathway represents a survival pathway for tumor cells. UCN-01 inhibits Chk1 and Chk2 kinases, thereby leading cells to exit the G₂ phase before DNA repair can be completed.⁸² This mechanism forces cells to undergo apoptosis. The clinical utility of UCN-01 was complicated by the development of UCN-01–related hyperglycemia.⁸⁴ This complication was reported in patients with and without diabetes mellitus. Grade 3 and 4 hyperglycemia, occasionally requiring hospitalization, was reported in patients who received UCN-01. The cause of hyperglycemia remains unclear. There is evolving evidence, however, that inhibition of the phosphatidylinositol-3 kinase signaling protein AKT and downstream targets of the insulin receptor are responsible for the UCN-01–induced hyperglycemia.⁸⁵ Other Chk1 inhibitors, such as CHIR124 and 17-AAG, are currently in clinical development.86, 87 They differ structurally from UCN-01 and have shown encouraging preclinical data.

Bryostatin (Tocris Bioscience) is a cell-cycle inhibitor that induces p21^Kip1, resulting in inactivation of CDK2 and inhibition of tumor cell growth.87, 88 Phase 1 studies of bryostatin found that myalgia is the DLT.88, 89 Phase 2 studies of bryostatin in GI cancers were conducted.90, 91 The combination of bryostatin with paclitaxel was explored in pancreatic cancer and proved to be ineffective. The paclitaxel/bryostatin combination was also explored in esophageal cancer. This regimen was, however, found to be associated with grade 3/4 myalgia in 50% of patients. In another phase 2 clinical trial, bryostatin did not find any substantial single-agent clinical activity in colon cancer.

R-roscovitine (Seliciclib; CYC202; Cyclacel Pharmaceuticals) is also an inhibitor of CDK2, CDK7, and CDK9. One phase 1 trial was completed and found that nausea, vomiting, asthenia, and hypokalemia were DLTs.⁹² One patient with HCC experienced a PR.

Other CDK2 inhibitors include SNS-032 (Sunesis Pharmaceuticals, South San Francisco, CA)⁹³ (formerly BMS-387032) and imidazopyridines⁹⁴ that are undergoing early phases of development. Several other new cell-cycle inhibitors are currently in various stages of development (Table 1).

Conclusion and Future Directions

It is well documented that abnormal cell division is the hallmark of the initiation and progression of cancers. The role of cell-cycle inhibition as an effective and durable cancer treatment modality remains yet to be fully elucidated, in part because of the complexity of the cell cycle and the known compensatory function of some of the cyclins and CDKs.⁸⁸ Although cell-cycle abnormalities are commonly observed in GI tumors, to date, efficacy of CDKis was at best modest in these tumors. Flavopiridol has been the most extensively studied CDKi and has shown disappointing results both alone and in combination therapy. Nevertheless, other cell-cycle inhibitors may prove to be clinically useful as monotherapy or in various combinations. PD-0332991 has shown encouraging activity as monotherapy in liposarcoma and non–small cell lung cancer and in combination with letrozole in breast cancer.95, 96, 97 Further research is also needed to explore the use of cell-cycle inhibitor-containing combinations in GI tumors, to identify the patients most likely to benefit, and to determine the mechanisms of resistance to this novel class of drugs. Finally, one of the most important obstacles to the effective application of personalized medicine has been the inability to perform meaningful testing for drug sensitivity on the patient's own tumor cells. This unmet need was recently addressed by a powerful new epithelial cell culture technique developed at Georgetown University and the National Institutes of Health. Termed conditionally reprogrammed cells, this approach allows for the rapid and prolonged culturing of primary epithelial cells and is now applied to a wide variety of both normal epithelium and epithelial cancers.98, 99, 100 In fact, we have recently applied the conditionally reprogrammed cell approach to successfully diagnose and treat a patient who was succumbing to a rare, malignant form of recurrent respiratory papillomatosis.¹⁰¹ The patient had endured 350 operations to clear his respiratory track of cancerous obstructions and had failed numerous other attempts at therapeutic intervention. Within 14 days of initial culturing of the patient's normal and malignant cells, the causative underpinning of the disease was defined, and, importantly, a drug approved by the US Food and Drug Administration was identified (vorinostat) which selectively killed the tumor cells and not the normal epithelium at therapeutic doses. The patient responded extremely well to vorinostat treatment and remains alive to date, a major advancement because the window for surgical intervention was closing as a result of the scarring associated with the previous operations. Obviously, much emphasis is being placed on validating the possibility of a broad application of this approach to many tumor types, including GI cancers, and we have recently reported on the use of prostate-derived conditionally reprogrammed cells for the testing of novel therapeutic compounds.100, 102 Should the conditionally reprogrammed cell approach continue to prove to be a robust method for drug sensitivity testing, the possibility exists that patient outcomes, similar to those described for the papillomatosis patient above, may become a reality for rapidly testing a GI patient's tumor cells against a battery of CDKis and other therapeutic compounds.