A Nonrandomized, Phase II Study of Sequential Irinotecan and Flavopiridol in Patients With Advanced Hepatocellular Carcinoma

ABSTRACT

BACKGROUND:

Flavopiridol, a Cdk inhibitor, potentiates irinotecan-induced apoptosis. In a phase I trial of sequential irinotecan and flavopiridol, 2 patients with advanced hepatocellular carcinoma (HCC) had stable disease (SD) for ≥14 months. We thus studied the sequential combination of irinotecan and flavopiridol in patients with HCC.

METHODS:

Patients with advanced HCC naïve to systemic therapy, Child-Pugh ≤B8, and Karnofsky performance score (KPS) ≥70% received 100 mg/m² irinotecan followed 7 hours later by flavopiridol 60 mg/m² weekly for 4 of 6 weeks. The primary end point was an improvement in progression-free survival at 4 months (PFS-4) from 33% to 54%, using a Simon's two-stage design. Tumors were stained for p53.

RESULTS:

Only 16 patients in the first stage were enrolled: median age, 64 years; median KPS, 80%; Child-Pugh A, 87.5%; and stage III/IV, 25%/75%. The primary end point was not met; PFS-4 was 20%, leading to early termination of the study. Ten patients were evaluable for response: 1 had SD >1 year and 9 had disease progression. Grade 3 fatigue, dehydration, diarrhea, neutropenia with or without fever, lymphopenia, anemia, hyperbilirubinemia, and transaminitis occurred in ≥10% of the patients. Of the 9 patients who progressed, 5 had mutant p53 and 4 had wild-type p53. The patient with stable disease had wild-type p53.

CONCLUSION:

Sequential irinotecan and flavopiridol are ineffective and poorly tolerated in patients with advanced HCC. Despite our limited assessments, it is possible that the presence of wild-type p53 is necessary but not sufficient to predict response in HCC.

Cyclin-dependent kinases (Cdks) are integral components of the cell cycle regulatory apparatus. Derangements in Cdk activity lead to cell cycle disinhibition, which is one of the hallmarks of malignancy.¹ In vitro studies have shown that Cdk inhibition leads to apoptosis.^2,3 Cell cycle dysregulation resulting from the loss of the Cdk inhibitors p16INK4A and p27 has been implicated in hepatocarcinogenesis.^4,5 These observations support a rationale for Cdk targeting in hepatocellular carcinoma (HCC).

Flavopiridol is a semisynthetic compound derived from the bark of the Disoxylum binectariferum plant found in India.⁶ It is a prototype competitive inhibitor of Cdk-1, -2, -4, -6, -7, and -9, inducing cell cycle arrest at the G1 or the G2/M transition point.⁷ It has also been shown to exhibit proapoptotic and antiangiogenic properties.⁸ When administered in a sequence-dependent fashion, flavopiridol has been shown to enhance the cytotoxicity of various chemotherapies, including taxanes and gemcitabine.^9,10 The hypothesized mechanisms underlying these interactions are thought to result from the crippling effects of chemotherapy on the cell cycle machinery that are subsequently potentiated by flavopiridol—a therapeutic “two-hit” event. At the molecular level, chemotherapy primes tumor cells by stimulating the expression of specific Cdks and/or apoptotic mediators that are then targeted by flavopiridol.⁸

The alterations in tumor cell cycle biology are illustrated by the sequential combination of the topoisomerase I inhibitor irinotecan with flavopiridol, which has been extensively studied by our group. Early preclinical studies in Hct116 colorectal cell lines showed that SN-38, the major metabolite of irinotecan, exerted a cytostatic effect in association with upregulation of p21, p53, and Drg1. Apoptosis was induced with the subsequent administration of flavopiridol, with maximum cell kill occurring when the drug was given 7 or 16 hours later.^11,12 Mechanistic studies in Hct116 cell lines have shown that flavopiridol suppresses homologous recombination repair in a p53-dependent manner, enhancing SN-38 cytotoxicity.¹³ These observations led to the development of a time- and sequence-dependent treatment schedule consisting of intravenous (IV) irinotecan followed 7 hours later by IV flavopiridol given weekly for 4 weeks, recycling every 6 weeks. This regimen was evaluated in a phase I trial conducted at Memorial Sloan-Kettering Cancer Center in patients with solid, treatment refractory, mainly gastrointestinal malignancies.¹⁴ One third of the patients experienced disease control, including 2 with HCC who had stable disease lasting more than 12 months. On the basis of these intriguing results generated in the presorafenib era, we elected to conduct a nonrandomized, single-arm, phase II trial studying the use of this regimen exclusively in patients with treatment-naïve advanced HCC.

PATIENTS AND METHODS

Inclusion and Exclusion Criteria

Patients ≥18 years of age with pathologically confirmed advanced HCC, Child's-Pugh score, ≤B8; Karnofsky performance score (KPS), ≥ 70%; and adequate hematologic (leukocytes ≥3,000/μL, neutrophils ≥1,500/μL, and platelets ≥75,000/μL), renal (normal creatinine or creatinine clearance, >60 mL/min/1.73 m²), and hepatic (AST and ALT, ≤2.5 × upper limit or normal) were eligible to participate. No prior systemic chemotherapy or biologic therapies for advanced disease were permitted. Prior surgery and liver-directed ablative therapies, but not external beam radiotherapy, of target lesions were allowed as long as subsequent disease progression in those regions, defined by revised World Health Organization (WHO) criteria,¹⁵ was present. Transplant recipients; patients with known brain metastases, a history of prior malignancy, clinically significant gastrointestinal bleeding within ≤1 month of study entry, known allergy to flavopiridol or irinotecan-related compounds, venous thromboembolism within ≤6 months of study entry, and intercurrent active systemic illness; and immunosuppressed and/or pregnant patients were excluded.

The protocol was reviewed and approved by the Institutional Review Board at Memorial Sloan-Kettering Cancer Center. All patients provided written informed consent. This study was conducted and its participants were protected in accordance with the Declaration of Helsinki.

Study Therapy and Dose Adjustments

Patients received IV irinotecan 100 mg/m² infused over 30 minutes, followed 7 hours later by IV flavopiridol 60 mg/m² given over 60 minutes. Both drugs were administered on days 1, 8, 15, and 22 of a 6-week cycle (ie, 4 weeks on/2 weeks off).

Dose delays or modifications for both irinotecan and flavopiridol were indicated for grade 3/4 hematologic and nonhematologic toxicities. The irinotecan dose was reduced by 20-mg/m² decrements to a minimum dose of 60 mg/m². The dose of flavopiridol was reduced to 50 mg/m². Study therapy was resumed once toxicities had improved to grade 1 or less. A maximum of 2 dose reductions for any reason was permitted before patients were removed from the study.

Response Assessment

Radiographic evaluation with computed tomography (CT) or magnetic resonance imaging (MRI) was performed after every 2 cycles of therapy. Responses were assessed by study investigators and were independently reviewed and reported by the study radiologist (LS), according to the WHO criteria,.¹⁵

Determination of p53 Mutation Status

The protocol was amended in April 2005 to include an assessment of p53 mutation status in tumor specimens. Pretreatment formalin-fixed paraffin-embedded tissue blocks were cut into 5-μm thin sections for immunohistochemistry (IHC). Two slides per patient were stained for p53 with a monoclonal antibody (PAb1801; Oncogene Calbiochem, Cambridge, MA) at a concentration of 0.2 μg/mL. Epitope retrieval was performed using a heat-induced, standard avidin-biotin peroxidase method with 0.01 mol/L citric acid (pH 6). Positive and negative controls were performed with each experiment. Specific staining for p53 was characterized by nuclear immunoreactivity. The slides were assessed for positive nuclear staining of 100 cells in 5 randomly selected high-powered fields. Mutant, or positive, p53 staining was defined by immunoreactivity in >20% of nuclei.¹⁴

Statistical Analysis

All statistical calculations were performed by using an intention-to-treat analysis. The primary end point of the study was progression-free survival at 4 months (PFS-4), calculated from the date that protocol treatment was started until the date of death or date of first evidence of radiographic disease progression, according to the revised WHO criteria.¹⁵ The smallest measurements recorded since the treatment started was taken as the reference. We hypothesized that study therapy would improve PFS-4 from a historical control of 33% to 54%.¹⁶ A projected enrollment of 32 patients was necessary to achieve 90% power with an 11% probability of 1-sided type I error.

PFS-4 was estimated with a Simon's minimax 2-stage design with a planned interim analysis at 6 months or after 16 patients had been enrolled, whichever came later. If PFS-4 was >33%, study enrollment would continue. A PFS-4 of <33% would indicate inefficacy of the investigational therapy, and the study would be terminated.

The relationship between p53 mutation status and clinical outcomes was analyzed in an exploratory fashion. Overall survival was calculated using the Kaplan-Meier method.

RESULTS

Patient and Disease Characteristics

Sixteen patients were enrolled in the study. The median age at HCC diagnosis was 64 years (range, 26–84). Ten (63%) patients were male, 14 (87.5%) were Caucasian, and 2 (12.5%) were Asian.

Three (19%) patients had viral hepatitis B as a risk factor for HCC, and 1 also had alcoholic cirrhosis and hepatitis C. Two (12.5%) other patients had hepatitis C, with 1 patient also having hemochromatosis. Five (31%) patients had a metabolic disorder including nonalcoholic steatohepatitis (NASH), diabetes, and hyperlipidemia. Of those, 1 (6%) also had a history of alcohol abuse. One (6%) patient had hemochromatosis alone, and 4 (25%) patients did not have any identifiable risk factors for HCC.

At study enrollment, the median KPS was 80% (range, 70%–90%), 14 (87.5%) patients had Child-Pugh A liver cirrhosis, and 12 (75%) had TNM stage IV disease. Eleven (69%) patients had undergone surgery for HCC; of those, 6 also underwent hepatic arterial embolization (HAE), either before surgery or for recurrent disease. Three (19%) underwent HAE and/or radiofrequency ablation alone. Two (12.5%) patients had not received any therapy before enrolling in the study.

Patient characteristics at study enrollment are outlined in Table 1.

Table 1.

Patient characteristics

Characteristic	N = 16
Median age, y (range)	64 (26–84)
Sex, n (%)
Male	10 (62.5)
Female	6 (37.5)
Karnofsky performance score, % (range)	80 (70–90)
Race/ethnicity n (%)
Caucasian	14 (87.5)
Asian	2 (12.5)
Child-Pugh score n (%)
A	14 (87.5)
B7-8	2 (12.5)
Disease stage n (%)
III	4 (25)
IV	12 (75)
Risk factors, n
Hepatitis B	4
Hepatitis C	3
Plus hemochromatosis	1
Metabolic disorder	5
Plus alcohol use	1
Hemochromatosis (1 with Hepatitis C)	2
Unidentified	4
Prior therapy, n (%)
Surgery only	5 (31)
Liver-directed ablative therapy (HAE, RFA) only	3 (19)
Surgery and ablative therapy	6 (37.5)
None	2 (12.5)

HAE = hepatic arterial embolization; RFA = radiofrequency ablation.

Patient Disposition and Treatment

Of the 16 patients enrolled, 2 progressed before starting therapy and were excluded. The 14 patients who proceeded with therapy received a median of 2 cycles (range, 1–8) of treatment. One of these patients was removed from the study after 1 cycle when pathologic revision of the diagnosis revealed metastatic neuroendocrine carcinoma.

Efficacy

Of the remaining 13 patients, 3 were not evaluable for response because of consent withdrawal (n = 1) and toxicity (n = 2). Of the remaining 10 patients, 1 (7%) had stable disease lasting 13 months, 8 (61.5%) had disease progression, and 1 (7%) had progression and died during the study.

The primary end point of the study was not met; PFS-4 was 20% (Figure 1). The trial was therefore terminated for inefficacy at the end of the first stage. Median overall survival for the entire cohort was 9.4 months.

Figure 1.

Time to progression with the irinotecan and flavopiridol regimen.

p53 Mutation Status and Response

IHC for p53 was performed on the pathology of the 10 evaluable patients for response. Five had mutations and 5 were wild-type for p53. Five patients with mutations and 4 with wild-type p53 had progression of disease while on protocol therapy. The patient with stable disease had wild-type p53. These relationships are shown in Table 2.

Table 2.

Clinical benefit and p53

Outcome	p53 Wild-type	p53 Mutation
Stable disease, n	1	0
Progression of disease, n	4	5

Toxicity

Fourteen patients were evaluable for toxicity. Two required dose reductions as a result of toxicity. Treatment-related grade 3 toxicities occurring in ≥10% of patients were neutropenia (28%), febrile neutropenia (21%), lymphopenia, (21%), anemia (14%), diarrhea (21%), dehydration (21%), hyperbilirubinemia (21%), transaminitis (21%), and fatigue (21%). One (7%) patient each had grade 4 neutropenia, dehydration, and transaminitis. Three patients with grade 3 neutropenia or lymphopenia also had episodes of grade 3 hyperbilirubinemia, but there was no strict temporal relationship between episodes as previously described.¹² The patient who experienced long-term stable disease had 8 episodes of grade 3/4 neutropenia, 4 of febrile neutropenia, and 1 of grade 3 hyperbilirubinemia. Other grade 3/4 toxicities unrelated to treatment were hyponatremia (21%), elevated INR (21%), hypoalbuminemia (14%), syncope (7%), hypoxia and dyspnea due to a respiratory infection (7%), hyperglycemia (7%), hypophosphatemia (7%), non-neutropenic fever (7%), and adrenal insufficiency (7%). Treatment-related toxicities are outlined in Table 3.

Table 3.

Treatment-related grade 3/4 toxicities

Toxicity	Grade 3	Grade 4
(N = 14)	n (%)	n (%)
Fatigue	3 (21)
Neutropenia	4 (28)	1 (7)
Febrile neutropenia	3 (21)
Lymphopenia	3 (21)
Anemia	2 (14)
Diarrhea	3 (21)
Dehydration	3 (21)	1 (7)
Hyperbilirubinemia	3 (21)
Aspartate aminotransferase (AST)	3 (21)	1 (7)

DISCUSSION

In this phase II study of sequential flavopiridol and irinotecan, there was no improvement in 4-month progression-free survival when compared with a historical control inferred from a phase II study of irinotecan monotherapy in HCC given at 125 mg/m² weekly for 4 weeks, followed by 2 weeks of rest.¹⁶

We did not identify any molecular markers that could distinguish the single patient who responded from those who progressed. Specifically, in contrast to other data, p53 mutation status was not associated with sensitivity to study therapy.¹⁴

Grade 3 and 4 gastrointestinal and myelotoxicities were frequent. It is unclear whether decreased hepatic functional reserve in this population played a role in the latter regard. Given the previously documented relationship between baseline and posttreatment hyperbilirubinemia, myelosuppression, and response to therapy,¹⁴ we analyzed our study population for a similar pattern but were unable to demonstrate any correlation between toxicity and response.

The therapeutic inefficacy of this drug combination might have been the result of inadequate intracellular drug concentrations. Chemoresistance in HCC has been attributed to tumor overexpression of the multidrug resistance gene (MDR)-1 and the P-glycoprotein drug-extrusion molecule.¹⁷ Drug concentrations of flavopiridol may have been reduced as a result of high protein binding.¹⁴ Pharmacokinetic studies would have been a valuable adjunct to this study in retrospect and might have provided some insight into this possibility.

There are several potential explanations for the absence of a relationship between p53 mutation status and clinical benefit from study therapy. In the previously reported phase I trial, p21 and differentiation-related gene-1 (Drg1), and p53, were measured in both pre- and posttreatment tumor specimens.¹⁴ Patients with wild-type p53 who had constant or decreased p21 and Drg1 levels after treatment experienced stable disease or a partial response, whereas those with mutated p53 or increased posttreatment p21 and Drg1 levels progressed.¹⁴ As the current study did not examine p21 or Drg1 levels or obtain posttreatment specimens, it remains unknown whether treatment inefficacy was related to unfavorable changes in the cell cycle machinery. In addition, the use of IHC for the determination of p53 mutation status has limited accuracy; when compared with direct sequencing, discordance rates range from 20% to 50%.^18,19 Since confirmatory DNA sequencing was not performed, it is possible that p53 mutation status was incorrectly classified in some of our patients. Furthermore, the primary disease sites were not specified for the patients whose specimens were tested for biomarkers in the phase I study,¹⁴ and so it is possible that the molecular associations reported therein may not be applicable to patients with HCC.

Most of the patients enrolled in this study underwent surgery and/or liver-directed ablative therapies for earlier stage disease. As p53 immunostaining was not necessarily performed on pretreatment specimens, this raises the question of whether therapy may have altered the molecular profile of the disease. In breast cancer, changes in hormone and Her-2 receptor status have been reported after treatment.^20,21 Since hypoxia can induce Drg1 in HCC²² and ischemia is a putative mechanism of cytotoxicity caused by HAE, hypoxia-induced Drg1 upregulation compounded by the effects of irinotecan may have exceeded the inhibitory capacity of flavopiridol. It is also conceivable that the molecular profile of HCC will evolve as the disease progresses from early to recurrent and metastatic stages. Thus, p53 and its associated network in early stage, untreated HCC may not be useful as a predictive biomarker of response to irinotecan and flavopiridol in those with heavily treated, advanced HCC.

Yet another consideration is whether the choice of study end point was appropriate. Although it is unlikely that an alternative end point would have yielded different results, it is recognized that composite end points such as PFS do not isolate the effect of the experimental therapy on outcome, especially in a disease like HCC, in which cirrhosis and survival are closely intertwined.²³ It is also questionable as to whether a PFS benefit translates into an overall survival benefit in HCC. Median survival for this cohort was 9.4 months, which is comparable to survival durations reported in the initial phase II study of sorafenib as well as the SHARP trial.^24,25 Eight patients received postprogression therapy including various chemotherapies, sorafenib, and liver-directed ablative therapies. Thus, although PFS was poor, it appears that patients were salvaged by postprogression therapy.

Although the results were negative and the molecular correlatives could not be meaningfully interpreted, this study highlights the need for continued efforts in this arena. The potential value of biomarker development is illustrated by the recently reported phase II trial of second line c-Met inhibition with tivantinib, which showed greater benefit in patients with c-Met-positive vs. c-Met-negative HCC.²⁶ It is also worth noting that our study population was heterogeneous, with a slight predominance of metabolic risk factors for cirrhosis over viral hepatitis. Although the sample size precluded an analysis of outcomes by risk factors, the elucidation of molecular differences by etiology represents another important area for further study.

In conclusion, sequential irinotecan and flavopiridol does not have clinically relevant antineoplastic activity and was cumbersome to administer and poorly tolerated in patients with advanced HCC and preserved hepatic function. In contrast to previous reports, wild-type p53 and the development of hepatobiliary and/or myelotoxicities were not associated with favorable outcomes with the study regimen. Although it is clear that p53 assessed by IHC is not by itself predictive of response, it is possible that wild-type p53 is necessary but not sufficient to predict response to this regimen in HCC.