Introduction

Whereas normal hepatic tissue is perfused predominantly from the portal vein, the blood supply to hepatic metastases comes only from the hepatic artery [1, 2]. Therefore, researchers have explored hepatic arterial infusion (HAI) of chemotherapeutic agents as an appropriate therapeutic strategy for patients with unresectable liver metastases [3]. Compared with intravenous (IV) administration, direct HAI produces a higher drug concentration at the tumor site and attenuates systemic adverse effects in normal tissues [4]. Among the many chemotherapeutic agents, investigators have examined paclitaxel given via regional arterial infusion because of its broad antitumor activity and prolonged chemical stability [5, 6]. Authors reported stable disease for at least 6 months and objective responses in 60 % of breast cancer patients with predominantly hepatic metastases who received HAI of paclitaxel at 200 mg/m2 over 24 h, including one patient whose cancer was controlled for 48 months [7].

Nanoparticle albumin-bound paclitaxel (nab-paclitaxel, Abraxane) may be more suitable for HAI than regular paclitaxel because of greater first-pass hepatic extraction of the former, generating tissue distribution to a greater extent over a shorter time frame [8–10]. Nab-paclitaxel capitalizes upon albumin binding to its receptor (gp60) on endothelial cells, which initiates transcytosis of free and albumin-bound plasma constituents across the endothelial cell into the extravascular space via caveolae [11–13] and accumulates in secreted protein acidic rich in cysteine-positive cells [14], escaping from elimination by P-glycoprotein (P-gp)-mediated efflux [15, 16].

Despite the anticipated benefits of HAI of nab-paclitaxel for cancer therapy, only 1 of 12 breast cancer patients with predominantly hepatic metastases experienced partial remission after receiving it over 1 h in a phase I trial [17]. Preliminary pharmacokinetic assessment revealed an apparent disadvantage of HAI of nab-paclitaxel over 1 h in that it led to approximately 42 % hepatic extraction of the drug based on peripheral blood mean systemic exposures (area under the curve [AUC]) [17], which was markedly lower than the reported 95 % mean hepatic extraction during a 24-h HAI of paclitaxel [18]. The evidence prompted us to hypothesize that longer HAI delivers a higher dose of nab-paclitaxel, resulting in greater hepatic extraction of the drug. To test this hypothesis, we amended a pilot phase I study of HAI of nab-paclitaxel (clinicaltrials.gov registration number NCT00732836) to administer to all patients prolonged HAI of nab-paclitaxel over 4 h at dose levels higher than the US Food and Drug Administration-recommended intravenously infused dose [19]. As described herein, we evaluated the safety, antitumor activity, and pharmacokinetic characteristics of nab-paclitaxel given via HAI in patients with advanced cancer having predominantly hepatic metastases.

Materials and methods

Patient enrollment and assessment

To be eligible for this phase 1 trial, patients at The University of Texas MD Anderson Cancer Center (MD Anderson) had to have predominantly hepatic metastases, defined as at least 40 % of the total tumor burden involving the liver, and had no standard-of-care therapy. Other eligibility criteria included an Eastern Cooperative Oncology Group (ECOG) performance status of no more than 2, an absolute neutrophil count of at least 1500/mm3, a platelet count of at least 100,000/mm3, a creatinine level no greater than 2 mg/dL, an alanine aminotransferase level no greater than five times the upper limit of normal, a bilirubin level no greater than 2 mg/dL, the ability to understand and willingness to sign an institutional review board-approved informed consent form, and full recovery from all previous therapies. Patients were excluded if they had clinically significant ascites, were pregnant, were breast-feeding, had untreatable bleeding diathesis, had evidence of portal vein thrombosis and clinically significant peripheral vascular disease, had at least grade 2 neuropathy, or had uncontrolled central nervous system metastasis. Concomitant use of known inducers or inhibitors of CYP2C8 and CYP3A4 was limited [17]. All patients were assessed according to the Common Terminology Criteria for Adverse Events (CTCAE) version 3.0 [20] and the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.0 [21]. This phase I trial 2007-0857 (clinicaltrials.gov registration number NCT00732836) was conducted at The University of Texas MD Anderson Cancer Center after the approval by the Institutional Review Board.

Treatment and pharmacokinetic sampling

After completing the informed consent process and registration, eligible patients were admitted to the inpatient service at MD Anderson after consultation with the interventional radiology service. HAI of nab-paclitaxel was administered at one of the six dose levels listed in Table 1 [17]. On the day of treatment, patients underwent angiographic placement of the catheter for HAI into a hepatic artery via femoral artery. Arteriography (using a non-iodinated contrast medium) of the celiac axis and superior mesenteric artery was performed to delineate the anatomy of hepatic artery distribution, identify accessory arteries, and confirm adequate portal venous flow and positioning of the catheter. If indicated, embolization was done to alter collateral blood flow patterns or to protect gastrointestinal organs from HAI [22]. After catheter placement, patients were taken to the nuclear medicine department for a catheter flow study. Once patients were transported to the designated floor, treatment was initiated with the assigned dose of nab-paclitaxel plus 1500 IU of heparin administered intraarterially over 4 h without prophylactic medications. The hepatic intraarterial catheter was removed once the intraarterial infusion was completed. The treatment was repeated once every 3 weeks.

Table 1 Baseline demographic data (n = 48)
Full size table

At dose levels of at least 260 mg/m2, two sets of pharmacokinetic studies were performed to compare peak drug concentrations and systemic exposure between IV infusion and HAI of nab-paclitaxel. At dose levels greater than 260 mg/m2, only one set of pharmacokinetic studies was performed during HAI, as IV infusion of nab-paclitaxel at these dose levels might not have been safe. Approximately 4 mL of whole blood sample was obtained at each time point as designed below during either IV administration of nab-paclitaxel or HAI of the first dose of nab-paclitaxel using an indwelling venous catheter placed in the arm (in the contralateral arm in patients receiving IV nab-paclitaxel), as illustrated in Fig. 1. For patients receiving 1-h HAI or IV infusion of nab-paclitaxel, blood samples were collected before and 5, 10, 20, and 40 min, 1 (within 5 min of the end of 60-min infusion), 1.5, 2, 3, 4, 6, and 24 h (on day 2) after the initiation of the infusion. For patients receiving 4-h HAI or IV infusion of nab-paclitaxel, blood samples were collected before and 0.5, 1, 2, 3, 4 (within 10 min of the end of 4-h infusion), 4.5, 5, 6, 7, 9, and 24 h (on day 2) after the initiation of the infusion. The collected blood samples were processed immediately, and the resultant plasma samples were stored at −70 °C for future analysis as described below.

Fig. 1
figure 1

Schema of the difference in uptake of nab-paclitaxel in the liver between the intravenous (IV) infusion and hepatic arterial infusion (HAI). With IV infusion, nab-paclitaxel was administered from one arm to the right heart, lungs, left heart, and the other arm. With HAI, it was administered from the liver, to the right heart, lungs, left heart, and arm

Full size image

Pharmacokinetic analyses

Analytes were extracted from plasma samples using solid-phase extraction, eluted with acetonitrile, evaporated using nitrogen gas, and reconstituted. Chromatographic separation was achieved using high-performance liquid chromatography, and detection was performed using tandem mass spectrometry with electrospray atmospheric pressure ionization. The lower limit of quantification for paclitaxel was 1 ng/mL, and the linear calibration range is 1–1000 ng/mL [10, 23, 24].

All drug concentration time data were analyzed in individual patient datasets. Pharmacokinetic parameter estimates for the individual datasets were generated via non-compartmental analysis using the WinNonlin software program (version 5.2, Certara, Princeton, NJ). The peak paclitaxel concentration (C max) and the corresponding time to peak (T max) were observed values. The elimination rate constant was calculated using log-linear regression analysis of the terminal phase of the whole blood paclitaxel concentration versus the time profile. The elimination half-life (t 1/2) was computed as ln2 divided by the elimination rate constant. The AUC from time 0 to infinity (AUCinf) was determined via summation of the AUC from time 0 to the last measurable concentration (calculated using the log-linear trapezoid rule) and the AUC of the extrapolated area (estimated by dividing the last measurable concentration by the elimination rate constant). The dose–area relationship (total drug dose divided by the AUCinf) was used to determine total-body clearance (CL). The volume of distribution (Vd) was calculated by dividing the CL by the elimination rate constant. Descriptive summary statistics were used to assess patient demographics and pharmacokinetic parameters. Categorical data were summarized using frequencies and percentages. Continuous data were summarized using means, medians, ranges, and coefficients of variation ± standard deviations. Statistical inferences were based on two-sided tests at a significance level of p < 0.05.

Results

Patient characteristics and assessment

We recruited 48 patients advanced cancer having predominantly hepatic metastases and meeting the inclusion and exclusion criteria into this study according to the 3 + 3 design including dose escalation and expansion in patients with responsive tumor types. The patients’ baseline characteristics are listed in Table 1. Thirty-eight patients underwent HAI of nab-paclitaxel over 1 h at three dose levels, whereas the remaining 10 patients did so over 4 h at three higher dose levels. We observed the antitumor activity in the patients, none of whom experienced dose-limiting toxicities or treatment-related death as described previously [17].

Additional 10 patients received a total of 34 cycles of therapy. No objective response was observed. Two patients had stable disease for at least 4 months: 1 patient with carcinoma of the ampulla of Vater and stable disease for 4.2 months, and another with non-small cell lung cancer and stable disease for 6.4 months. No dose-limiting toxic effects or treatment-related deaths were observed. One patient required a dose reduction to 260 mg/m2 for grade 3 thrombocytopenia and grade 2 hyperbilirubinemia after 2 cycles of therapy at 280 mg/m2.

Pharmacokinetic evaluation

To evaluate hepatic extraction of nab-paclitaxel administered using HAI over IV infusion, we used a safe, feasible method of collecting two sets of peripheral blood samples for pharmacokinetic comparison. We collected 44 sets of evaluable blood samples from 27 patients: 19 from patients receiving IV infusion of nab-paclitaxel, and 25 sets from patients receiving HAI of it. Tables 2 and 3 list data on nab-paclitaxel CL, t1/2, Vd, C max, and AUC [17]. Compared with IV infusion of nab-paclitaxel, HAI of nab-paclitaxel over 1 h led to markedly lower C max and AUC values. Also, patients receiving HAI of nab-paclitaxel over 4 h had markedly lower C max, but similar AUC values when compared with those receiving HAI of nab-paclitaxel over 1 h.

Table 2 Pharmacokinetic assessment of nab-paclitaxel delivered via IV infusion
Full size table
Table 3 Pharmacokinetic assessment of nab-paclitaxel delivered via HAI
Full size table

Figure 2 shows mean paclitaxel blood concentration–time plots in patients receiving IV infusion or HAI of nab-paclitaxel over 1 or 4 h. All these patients exhibited rapid distribution and biphasic elimination of nab-paclitaxel regardless of the route or duration of infusion. As expected, 4-h IV infusion of nab-paclitaxel led to a markedly lower C max associated with a slightly longer T max when compared with 1-h IV infusion. In patients receiving HAI of nab-paclitaxel, 4-h infusion resulted in a markedly lower C max and longer T max than did 1-h infusion, suggesting a greater hepatic first-pass effect associated with the longer HAI.

Fig. 2
figure 2

Mean nab-paclitaxel blood concentration–time profiles after intravenous infusion (a) and hepatic arterial infusion (b) over 1 and 4 h

Full size image

Hepatic extraction of nab-paclitaxel after HAI

As shown in Fig. 1, peripheral blood concentrations of nab-paclitaxel may have reflected actual hepatic venous blood concentrations of the drug, assuming that pulmonary uptake of nab-paclitaxel was negligible. Table 4 demonstrates the hepatic extraction of nab-paclitaxel after HAI according to the formula AUC0-inf = 1 − (AUC0–inf[HAI]/AUC0–inf[IV]) [17]. To limit the influence of drug distribution and elimination, we use the following formulas for hepatic extraction calculation: 1 − (AUC0–1 h[HAI]/AUC0–1 h[IV]) for 1-h infusion and 1 − (AUC0–4 h[HAI]/AUC0-4 h[IV]) for 4-h infusion. Although HAI of nab-paclitaxel over 4 h produced markedly lower C max and longer T max values, 4-h HAI was not associated with increased hepatic extraction of nab-paclitaxel according to the calculations based on AUC described above.

Table 4 Hepatic extraction of nab-paclitaxel administered via HAI over IV
Full size table

Discussion

This pilot study of HAI of nab-paclitaxel in cancer patients with predominantly hepatic metastases demonstrated that the treatment was well tolerated and had antitumor activity [17]. Pharmacokinetic evaluation displayed that HAI of nab-paclitaxel led to lower C max and AUC values than those achieved with IV infusion and that HAI of nab-paclitaxel over 4 h resulted in lower C max and longer T max values than those achieved with HAI over 1 h, providing evidence for enhanced hepatic extraction of nab-paclitaxel with prolonged HAI. However, we did not observe increased hepatic extraction of nab-paclitaxel after HAI over 1 or 4 h when we used the formula based on total or IV AUC for hepatic extraction calculation.

In cancer patients with extensive hepatic metastases, surgical resection, irradiation, and transcatheter arterial chemoembolization are largely ineffective. Uncontrolled local disease eventually evolves to become fatal in the majority of these patients. Efficacious regional therapy for local and systemic control of tumor progression is an appropriate option to provide symptomatic palliation and a survival benefit in patients whose extensive hepatic metastases are lethal. Accordingly, researchers have explored clinical development of HAI of cytotoxic agents for its apparent therapeutic advantages over IV infusion: higher local drug concentrations [25] and lower systemic exposure [26].

We made several important observations in this study. First, short HAI therapy has its own apparent advantages in daily clinical practice if its antitumor activity is not compensated. However, prolonged HAI may increase hepatic extraction of paclitaxel and nab-paclitaxel, which in turn enhances its antitumor activity. About 5 % of patients had partial responses, and 8 % had stable disease at least 6 months after HAI of nab-paclitaxel over 1 h as described previously [17], whereas 30 % had partial responses and 30 % had stable disease at least 6 months after HAI of paclitaxel over 24 h in another study [7]. Second, prolonged HAI of nab-paclitaxel over 24 h is feasible but may be impractical in routine clinical care because of the relatively short chemical stability of nab-paclitaxel. A reconstituted suspension of nab-paclitaxel in a vial may be placed in the original carton to protect it from bright light at 2–8 °C for a maximum of 8 h if it is not used immediately. At ambient temperatures and under ambient lighting conditions, reconstituted nab-paclitaxel in an infusion bag may be stored for up to 4 h [27]. Third, pharmacokinetic evaluation of hepatic extraction of nab-paclitaxel is complicated. Although delayed T max and decreased C max of nab-paclitaxel associated with prolonged HAI suggested enhanced hepatic extraction of nab-paclitaxel, AUC-based assessment of hepatic extraction did not support these findings. The effectiveness of regional drug delivery via HAI to the sites of predominantly hepatic metastases can be measured in several ways. The direct way is to measure tissue levels of drugs in biopsy specimens collected a short time after infusion, although biopsy carries significant risks of the patient. An alternative method is to collect serial blood samples from the hepatic artery and vein during and after the infusion, which also carries risk related to placement of a hepatic vein catheter and the possibility of drug contamination of an arterial port owing to insufficient flushing prior to sampling. Further preclinical and clinical studies are required to develop reliable methods of hepatic extraction evaluation associated with passive and active uptake, which may be both time and concentration dependent [11–14, 28, 29].

When considering the clinical relevance of our findings, several limitations should be kept in mind. First, our small sample size resulting from different confounding factors limits the generalizability of our findings, a limitation of many clinical trials as well. Second, the ultimate purpose of the cancer therapy using different routes and durations of administration is to enhance drug uptake in the tumor cells. This depends on not only passive uptake through increased local drug concentrations, but also active uptake through their intrinsic drug transportation system. Differences in drug distribution in the liver between IV administration and HAI may result from the differences in the biological properties of the drug and the microenvironmental characteristics of different tissues. From this point of view, changes in infusion route and duration may not be critical to increasing hepatic extraction. Third, tumor responses to treatment with cytotoxic agents are not always dose dependent, as many resistant mechanisms cannot be overcome by solely increasing local drug concentrations [30, 31]. Although previous studies demonstrated a survival benefit of regional therapy with other cytotoxic agents [32–34], convincing evidence supporting an advantage of HAI of nab-paclitaxel over IV infusion in patients with predominant hepatic metastases has yet to be obtained.

HAI of nab-paclitaxel at doses of up to 300 mg/m2 over 4 h is safe in patients with predominantly hepatic metastases. We observed no dose-limiting toxic effects or treatment-related deaths. Pharmacokinetic studies revealed that prolonged HAI of nab-paclitaxel over 4 h resulted in markedly lower C max and longer T max than did 1-h HAI. Also, HAI of nab-paclitaxel led to markedly lower C max and AUC than did IV infusion. The evidence suggested a greater hepatic first-pass effect associated with prolonged HAI of nab-paclitaxel over 4 h than did 1-h HAI and IV infusion. However, AUC-based calculation of hepatic extraction of nab-paclitaxel did not support these findings. Further preclinical and clinical studies are required to develop reliable methods of evaluation of hepatic extraction associated with passive and active uptake, which may be both time and concentration dependent.