Extending Principles Learned in Model Systems to Clinical Trials Design

Introduction

Irinotecan(Drug information on irinotecan) (CPT-11 [Camptosar]), [7-ethyl-10-(4-[1-piperidino]-1-piperidino)-carbonyloxy-camptothecin], formerly known as CPT-11, is one of approximately five camptothecin derivatives currently under clinical investigation. Of these, irinotecan and topotecan(Drug information on topotecan) (Hycamtin) (9-dimethylamino methyl-10-hydroxy camptothecin) have demonstrated significant clinical activity.

Given as a single intravenous dose, irinotecan caused objective responses in previously treated patients with sarcoma of the stomach, melanoma, colon adenocarcinoma, and non-small-cell lung (NSCLC) cancer.[1] Irinotecan, administered weekly by short infusion, has demonstrated antitumor activity against refractory or relapsed small-cell lung cancer (SCLC),[2,3] and responses have been measured in non-Hodgkin’s lymphoma, acute lymphocytic leukemia,[4] NSCLC,[5,6] and colon adenocarcinoma[7-11] with different schedules. In most studies, dose-limiting toxicity has been neutropenia and severe diarrhea.[12]

Despite these encouraging clinical results, however, it is clear that irinotecan (and other camptothecin derivatives) has not fulfilled the expectations predicted by results in preclinical models.[13-22] One explanation (unfortunately offered with increasing frequency) is that the "preclinical xenograft models do not predict for sensitivity of human cancer." In this article, we will discuss whether this explanation is, in fact, correct, or whether the difference between the response rates predicted by preclinical models and current clinical experience may be due to a failure to incorporate some of the principles learned from the models into the design of clinical trials.

Our analysis suggests that if differences in host tolerance (mouse vs human) are taken into account, xenograft models predict clinical response rates quite accurately. Furthermore, if the principles learned in preclinical models are incorporated into the design of clinical trials, they may significantly enhance clinical response rates. Thus, rather than providing an example of how preclinical results have failed to translate to the clinical setting, the camptothecin agents appear to offer a paradigm for better integrated, pharmaco- kinetically driven, preclinical and clinical development of new agents.

Materials and Methods

Immune Deprivation of Mice--Female CBA/CaJ mice, 4 weeks of age, were immune-deprived by thymectomy, followed 3 weeks later by whole-body irradiation (1,200 cGy) using a cesium-137 source. The mice received 3 × 10⁶ nucleated bone marrow cells within 6 to 8 hours of irradiation.[20] Tumor pieces of approximately 3 mm³ were implanted in the space of the dorsal lateral flanks of the mice to initiate tumor growth. Tumor-bearing mice were randomized into groups of seven before therapy was initiated.

Tumor Lines--Each of the xenografts used has been described previously.[20,22]

Growth Inhibition Studies--Each mouse bearing bilateral subcutaneous tumors received the agent when tumors were approximately 0.20 to 1 cm in diameter. The procedures (tumor growth measurement) have been reported previously.[23] Briefly, two perpendicular diameters were determined at 7-day intervals using digital Vernier calipers interfaced with a microcomputer. Tumor volumes were calculated, based on the assumption that the tumors were spherical, using the formula [ (p/6) × d³], where d is the mean diameter. Mice were observed for at least 12 weeks after treatment was started.

Irinotecan Formulation and Administration--Irinotecan powder, provided by The Pharmacia & Upjohn Company (Kalamazoo, Michigan), was dissolved in sorbitol (70% with water) and lactic acid(Drug information on lactic acid). The irinotecan was heated at 65° C for 20 minutes and cooled to room temperature. The pH of the solution was adjusted to 3.9 by adding hydrochloric acid. The solution was diluted in sterile water to 20 mg/mL, filter-sterilized, and stored in foil-wrapped tubes (-20 °C). The drug was further diluted prior to administration (0.1 mL/20 g body weight).

Irinotecan antitumor activity was evaluated on a schedule of intravenous daily administration for 5 days per week for 1 week, or for 2 consecutive weeks [days 1-5, 8-12; abbreviated (d ×5)2] repeated every 21 days for 3 cycles of therapy {[(d ×5)2]3} over 8 weeks. Control mice received vehicle [0.26 mL of sorbitol (70% with water), 0.9 mg of lactic acid per mL, pH 3.9], diluted appropriately.

Sample Collection and Drug Analysis--Following a single intravenous dose of irinotecan (10 mg/kg), blood samples were collected from non-tumor-bearing mice (four animals per time point) at 0.125, 0.25, 0.5, 1, 2, 4, 8, and 10 hours after the dose. All samples were immediately centrifuged at 12,500g for 2 minutes on a tabletop centrifuge. Plasma proteins were precipitated by the addition of 200 µL of plasma to 800 µL of cold methanol (-30 °C), followed by vigorous agitation on a vortex mixer and centrifuging again at 1,250g for 2 minutes. The supernatant was decanted and stored at -70 °C until analysis.

Concentrations of irinotecan and its more potent metabolite SN-38 (7-ethyl-10-hydroxy-camptothecin) in plasma were determined by an isocratic high-pressure liquid chromatography (HPLC) assay with fluorescence detection; this technique has been described in detail previously.[23] The lower level of quantitation was 2.5 ng/mL for both irinotecan and SN-38. All calibrators and controls were prepared in murine plasma.

Pharmacokinetic Analysis--A four-compartment model using maximum likelihood estimation (ADAPT II) was fit simultaneously to the irinotecan and SN-38 plasma concentration data after intravenous administration.[24] Model parameters included volume of the central compartment for irinotecan and SN-38 (Vc), irinotecan elimination rate (k₁₀), irinotecan intercompartment rate constants (k₁₂, k₂₁), SN-38 elimination rate constant (k₃₀), carboxylesterase conversion of irinotecan to SN-38 (k₁₃), and SN-38 intercompartment rate constants (k₃₄, k₄₃). Irinotecan and SN-38 undergo linear disposition at intravenous doses < 20 mg/kg in mice; thus a linear rate constant (k₁₃) was used to represent hydrolysis by carboxylesterase.[25] Systemic clearance of irinotecan was calculated by Vc and the sum of k₁₃ and k₁₀, and SN-38 systemic clearance was calculated from Vc and k₃₀. Area under the irinotecan and SN-38 plasma concentration-time curve from zero to infinity (AUC_0-¥) was calculated using the log-linear trapezoidal method.[26]

Tumor Response and Statistical Analysis--For comparison of different treatment regimens, tumor responses were analyzed for: (1) the time (days) individual tumors required to reach four times their volume at initiation of therapy, and (2) tumor regression. The proportion of tumors that failed to reach four times their volume at the start of treatment was estimated by the Kaplan-Meier method,[27] in which time to reach four times the original volume was censored in mice that died for any reason. The exact log-rank test[28,29] was used to assess differences among treatment groups with respect to time for tumors to increase fourfold in volume, as described previously.[23]

For individual tumors, partial response (PR) was defined as a volume regression > 50% but with measurable tumor at all times. Complete response (CR) was defined as disappearance of measurable tumor mass at some point after therapy was initiated. Maintained complete response (MCR) was CR with-out tumor regrowth within the study period (usually 12 weeks). Procedures for evaluating intracranial tumor responses have been reported previously.[30]

Results

Schedule-Dependent Antitumor Activity

In human xenograft models, different camptothecin analogs demonstrate both qualitative and quantitative differences in antitumor activity. However, certain principles have emerged that may be more generally applied to this class of topoisomerase I poisons.

Our group[19,20] and others[31-33] have clearly shown that the antitumor activity of camptothecin agents is highly schedule-dependent. The duration of treatment appears to be critical for optimal antitumor activity. Thus, as shown in Figure 1, irinotecan administered to mice bearing a human neuroblastoma on the (d ×5)2 schedule is more effective than the same total dose given over 5 days [(d × 5)1]. Similar results have been obtained with 9-aminocamptothecin and topotecan.

Data in Figure 2 further substantiate the schedule-dependent activity of irinotecan. In this experiment, mice bearing a colon tumor were treated on the (d × 5)2 schedule at 40 or 10 mg/kg/dose for one cycle {[(d ×5)2]1}, or on the same schedule at lower dose levels, where cycles were repeated every 21 days {[(d ×5)2]3}. Two points are noteworthy:

On the [(d ×5)2]1 schedule, the antitumor activity of 10 and 40 mg/kg/dose was identical, suggesting self-limiting antitumor activity.

Much better responses were obtained with the [(d ×5)2]3 schedule, despite a lower total dose administered (400 vs 150 mg/kg total dose).

Similar results have been obtained with other camptothecin analogs that, unlike irinotecan, do not require activation. Since the cytotoxicity of camptothecins depends on active DNA replication (at least at pharmacologically achievable levels), the most likely explanation is that increased antitumor activity is a consequence of killing more cells in S-phase by giving the drug over a longer period.

Table 1, Table 2, and Table 3 summarize results for rhabdomyosarcoma, colon carcinomas, and pediatric brain tumor xenografts treated with the optimal schedule {[(d × 5)2]3} of irinotecan. Figure 3 shows the responses to different cytotoxic drugs of a childhood glioblastoma xenograft (SJ-GBM2) established from a previously treated patient. Each agent was administered at the maximum tolerated dose (MTD) to mice bearing SJ-GBM2. The only agents that had significant activity were irinotecan and topotecan, whereas doxorubicin(Drug information on doxorubicin), vin- cristine, and melphalan(Drug information on melphalan) (Alkeran) were inactive.

Additional data from adult and pediatric brain tumors, as subcutaneous grafts, are presented in Table 4. Against intracranial implanted D456 (glioblastoma) or D341 (medulloblastoma) tumors, irinotecan enhanced survival time by 114% and 73%, respectively. The response rates of these tumors as xenografts clearly exceed the response rates of their clinical counterparts, when these have been determined (eg, colon carcinoma). Why, then, are the models overpredicting the clinical activity of irinotecan?