- © 2003 by American Society of Clinical Oncology
Cantuzumab Mertansine, a Maytansinoid Immunoconjugate Directed to the CanAg Antigen: A Phase I, Pharmacokinetic, and Biologic Correlative Study
- Anthony W. Tolcher,
- Leonel Ochoa,
- Lisa A. Hammond,
- Amita Patnaik,
- Tam Edwards,
- Chris Takimoto,
- Lon Smith,
- Johann de Bono,
- Garry Schwartz,
- Theresa Mays,
- Zdenka L. Jonak,
- Randall Johnson,
- Mark DeWitte,
- Helen Martino,
- Charlene Audette,
- Kate Maes,
- Ravi V.J. Chari,
- John M. Lambert and
- Eric K. Rowinsky
- From the Institute for Drug Development, Cancer Therapy and Research Center and The University of Texas Health Science Center at San Antonio; and Brooke Army Medical Center, San Antonio, TX; GlaxoSmithKline, Collegeville, PA; and ImmunoGen Inc., Cambridge, MA.
- Address reprint requests to Anthony W. Tolcher, MD, Institute for Drug Development, Cancer Therapy and Research Center, 7979 Wurzbach, Suite #400, San Antonio, TX 78229; email: atolcher{at}saci.org.
Abstract
Purpose: To determine the maximum tolerated dose and pharmacokinetics of cantuzumab mertansine, an immunoconjugate of the potent maytansine derivative (DM1) and the humanized monoclonal antibody (huC242) directed to CanAg, intravenously (IV) once every 3 weeks and to seek evidence of antitumor activity.
Patients and Methods: Patients with CanAg-expressing solid malignancies were treated with escalating doses of cantuzumab mertansine administered IV every 3 weeks. The pharmacokinetic parameters of cantuzumab mertansine, the presence of plasma-shed CanAg, and the development of both human antihuman and human anti-DM1 conjugate antibodies also were characterized.
Results: Thirty-seven patients received 110 courses of cantuzumab mertansine at doses ranging from 22 to 295 mg/m2. Acute, transient, and reversible elevations of hepatic transaminases were the principal toxic effects. Nausea, vomiting, fatigue, and diarrhea were common but rarely severe at the highest dose levels. Dose, peak concentration, and area under the concentration–time curve correlated with the severity of transaminase elevation. The mean (± SD) clearance and terminal elimination half-life values for cantuzumab mertansine averaged 39.5 (±13.1) mL/h/m2 and 41.1 (±16.1) hours, respectively. Strong expression (3+) of CanAg was documented in 68% of patients. Two patients with chemotherapy-refractory colorectal carcinoma had minor regressions, and four patients had persistently stable disease for more than six courses.
Conclusion: The recommended dose for cantuzumab mertansine is 235 mg/m2 IV every 3 weeks. The absence of severe hematologic toxic effects, preliminary evidence of cantuzumab mertansine tumor localization, and encouraging biologic activity in chemotherapy-refractory patients warrant further broad clinical development of this immunoconjugate in CanAg-expressing tumors.
CANTUZUMAB MERTANSINE (SB-408075, huC242-DM1) is a tumor-activated immunoconjugate derived from the conjugation of approximately four molecules of the potent maytansinoid antimicrotubule agent DM1 (N2′-deacetyl-N2′-(3-mercapto-1-oxopropyl)-maytansine C35H48ClN3O10S) to the humanized monoclonal antibody huC242 (Fig 1⇓). HuC242 binds specifically to the extracellular domain of a tumor-associated carbohydrate epitope of CanAg (a novel glycoform of MUC1), which is strongly expressed in most pancreatic, biliary, and colorectal cancers as well as in a large proportion of non–small-cell lung (40%), gastric (55%), uterine (45%), and bladder (40%) cancers.1–,3 In contrast, only minimal immunostaining of normal tissues has been reported.1,2 Once cantuzumab mertansine is bound to the external domain of CanAg, the complex is internalized, and the DM1 molecules are released intracellularly by cleavage of the DM1-huC242 disulfide linkage.
Schematic of immunoconjugate structure of DM1 (n = 3 to 4).
DM1 is a derivative of the natural microbial fermentation product and extraordinarily potent antimicrotubule agent ansamitocin P-3.4 Maytansine was evaluated by the National Cancer Institute in phase I and II studies in the 1970s.5–,9 Although complete and partial regressions of advanced non-Hodgkin’s lymphoma, melanoma, thymoma, acute lymphocytic leukemia, and ovarian and breast carcinomas were noted in these early clinical trials, severe toxic effects, particularly nausea, vomiting, diarrhea, elevations of hepatic function tests and, less commonly, weakness and lethargy precluded further development of maytansine in favor of other antimicrotubule agents.5–,9
On the basis of the extremely potent cytotoxic properties and clinical activity in limited early studies, maytansine was believed to be an ideal cytotoxic agent for conjugation to tumor-specific monoclonal antibodies. Chemical derivatization of maytansine resulted in the synthesis of a maytansine derivative DM1, which is three- to 10-fold more potent than maytansine, with an inhibitory concentration (IC50) in the picomolar range and broad cytotoxic activity against a broad range of human tumors in vivo.4 The free sulfydryl group of DM1 enabled stable disulfide linkage to huC242 that was modified to contain sulfhyoryl groups. Experimental studies demonstrated that the conjugation of DM1 to huC242 (cantuzumab mertansine) permits the selective distribution of DM1 molecules to specific antigen-bearing cells, resulting in enhancement of antitumor activity and reduction of normal tissue toxicity.4,10,11
In addition to its extraordinary potency in vitro, cantuzumab mertansine demonstrated impressive antitumor activity against a broad range of CanAg-expressing, well-established human tumor xenografts.4,10 Treatment of mice bearing established human xenografts of Colo 205, HT-29, and LoVo colon cancer resulted in both cures and complete responses at relatively nontoxic cantuzumab mertansine doses of 48 mg/m2 intravenously (IV) daily for 5 days.10 In contrast, tumor regressions were not observed in mice treated with maytansine, the “naked” huC242 antibody, a mixture of nonconjugated maytansine and huC242, fluorouracil (FU), or irinotecan as single agents at their respective maximum tolerated doses (MTDs). Furthermore, treatment of mice bearing established human lung cancer cell line H441 or pancreatic cell lines SW-1990, BxPC-3, and SU-8686 with even lower doses of cantuzumab mertansine (24 to 27 mg/m2 daily for 5 days) resulted in complete responses.12
The toxicologic and pharmacologic profiles of cantuzumab mertansine were evaluated in both mice and cynomolgus monkeys—the latter animal model exhibiting comparable normal tissue expression of CanAg as in humans.1 In the single-dose toxicologic studies, the lethal dose (LD) in 10% and 50% of mice (LD10 and LD50) was 228 and 333 mg/m2, respectively.1 Histomorphologically, gastrointestinal, hematopoietic, and neuronal tissues were the most sensitive tissues to the toxic effects of cantuzumab mertansine. Transient and reversible elevations of serum lipase, amylase, ALT, and AST also were observed in monkeys. In addition, pathologic studies revealed modest neuron axonal degeneration in monkeys treated with a single dose at 341 mg/m2, whereas mild axonal degeneration was noted in animals treated with lower doses of 58 to 228 mg/m2 weekly, thereby demonstrating the potential for both peak and cumulative dose-related neurotoxicity. The plasma clearance of cantuzumab mertansine in mice was biphasic, with terminal half-life (t1/2) values averaging approximately 48 hours.1 In monkeys treated with a single IV dose of 228 mg/m2, pharmacokinetics were dose independent, with t1/2 and volume of distribution averaging 40.3 hours and 59 mL/kg, respectively.1
The rationale for the clinical development of cantuzumab mertansine included the potential for targeted delivery of the potent maytansinoid DM1 on the basis of the prevalence of CanAg expression on several common human tumors and the relative lack of expression on normal tissues, as well as the impressive activity of cantuzumab mertansine in a broad spectrum of CanAg-expressing experimental tumors at nontoxic doses. The principal objectives of this phase I and pharmacokinetic study were to (1) determine the MTD of cantuzumab mertansine administered as an IV infusion every 3 weeks, (2) determine the toxic effects of cantuzumab mertansine on this schedule, (3) characterize the pharmacokinetic behavior of cantuzumab mertansine, (4) seek preliminary evidence of anticancer activity in patients with advanced solid malignancies, and (5) characterize the expression of CanAg on tumor cells and shed CanAg expression and explore the relationships between these parameters and both the toxicity and pharmacokinetics.
PATIENTS AND METHODS
Patient Selection
Patients with histologically or cytologically confirmed solid malignancies refractory to standard therapy or for whom no standard therapy existed were eligible. Immunohistochemical assessment of CanAg expression was performed in all patients with available tissue. As CanAg is highly expressed in the majority of pancreatic and colorectal carcinomas, patients with these malignancies could be treated without prior confirmation of CanAg expression; however, patients with other malignancies required immunohistochemical confirmation of CanAg expression before enrollment. Eligibility also included an age ≥ 18 years; a life expectancy of at least 12 weeks; an Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 2; no prior chemotherapy within 4 weeks (6 weeks for mitomycin or a nitrosourea); adequate hematopoietic (hemoglobin ≥ 9 g/dL, absolute neutrophil count [ANC] ≥ 1,500/μL, platelet count ≥ 100,000/μL), hepatic (bilirubin ≤ 1.5 mg/dL, AST, ALT, and alkaline phosphatase less than or equal to three times the upper limit of normal, or less than or equal to five times institutional upper limit of normal if the elevation was from hepatic metastases), and renal (serum creatinine ≤ 1.5 mg/dL) functions; measurable or evaluable disease; no evidence of brain metastases; and no coexisting medical problem of sufficient severity to limit compliance with the study. Patients gave written informed consent before treatment for all clinical and research aspects of the study according to federal and institutional guidelines.
Dosage and Drug Administration
IV cantuzumab mertansine was administered at a rate of 1 mg/min for 30 minutes and then increased to 3 mg/min if hypersensitivity phenomena were not observed. Treatment courses were repeated every 3 weeks. An accelerated dose-escalation method was used to guide dose escalation in cohorts of new patients, and toxicity was graded according to the National Cancer Institute’s (NCI) Common Toxicity Criteria, version 2. Three patients were treated with cantuzumab mertansine at the starting dose of 22 mg/m2, which represented one tenth of the LD10 in mice. The dose was doubled in each new cohort of patient(s) until moderate toxicity (grade ≥ 2) was observed, at which time the cohort size was a minimum of three patients and a more conservative dose-escalation scheme was invoked. If one of three patients experienced dose-limiting toxicity (DLT), the cohort was expanded to six patients. The MTD was defined as the highest dose at which fewer than two of six patients experienced treatment-related DLT during the first course of therapy. DLT was defined as any grade 3 or 4 nonhematologic toxicity (including grade ≥ 3 nausea or vomiting despite optimal antiemetics), grade 4 thrombocytopenia, or grade 4 neutropenia (ANC < 500/μL) lasting more than 4 days or accompanied by fever. In patients with hepatic transaminase elevations before treatment because of liver metastases, DLT was defined as an elevation of at least two grade levels above pretreatment.
Cantuzumab mertansine was supplied in 20-mL single-use vials by ImmunoGen, Inc. (Cambridge, MA). Each vial contained protein at a concentration of 0.7 mg/mL in a buffered solution (acceptable pH, 6.5 ± 0.5) consisting of monobasic potassium phosphate (0.57 mg/mL), monobasic sodium phosphate monohydrate (0.20 mg/mL), dibasic sodium phosphate (0.555 mg/mL), and sodium chloride (8.16 mg/mL) in purified water.
Pretreatment and Follow-Up Studies
A complete medical history, physical examination, concurrent medication profile, assessment of performance status, and routine laboratory studies were performed pretreatment and weekly during treatment. Routine laboratory studies included a complete blood count (CBC), differential WBC count, prothrombin and partial thromboplastin times, electrolytes, blood urea nitrogen, serum creatinine, uric acid, glucose, alkaline phosphatase, lactate dehydrogenase, ALT, AST, total bilirubin, calcium, total protein, albumin, cholesterol, triglycerides, amylase and lipase, and urinalysis. Pretreatment studies also included an ECG, relevant radiologic studies for the evaluation of all measurable and evaluable sites of disease, and an assessment of appropriate tumor markers. Radiologic studies for disease status were repeated after every other course. Patients were able to continue treatment if they did not develop progressive disease or experience intolerable toxicity. A complete response was scored if there was disappearance of all measurable and evaluable disease for at least two measurements performed at least 4 weeks apart without worsening of disease-related symptomatology or declining performance status. A partial response required at least a 50% reduction in the sum of the product of the bidimensional measurements of all lesions documented by at least two measurements separated by at least 4 weeks. Any concurrent increase in the size of any lesion by 25% or more or the appearance of any new lesion was considered disease progression. In course 1, patients were observed for 3 hours following treatment. ECGs were performed 1, 3, 6, and 24 hours posttreatment. A complete physical and neurologic exam was performed 24 hours following cantuzumab mertansine dosing along with a CBC and chemistry profile including lipase and amylase. Laboratory assessments also were performed on days 5, 8, 15, and 22 of course 1 and weekly for subsequent courses.
Plasma Pharmacokinetic Sampling and Assay
Blood samples were collected into heparinized tubes before the first infusion and at the end of infusion. Samples also were collected at 15 minutes and 3, 8, 24, 48, and 96 hours after the end of the infusion and on days 8, 15, and 22. Such intensive sampling was performed on the first and fifth courses, and peak and trough blood samples were obtained immediately before every course. The blood samples were centrifuged at 3,000 rpm for 10 minutes immediately after collection, and the plasma was transferred to separate tubes and frozen to −20°C until assayed.
Determination of Intact Cantuzumab Mertansine (huC242-DM1 Conjugate) in Plasma
The intact cantuzumab mertansine immunoconjugate was measured using a dual antibody enzyme-linked immunosorbent assay (ELISA) method using two specific murine IgG1 monoclonal antibodies developed at ImmunoGen, Inc. One antibody, antihuC242, is specific for huC242 (an anti-idiotype), and the second antibody is specific for the DM1 moiety.
Briefly, ELISA plates were coated with antihuC242 antibody (KM142) to capture cantuzumab mertansine from the test samples. Standard curves using cantuzumab mertansine concentrations ranging from 1.56 to 200 ng/mL diluted in 2% normal human plasma were run on each plate. The amount of cantuzumab mertansine that was captured in each well was detected using biotinylated antimaytansinoid antibody (ND137) followed by streptavidin-horseradish peroxidase (HRP) and 2,2′-azino-bis (3-ethylbenzethiazoline-6-sulfonic acid) (ABTS) to develop the signal absorbance, which was recorded at 405 nm. For each sample dilution, the mean of the triplicates was calculated, and the concentration of the intact conjugate was determined by comparison with the standard curve. The lower limit of detection (LLD) for this assay was 31.25 ng/mL in plasma.
Determination of Total huC242 Antibody (huC242-DM1 and Free huC242) in Plasma
The total huC242 antibody, including both unconjugated huC242 and huC242-DM1 containing varying amounts of DM1, was measured in patient plasma samples using an ELISA method, which measured the competition of binding of biotinylated huC242 antibody to anti-huC242 antibody by the standard huC242 antibody and the test samples. Briefly, anti-huC242 antibody was captured using goat antimouse IgG(FC)-coated ELISA plates. Test samples and standard curve solutions were mixed with an equal volume of 20 ng/mL biotinylated-huC242 antibody, and then the mixtures were added to the coated ELISA plates. The standard curves were made using concentrations of huC242 antibody ranging from 3.9 to 500 ng/mL. The plates were developed using streptavidin-HRP and ABTS, and the absorbance was recorded at 405 nm. For each sample dilution, the mean of triplicates was calculated and the concentration of total huC242 antibody was determined by comparison with the standard curve. The LLD for this assay was approximately 1,560 ng/mL in plasma.
Detection of Human Anti-huC242 Antibodies and Anti-DM1 Antibodies
Human anti-huC242 antibodies (HAHA) and anti-DM1 antibodies (HADA) were detected in the plasma of patients with ELISA methods that use the multivalency of the Ig molecules. For detecting HAHA, ELISA plates were coated with huC242 antibody to capture any anti-huC242-specific human antibodies from the plasma. Bound human anti-huC242 antibody was then detected by its capture of biotinylated huC242, followed with streptavidin-HRP and ABTS reagent to develop the signal, and absorbance was recorded at 405 nm. Values were compared with a standard curve (4,000 ng/mL to 7.81 ng/mL in 2% human serum) of anti-huC242 (a murine IgG1 monoclonal antibody), which also served as a positive control. All wells were run in triplicate. The LLD for this assay was 781 ng/mL.
For detection of HADA, ELISA plates were coated with bovine serum albumin conjugated with DM1 (BSA-DM1) to capture any specific human anti-DM1 antiserum. Biotinylated antigen (generated by the reaction of the thiol of DM1 with a maleimide-biotin derivative) was then added to detect any bound human anti-DM1 antibody by capture of the biotinylated DM1, followed by streptavidin-HRP and ABTS to develop the signal, and absorbance was recorded at 405 nm. Values were compared with a standard curve (4,000 ng/mL to 1.95 ng/mL in 2% human serum) of antimaytansinoid antibody (ND137; a murine IgG1 monoclonal antibody), which served as a positive control. All wells were run in triplicate. The LLD for this assay was 390 ng/mL.
Determination of Nonprotein-Bound Maytansinoid in Plasma
Non-protein-bound maytansinoid (“free” maytansinoid) was detected using a competition ELISA method to detect maytansinoids in a protein-free extract of patient plasma samples. Protein-free extracts were made by precipitation of plasma proteins (including, therefore, cantuzumab mertansine) by the addition of an equal volume of chilled (−20°C) acetone to ice-cold aqueous samples. After 30 minutes’ incubation on ice, precipitated protein was removed by centrifugation. Fifty microliters of appropriately diluted extracts, control samples, and a standard curve (50 pM to 400 pM maytansine) were mixed with 50 μL of biotinylated antimaytansinoid antibody, and then 50 μL of this mixture was transferred to ELISA plates coated with BSA-DM1. Streptavidin-HRP and the 3,3′,5,5′-tetramethylbenzidine-substrate system were used for development of the assay, reading the absorbance at 630 nm. Samples were assayed in duplicate. The LLD was approximately 1.2 nmol/L in plasma.
Detection of Shed Antigen (Shed CanAg)
Shed CanAg was detected in plasma using an ELISA that took advantage of the multivalency of the antigen. Briefly, Immulon-2 plates were coated with the murine C242 antibody, which was used to capture CanAg from test samples. A standard curve was made from CanAg prepared as described previously3 and calibrated against a commercial antigen standard (CanAg Diagnostics, Gothenburg, Sweden). Bound antigen was detected by its capture of biotinylated murine C242 antibody, and the assay was developed with streptavidin-HRP and TMB reagent, recording the absorbance at 630 nm. The LLD was 10 units/mL.
Pharmacokinetic and Pharmacodynamic Analyses
Individual plasma concentration data sets for cantuzumab mertansine were analyzed by noncompartmental methods (WinNonLin Professional 3.1, Pharsight, Inc., Mountain View, CA). Peak concentrations (Cmax) were determined by inspection of plasma concentration-time data. Elimination rate constants were estimated by linear regression of the last three data points on the terminal log-linear portion of the concentration-time curves, and t1/2 was calculated by dividing 0.693 by the elimination rate constants. The area under the concentration-time curve (AUC) was calculated using the linear trapezoidal rule up to the last measurable data point (for AUC0–t), then extrapolated to infinity (AUC0–∞). The systemic clearance (CL) was determined by dividing the dose (in mg/m2) by the AUC∞. The apparent volume of distribution at steady state (Vdss) was determined using the following formula: Vdss = (dose × AUMC/AUC2) − (dose × duration of infusion)/(2 × AUC), where AUMC is the area under the moment curve extrapolated to infinity. Dose proportionality was assessed using a one-way analysis of variance (ANOVA) using Scheffe’s multiple comparison test of the effect of cantuzumab mertansine dose on clearance.
The relationships between cantuzumab mertansine pharmacokinetic parameters reflecting drug exposure (eg, AUC and Cmax) and indices reflecting hepatic toxicity (AST, ALT, bilirubin, and alkaline phosphatase) and myelosuppression (ANC and platelets) in the first course were explored. Relevant parameters of myelosuppression that were evaluated included ANC and platelet nadir values, and the percentage decrements in the ANC and platelet counts were calculated as follows: 100 × ([pretreatment counts − nadir counts]/pretreatment counts). For hepatic toxicity, both categorical toxicity grade and percentage elevation in hepatic function test (100 × [peak value − pretreatment value]/pretreatment value) were evaluated. Pharmacodynamic relationships were fit to both linear and sigmoidal maximal effect (Emax) models of drug action (ie, percentage change in toxicity parameter = Emax × AUCγ0/AUC50γ + AUCγ), where Emax was fixed at 100% and AUC50γ is the AUC at which the effect is 50% of Emax. The exponent γ is a constant that describes the sigmoidicity of the curve. The sigmoidal Emax model was fit to the data by nonlinear least squares regression. The coefficient of determination (R2) and the SEs for the estimated parameters were used as measures of goodness of fit for the pharmacodynamics model. Parameter values were expressed as means and SD values. Differences in mean AUC0–∞ values between patients who did or did not experience severe hematologic toxicity were compared using the Student’s t test (two-sided). A multivariate regression analysis was used to examine the association between pharmacokinetic parameters and parameters of hepatic toxicity. Separate regression models were estimated for either percentage elevation in AST or percentage elevation in ALT, with independent variables consisting of dose (mg/m2), clearance, and volume. Statistical significance was declared if the P value on the corresponding coefficient was less than 0.05.
Immunohistochemical Staining of Tumor Tissues
One patient with readily accessible tumor tissue for biopsy consented to a biopsy procedure 24 hours after treatment. The tissue was coated with Tissue-Tek OCT embedding compound (Sakura Finetech USA, Torrance, CA), snap frozen, and sectioned on a cryostat, and then 6-μm sections were air dried and fixed in acetone for 2 minutes at room temperature. Tissues were blocked with Peroxidase Blocking Reagent and Biotin Blocking System (Dako, Carpenteria, CA) to reduce nonspecific endogenous antigen sites. The antibodies, murine C242, antihuman IgG, and anti-DM1, were biotinylated with FluoReporter Mini-Biotin-Protein Labeling Kit (Molecular Probes, Eugene, OR). Tissues were incubated with the antibodies for 30 minutes, and peroxidase-conjugated streptavidin (LSAB2; Dako, CA) was used to detect the presence of antibody binding on the tissue sections. The sections were developed in Liquid DAB (3,3′-diaminobenzidine) Substrate-Chromogen System (Dako, CA) and counterstained with hematoxylin (Sigma Co., St. Louis, MO).
The distribution of the epitope of CanAg recognized by the murine C242 antibody also was determined on sections cut from formalin-fixed, paraffin-embedded tumor biopsies from all patients where available and examined by immunohistochemical staining using the avidin-biotin immunoperoxidase technique. A control murine monoclonal IgG1 antibody was obtained from Coulter Immunology (Hialeah, FL).
RESULTS
General
Thirty-seven patients, whose pertinent demographic characteristics are displayed in Table 1⇓, received a total of 110 courses of cantuzumab mertansine at doses ranging from 22 to 295 mg/m2. The total number of new patients treated and the number of courses at each dose level, as well as the overall dose-escalation scheme, are depicted in Table 2⇓. The median number of courses administered per patient was two (range, one to 10). Doses were increased in two patients, whereas the dose was reduced because of DLT in three patients.
Patient Characteristics
Dose-Escalation Scheme
After no or negligible drug-related effects were noted in the first three patients treated at the 22-mg/m2 dose level, the dose of cantuzumab mertansine was doubled twice, to 44 and 88 mg/m2, respectively. The first patient treated at the 88-mg/m2 dose level experienced grade 2 fatigue, precluding further 100% dose-escalation increments and single-patient cohorts. Thereafter, no DLT was observed in three patients treated with cantuzumab mertansine at the 132- and 176-mg/m2 dose levels. At the next highest dose level, 235 mg/m2, one of the three initial patients experienced dose-limiting elevations in hepatic transaminases, resulting in three additional patients entered at this dose level, none of whom had DLT. Two of the three patients treated at 295 mg/m2 experienced DLT, including a grade 3 AST and grade 3 fatigue in course 1. At this juncture, 10 additional patients were subsequently treated with the 235 mg/m2 dose. At the 235-mg/m2 dose level, one of these additional patients had DLT (grade 3 elevation in hepatic transaminase) during course 1, for a total of two of 16 patients with DLT. Because there was a preponderance of clinically significant episodes of hepatic transaminitis in patients who had extensive hepatic metastases, five additional patients with significant hepatic metastases were treated with cantuzumab mertansine at 176 mg/m2, which resulted in notable but asymptomatic grade 3 elevation of lipase in one of these individuals but no other DLT during the first course. On the basis of these results, the recommended dose of cantuzumab mertansine for phase II studies is 235 mg/m2 IV every 3 weeks.
Toxicity
Hepatic toxicity.
The principal toxic effects of cantuzumab mertansine treatment were reversible elevations of hepatic transaminases and occasionally alkaline phosphatase and bilirubin. The onset of transaminitis was typically within 48 hours following treatment, with median peak values observed on day 5 (range, days 2 to 8) and recovery toward pretreatment values by day 8. Elevations in alkaline phosphatase values occurred somewhat later, with median peak values noted by day 8 and recovery toward pretreatment values by day 15. Most drug-related hepatic toxic effects completely resolved by the time of scheduled retreatment, and no treatment delays were required for persistent elevations, except in those cases in which elevations were clearly attributable to progressive disease. In addition, although elevations in hepatic parameters recurred in subsequent courses in patients who experienced hepatotoxicity in course 1, the magnitude of hepatotoxicity did not worsen significantly with cumulative treatment.
The distributions of grades of related hepatic toxic effects as a function of dose are shown in Table 3⇓. Clinically significant hepatic function test elevations attributable to cantuzumab mertansine were not observed at doses below 176 mg/m2. Of eight patients treated at the 176-mg/m2 dose level, one patient, who had grade 2 elevation in serum alkaline phosphatase values before treatment because of underlying malignancy, experienced further elevations to grade 3 concomitant with progressive growth in hepatic metastases that, therefore, was not considered a DLT. At the next higher dose level, 235 mg/m2, 13 of 16 patients had elevations of liver function tests from pretreatment values by one or two grades, and two of 16 patients ultimately experienced DLT.
Hepatic Toxicity Related to Cantuzumab Mertansine
Severe dose-limiting toxicity occurred in two of three patients treated at 295 mg/m2. One patient, who experienced a three-grade increase in AST (grade 0 to 3), associated with grade 2 nausea and vomiting in course 1, received no further treatment; the second patient who had a two-grade increase in both AST (grade 1 to 3) and ALT (grade 0 to 2) had grade 3 fatigue, and the dose for this patient was reduced to 235 mg/m2 for two subsequent courses, with the patient experiencing mild transaminitis (grade 1).
Possible determinants of hepatotoxicity were explored. Neither the presence of any single elevated hepatic function (AST, ALT, alkaline phosphatase, bilirubin) nor the combinations thereof at pretreatment predicted drug-related elevations in hepatic functions (P > .05 for elevations in AST, ALT, alkaline phosphatase, and bilirubin). Although hepatotoxicity appeared to be related to the extent of hepatic metastases as assessed by computed tomography (CT) scanning, the limited numbers of patients with hepatotoxicity at any one dose level and difficulties measuring the total liver involvement on the basis of CT therefore precluded the formulation of rigid quantitative scoring of liver disease for toxicity prediction.
Hematologic toxicity.
The distribution of NCI grades of neutropenia and thrombocytopenia are displayed in Table 4⇓. Effects on neutrophils and platelets were rarely clinically significant (grade 3 or 4) and never dose limiting. In fact, the most severe hematologic event consisted of grade 3 thrombocytopenia (nadir platelet count 40,000/μL) and uncomplicated, brief (< 5 days) grade 4 neutropenia in the first course of a patient treated with cantuzumab mertansine at the 235-mg/m2 dose level. Neither dose reduction nor treatment delay was required because of severe or unresolved hematologic toxicity.
Hematologic Toxic Effects
Neurotoxicity.
Cumulative and dose-related peripheral sensory neuropathy was experienced by 16 patients (43%). Neurosensory symptoms were characterized by painful and nonpainful paraesthesias involving the feet and hands that persisted following discontinuation of treatment. Eight of the 17 patients (47%) treated at the 235-mg/m2 dose level and two of three patients (67%) at the 295-mg/m2 dose level experienced these symptoms. Only one of eight patients treated at this dose level had received prior oxaliplatin therapy. Two patients, both treated at 235 mg/m2, experienced severe (grade 3) peripheral neuropathy after one and four courses, respectively. Interestingly, the patient who experienced an accelerated onset of severe peripheral neuropathy after the first course of cantuzumab mertansine received prior oxaliplatin and was the same patient who experienced the aforementioned grade 4 neutropenia and grade 3 thrombocytopenia during course 1.
Miscellaneous Toxic Effects
Nausea and vomiting, diarrhea, arthralgias, fever, peripheral neuropathy, and fatigue were common. The distributions of the NCI grades of these toxic effects as a function of dose are depicted in Table 5⇓. Twenty-five patients (68%) and 12 patients (32%) experienced nausea and vomiting, respectively, at some time during treatment. Nausea and vomiting were generally mild or moderate (grade 1 or 2); however, one patient at the 235-mg/m2 dose level experienced grade 3 vomiting at a later course of cantuzumab mertansine. Nausea and vomiting were usually managed and/or prevented with prochlorperazine or a serotonin (5HT3) receptor antagonist. Thirteen patients (35%) experienced diarrhea at some time during treatment. The diarrhea was generally mild to moderate; however, one patient experienced grade 3 diarrhea within 24 hours of the sixth course of cantuzumab mertansine; the diarrhea was not believed to be related to the investigational agent.
Nonhematologic Toxicity (excluding hepatic)
Three patients experienced hypersensitivity reactions. One patient experienced mild flushing, diaphoresis, dyspnea, transient bradycardia, and nausea within 15 minutes of the second course of cantuzumab mertansine at the 22-mg/m2 dose level. The symptoms resolved with temporary interruption of the infusion and IV dexamethasone and antihistamine administration. The infusion was restarted within 1 hour, and the patient had no further recurrence of symptoms on this course. Two additional patients treated at 235 mg/m2 experienced a similar, although more modest, reaction during the first course. Again, the patients received antihistamine and corticosteroids and were successfully retreated without further recurrence of symptoms during this or subsequent courses.
Three patients who had no clinical manifestations of pancreatitis experienced isolated grade 3 elevations of serum lipase. These events occurred at the 22-mg/m2 (course 2), 132-mg/m2 (course 3), and 176-mg/m2 (course 1) dose levels.
Other mild to moderate (grade 1 or 2) nonhematologic toxic effects that were possibly related to cantuzumab mertansine included anorexia in 22% of patients, constipation in 3%, and intestinal ileus in 3%. These effects were noted across the entire cantuzumab mertansine dose range, and definite temporal relationships could not be discerned for any of these toxic effects. The underlying malignant processes might have contributed to these events.
Pharmacokinetics and Pharmacodynamics
Thirty-six patients had plasma sampling performed in the first course for pharmacokinetic studies, and 33 patients had complete sampling at all time points. All data sets were analyzed by noncompartmental methods. The mean pharmacokinetic parameters of the intact immunoconjugate cantuzumab mertansine at each dose level are listed in Table 6⇓. The dose proportionality was confirmed by the demonstration of cantuzumab mertansine clearance being constant over the dose range administered (P = .588, one-way ANOVA). The mean (±SD) Vdss for cantuzumab mertansine approximated the intravascular plasma volume, averaging 1,497 ± 584 mL/m2, whereas the mean plasma clearance was 39.5 ± 13.1 mL/h/m2 and the elimination t1/2 was long, at 41.1 ± 16.1 hours. Neither cantuzumab mertansine clearance nor Vdss increased with increasing with BSA (R2 = 0.011, P = .56, and R2 = 0.015, P = .49, respectively), thereby indicating that alternative dosing methods including flat dosing could be used in future phase II studies.
Noncompartmental Pharmacokinetic Parameters of Intact Cantuzumab Mertansine
Five patients had complete pharmacokinetic sampling performed in both courses 1 and 5. A representative patient’s plasma concentration data fit for both course 1 and 5 at 235 mg/m2 is shown in Fig 2⇓. In these individuals, the clearance of cantuzumab mertansine did not change following repetitive treatments; clearance values (±SD) averaged 37.9 (±7.18) and 38.2 (±14.1) for course 1 and 5, respectively (P > .05).
A representative patient’s plasma cantuzumab mertansine concentration versus time curve for courses 1 and 5 at the 235-mg/m2 dose level.
The total concentration of huC242 antibody also was measured in the plasma. The ELISA detects total huC242 irrespective of whether the antibody molecules are unconjugated or conjugated to DM1 molecules. The pharmacokinetic parameters for the total antibody are depicted in Table 7⇓. The modest differences between peak concentrations of the total antibody versus immunoconjugate, illustrated in Fig 3⇓, are attributable to the different ELISAs used. The clearance of total antibody was markedly slower than the immunoconjugate, with a mean clearance value of 11.0 ± 8.7 mL/h/m2 and a prolonged mean t1/2 of 230.4 ± 97.5 hours. As the clearance of the total antibody was markedly slower than that of the immunoconjugate, the concentrations of unconjugated antibody exceeded those of the immunoconjugate at later time points postinfusion. Representative concentration–time profiles for both the immunoconjugate and total antibody, including peak and trough values, in courses 1 through 5 are depicted in Fig 3⇓.
Noncompartmental Pharmacokinetic Parameters of Total Antibody
Representative patient’s plasma concentration versus time curve for courses 1 and 5 for the intact immunoconjugate (open circles) and total antibody (closed circles), and peak and trough values for courses 2, 3, and 4. Symbols in parentheses represent data below the level of detection for assay.
Free maytansinoid (DM1 non-protein bound) plasma concentrations were measured in two patients at the highest dose level of 295 mg/m2 using an ELISA of a protein-free extract of blood. DM1 was detected in the plasma up to 48 hours following treatment but decreased below the lower limit of detection (1.2 nmol/L) at 96 hours posttreatment. The free maytansinoid concentrations in the plasma of the two patients were 52 and 56 nmol/L immediately following cantuzumab mertansine infusion and declined to 4.6 and 5.5 nmol/L by 48 hours. At all time points, the free maytansinoid represented less than 1% of the total circulating maytansinoid (huC242-conjugated immunoconjugate plus free maytansinoid).
The relationships between the pharmacokinetic parameters that reflect intact cantuzumab mertansine exposure (Cmax and AUC) and the principal toxic effects encountered in this study also were explored. Scatterplots of the percentage increase in AST and ALT values as functions of Cmax and AUC0–∞ are depicted in Fig 4A–D⇓; all relationships were roughly linear (R2 = 0.137 to 0.312), with the strongest being those relating transaminitis to the magnitude of Cmax. The mean cantuzumab mertansine AUC0–∞ values at the 235- and 295-mg/m2 dose levels were greater, albeit not significantly so, for patients who experienced severe hepatic toxicity (grade 3 or 4 transaminitis) compared with patients without toxicity of this magnitude (eg, for ALT, 75.74 ± 31.28 mg/mL × hr v 55.80 ± 23.52 mg/mL × hr; P = .12).
Scatterplots depicting the relationships between percentage increase in AST values during the first course of cantuzumab mertansine and Cmax (A) and AUC (B) and the percentage increase in ALT values during the first course and Cmax (C) and AUC (D).
Although a rough relationship of Cmax and AUC0–∞ to elevation in AST and ALT was observed using univariate analysis, a multivariate analysis of the relationship of independent variates of dose, AUC0–∞, Cmax, cantuzumab mertansine clearance, and Vdss indicated that none of the variables was statistically significant (P > .05).
In contrast to transaminitis, elevations in alkaline phosphatase or bilirubin could not be related to either Cmax or AUC values (data not shown). With regard to hematologic effects, no pharmacodynamic relationships were apparent, and the few patients who experienced grade 3 or 4 neutropenia did not have lower clearance rates for cantuzumab mertansine compared with other patients.
Circulating Shed CanAg
There was marked interindividual variability in pretreatment levels of shed CanAg levels in plasma. Thirty-three of 37 patients (89%) had quantifiable plasma levels of shed CanAg before treatment; the median value was 184 units (range, from 10 [lower limit of detection] to 31,240 units; Fig 5⇓). Following the first dose of cantuzumab mertansine, the shed CanAg levels decreased to undetectable levels in 25 of 30 patients (83%) who had complete pre- and posttreatment data sets (Fig 5⇓). An analysis of all paired data sets indicated that this effect was highly significant (P < .0001, two-tailed paired t test). In eight patients who had multiple measurements of shed CanAg performed in the peritreatment period, plasma clearance of shed CanAg was rapid, with reductions to undetectable levels noted by the first time point assayed (3-hour and 8-hour samples in five and three patients, respectively). In 21 patients (70%), shed CanAg levels decreased to undetectable levels that persisted until the start of course 2 (day 22). Shed CanAg decreased to undetectable levels in patients treated across the entire cantuzumab mertansine dose range, indicating that CanAg clearance was not dose-dependent, at least not in the dose range examined in this study.
Scatterplots of individual circulating shed CanAg values pretreatment and following the first course (day 22) of cantuzumab mertansine. The patient with the highest CanAg value (31,240 units) sampled to 8 hours postinfusion only. Dashed line indicates lower limit of detection for assay.
The relationships between the magnitude of pretreatment shed CanAg levels and both the clearance of cantuzumab mertansine and the principal clinical toxic effects encountered during the first course of treatment were examined. No relationships between the pretreatment shed CanAg values and the clearance of cantuzumab mertansine were evident (data not shown). Similarly, neither linear nor nonlinear models were satisfactory in describing the relationships between pretreatment shed CanAg values and the percentage increases in relevant hepatic function parameters (data not shown), although only a few patients (four) had high CanAg values (> 1,500 units; Fig 5⇑).
Formation of Human Antihuman and Human Anti-DM1 Antibodies
Neither HAHA nor HADA was detected in any of the 37 patients entered onto the study, including the patients who experienced hypersensitivity reactions.
Immunohistochemistry for Tumor CanAg Expression
Thirty-five of the 37 patients entered had tumor specimens available for immunohistochemical determination of CanAg expression, and the results are shown in Table 8⇓. A homogeneous distribution pattern of CanAg staining was observed in 15 patients (44%), whereas most patients had either heterogeneous (14 [40%] patients) or focal (five [14%] patients) staining. The 30 colorectal cancer patients exhibited a broad range of immunostaining distribution and intensity scores. Three patients with pancreatic carcinomas demonstrated homogeneous 3+, heterogeneous 2+, and focal 3+ immunostaining, and the single non–small-cell lung cancer patient had heterogeneous 3+ immunostaining. Relationships between CanAg immunostaining and the duration that patients remained on study were explored. Neither the distribution pattern nor the intensity of immunostaining was related to the number of cumulative courses.
Tumor Immunohistochemistry for CanAg
One patient with accessible subcutaneous colon carcinoma metastases consented to a tumor biopsy 24 hours following cantuzumab mertansine infusion. Immunohistochemical staining of the tissue indicated 2+ homogeneous CanAg expression and DM1 localization to the tumor using a murine monoclonal anti-DM1 antibody as depicted in Fig 6⇓. This patient had evidence of a minor response of pulmonary metastases after two and four courses and received a total of six courses.
Immunohistochemical detection for (A) CanAg expression (2+ homogeneous), (B) human IgG, and (C) DM1-huC242 (cantuzumab mertansine) in the tumor biopsy obtained 24 hours after the first infusion of cantuzumab mertansine.
Antitumor Activity
No major objective responses were noted; however, two patients had minor responses. One patient, a 48-year-old female with metastatic colorectal carcinoma previously treated with FU and irinotecan, had a greater than 33% reduction in the size of a retroperitoneal lymph node following treatment with cantuzumab mertansine at 235 mg/m2 and continued on treatment for a total of eight courses. The second patient, a 57-year-old female with metastatic colorectal carcinoma refractory to both FU and irinotecan, had a 36% reduction in the size of the lung metastases following cantuzumab mertansine treatment at the 235-mg/m2 dose level and continued for a total of six courses. Four other patients had persistent stable disease exceeding 6 months. Seven patients experienced decrements of elevated carcinoembryonic antigen (CEA) values of greater than 30%, the most vivid of which was a 76% decrease after one course. Interestingly, five of 14 patients who had elevated CEA values pretreatment and were treated at the two highest dose levels, 235 and 295 mg/m2, experienced decrements of at least 30% or greater. The possibility of cantuzumab mertansine interfering with the CEA radioimmunoassay was ruled out in comixing studies (data not shown).
DISCUSSION
Cantuzumab mertansine represents the first in a series of DM1-linked immunoconjugates currently in preclinical and early clinical development.10,11,13,14 Immunoconjugates represent novel delivery modalities for highly potent cytotoxic agents, particularly those that cannot feasibly be delivered via conventional systemic administration routes. The maytansinoid antimicrotubule agents exemplify such agents. Maytansine, although active at extremely low concentrations, exhibited too narrow a therapeutic index, with significant and sometimes unpredictable gastrointestinal toxic effects that precluded further clinical development. Cantuzumab mertansine was selected for clinical development on the basis of the selective delivery of the potent antimicrotubule agent DM1, the impressive antitumor activity of this agent in CanAg-expressing human tumor cell lines and xenograft models at nontoxic doses, and the prevalence of CanAg expression in common human tumors.4,10
The extrapolation of cantuzumab mertansine toxicology data from mice and monkeys to humans closely approximated the level at which clinically relevant toxicity was observed in this study. The starting dose for this phase I study, 22 mg/m2, was equivalent to one tenth of the LD10 in mice, which is a conventional algorithm for cytotoxic dose selection in phase I studies. In the current study, cantuzumab mertansine doses were escalated more than 10-fold before the MTD was reached and significant toxicity was noted. Also predicted from the preclinical toxicology studies, the principal toxic effects of cantuzumab mertansine were reversible elevations in hepatic transaminases and other nonhematologic gastrointestinal and neurosensory toxic effects. Although elevations of AST and ALT were common at the 235-mg/m2 dose level, the increments were rarely severe. Furthermore, at dose levels of 176 mg/m2 and below, the occurrence of hepatic toxicity was rare even in patients with significant preexisting hepatic metastases.
The etiology of the reversible transaminase elevations may be attributable to one or a combination of mechanisms. Hepatocytes do not express the CanAg antigen; therefore, a direct toxic injury secondary to cantuzumab mertansine binding to normal hepatocytes is unlikely. DM1, however, may possess intrinsic hepatotoxic properties. In the previous clinical phase I and II studies of maytansine, a pattern of transient, but also at times severe, hepatotoxicity was observed.5–,9 Because cantuzumab mertansine is degraded into individual components, including the release, albeit in extremely low concentrations, of free maytansinoid into the systemic circulation, this may contribute, at least in part, to the transaminitis observed.
An alternative mechanism for the hepatic transaminase elevations may include cantuzumab mertansine localizing to hepatic metastases with a “bystander injury” to adjacent normal hepatocytes. The observation that patients with bulky hepatic metastases were at greatest risk for the development of this toxicity, even if they had normal or near normal transaminases at pretreatment, may support this bystander injury effect. Finally, hepatic transaminitis may represent a toxicity associated with the immunoconjugate “class” of antineoplastics. Reversible and sometimes severe hepatic transaminase elevations have been observed in clinical studies of other immunoconjugates, including gemtuzumab (calicheamicin-anti-CD33 immunoconjugate) and an antiB4-blocked ricin immunoconjugate, both of which target hematopoietic and lymphoid antigens not cross-reactive to hepatobiliary cells.15–,17 Common to most therapeutic IgG immunoglobulins, despite widely different antigen targets, is their ultimate clearance mediated by the reticuloendothelial system. This pathway may result in the delivery of high concentrations of the immunoconjugate and the cytotoxic agent directly to hepatobiliary cells.
In addition to the hepatic toxic effects observed with cantuzumab mertansine, a constellation of nonhematologic manifestations that included nausea, vomiting, fever, and fatigue were observed at the two highest dose levels and precluded treatment beyond the observed hepatic biochemical abnormalities. Nausea and vomiting, including delayed onset, occurred frequently in patients who experienced significant transaminase elevations, but it could be successfully managed, in most patients, with prochlorperazine and/or serotonin 5HT3 receptor antagonists. The perturbations in hepatic function also may have contributed to fatigue reported in 65% of patients treated at the 235 mg/m2 dose level and 100% of patients (including one grade 3 fatigue) treated at the 295-mg/m2 dose level. Fever, although not severe, did occur in 35% and 33% of patients treated at the 235- and 295-mg/m2 dose levels, respectively.
Hypersensitivity reactions were uncommon events with cantuzumab mertansine, and only three patients (8%) experienced hypersensitivity reactions. Furthermore, no patient’s treatment was permanently discontinued because of a hypersensitivity reaction. Premedication with antihistamines and dexamethasone permitted retreatment with cantuzumab mertansine in all cases. These reactions may represent idiosyncratic reactions to the immunoconjugate because they occurred during the first course without prior sensitization to the agent. Moreover, the absence of detectable HAHA and HADA in all 37 patients, including the three patients with hypersensitivity reactions, and evidence of comparable clearance values of the immunoconjugate despite multiple courses of treatment, further supports the absence of clinically significant immunogenicity with cantuzumab mertansine. This is particularly noteworthy because huC242 is the first clinical evaluation of an antibody that was humanized via the variable domain resurfacing method.18
The degradation of cantuzumab mertansine occurs via cleavage of the disulfide linkage intracellularly and in the plasma, resulting in the progressive depletion of DM1 molecules eventually yielding the individual components of unconjugated antibody and “free” DM1. The free maytansinoid concentration, where measurable, was less than 1% of the total circulating maytansinoid and was readily cleared. There was a good correlation between the mean elimination half-life observed in the current study with that derived from preclinical animal studies. Although the plasma elimination half-life of 41.1 hours is longer than that of many conventional cytotoxic agents, it is considerably shorter than that reported for “naked” antibodies such as trastuzumab, rituximab, and the huC242 antibody product of cantuzumab mertansine.19,20 Because of the marked differences in clearances between the conjugated and unconjugated antibodies, the plasma concentration of unconjugated antibody eventually exceeds the immunoconjugate. Although there is a theoretical concern that cantuzumab mertansine’s antitumor effect may be abrogated from the competition between the conjugated and unconjugated antibody for antigen sites, this does not occur in preclinical models examining this issue.21 Furthermore, abrogation would only occur when CanAg antigen-binding sites were near or totally saturated—an implausible situation, based on the density of CanAg expression on tumor cells.
The elevations of hepatic transaminases were a function of cantuzumab mertansine dose, peak plasma concentration (Cmax), and to a lesser extent, AUC0–∞. One strategy to reduce the risk of peak dose–and peak plasma concentration–mediated hepatotoxicity while concurrently maximizing the total dose of cantuzumab mertansine delivered to CanAg-expressing tumor cells would include the evaluation of fractionated dosing schedules. With the elimination half-life of this agent (41.1 hours), frequent dosing even up to every other day would be predicted to be feasible. At present, phase I studies are examining cantuzumab mertansine administration either weekly or three times a week for 3 weeks every 28 days and will address whether this strategy is valid.
Because of the prevalence of CanAg expression in colorectal carcinomas, patients with this tumor type represented the vast majority of patients entered onto this study. Historically, colorectal carcinomas have demonstrated intrinsic chemotherapy resistance to the anticancer effects of antimicrotubule agents in general, and to maytansine in particular.9,22 Although no objective responses were observed with cantuzumab mertansine in this phase I study, two minor responses and persistent stable disease beyond 6 months in several patients was encouraging. Defining the optimal population to select for maximal therapeutic effect, however, remains a critical facet for the future clinical development of cantuzumab mertansine. Evidence from the current study demonstrates that a broad spectrum of CanAg expression exists between patients with similar histologic tumors, and also within an individual’s tumor. Several lines of evidence indicate that the activity with this agent may be dependent on the distribution and intensity of CanAg expression within tumors. Mechanistically, the number of DM1 molecules delivered to an individual cell is dependent on the intensity of CanAg expression, and, plausibly, the optimal population for cantuzumab mertansine treatment will be those patients who express high numbers of CanAg molecules on individual cells throughout their tumor (eg, homogeneous v heterogeneous or focal). The latter is supported by preclinical studies that indicate that mice bearing tumors with high CanAg expression were the most sensitive to cantuzumab mertansine treatment.4,10 Furthermore, clinical experience with trastuzumab indicates that clinical antitumor activity is dependent on target antigen expression within a tumor.23 In the present study, only a minority of patients’ tumors (32%) exhibited maximal CanAg expression (homogeneous 3+), which may have contributed to the absence of antitumor activity observed. Furthermore, on the basis of the aforementioned intrinsic resistance of colorectal carcinomas, studies should be initiated in CanAg-expressing tumors that also exhibit intrinsic sensitivity to antimicrotubule agents, such as non–small-cell lung, bladder, and uterine cancers. Identifying these favorable populations and tumor classes for cantuzumab mertansine treatment therefore remains an important next step for phase II clinical studies.
In conclusion, the immunoconjugate cantuzumab mertansine can be feasibly and safely administered to patients with advanced malignancies at doses that exceed those that portend antitumor activity in preclinical models. The preliminary biologic evidence of tumor localization, minor antitumor activity, and prolonged disease stability in patients with chemotherapy-refractory disease warrants broad clinical development of cantuzumab mertansine in CanAg-expressing malignancies.
- Received May 20, 2002.
- Accepted September 23, 2002.
