Humanized Anti-Trop-2 IgG-SN-38 Conjugate for Effective Treatment of Diverse Epithelial Cancers: Preclinical Studies in Human Cancer Xenograft Models and Monkeys

This article is featured in Highlights of This Issue, p. 3051

Human trophoblast cell-surface antigen (Trop-2), also known as GA733-1 (gastric antigen 733-1), EGP-1 (epithelial glycoprotein-1), and TACSTD2 (tumor-associated calcium signal transducer), is expressed in a variety of human carcinomas and has prognostic significance in some, being associated with more aggressive disease (1–14). Studies of the functional role of Trop-2 in a mouse pancreatic cancer cell line transfected with murine Trop-2 revealed increased proliferation in low serum conditions, migration, and anchorage-independent growth in vitro, and enhanced growth rate with evidence of increased Ki-67 expression in vivo and a higher likelihood to metastasize (15).

Trop-2 antigen's distribution in many epithelial cancers makes it an attractive therapeutic target, but its normal tissue expression profile casts some doubt on Trop-2 as a suitable target (11, 14). Stein and colleagues (4) characterized an antibody, designated RS7-3G11 (RS7), that bound to EGP-1 (16, 17), which was present in a number of solid tumors, but the antigen was also expressed in some normal tissues, usually in a lower intensity, or in restricted regions. Targeting and therapeutic efficacies were documented in a number of human tumor xenografts using radiolabeled RS7 (18–20), but this internalizing antibody did not show therapeutic activity in unconjugated form (18). However, in vitro it has demonstrated antibody-dependent cellular cytotoxicity (ADCC) activity against Trop-2 positive carcinomas (21).

We reported the preparation of antibody–drug conjugates (ADC) using an anti-CEACAM5 (CD66e) IgG coupled to several derivatives of SN-38, a topoisomerase-I inhibitor that is the active component of irinotecan, or CPT-11 (22, 23). The derivatives varied in their in vitro serum stability properties, and in vivo studies found 1 form (designated CL2) to be more effective in preventing or arresting the growth of human colonic and pancreatic cancer xenografts than other linkages with more or less stability. Importantly, these effects occurred at nontoxic doses, with initial testing failing to determine a dose-limiting toxicity (23). These results were encouraging, but also surprising, because the CEACAM5 antibody does not internalize, a property thought to be critical to the success of an ADC. We speculated that the therapeutic activity of the anti-CEACAM5-SN-38 conjugate might be related to the slow release of SN-38 within the tumor after the antibody localized. Because irinotecan performs best when cells are exposed during the S-phase of their growth cycle, a sustained release is expected to improve responses (24–26). Indeed, SN-38 coupled to nontargeting, plasma extending agents, such as polyethylene glycol (PEG) or micelles, has shown improved efficacy over irinotecan or SN-38 alone (27–33), lending additional support to this mechanism.

Given the RS7 antibody's broad reactivity with epithelial cancers and its internalization ability, we hypothesized that an RS7-SN-38 conjugate could benefit not only from the sustained release of the drug, but also from direct intracellular delivery. Therefore, we prepared and tested the efficacy of SN-38 conjugates using a humanized version of the murine RS7 antibody (hRS7). A slight modification was made to the SN-38 derivative that was reported previously (22, 23), which improved the quality of the conjugate without altering its in vitro stability or its efficacy in vivo, and thus this new derivative (designated CL2A) is currently the preferred agent for SN-38 coupling to antibodies. Herein, we show the efficacy of the hRS7-SN-38 conjugate in several epithelial cancer cell lines implanted in nude mice at nontoxic dosages, with other studies revealing that substantially higher doses could be tolerated. More importantly, toxicity studies in monkeys that also express Trop-2 in similar tissues as humans showed that hRS7-SN-38 was tolerated at appreciably higher amounts than the therapeutically effective dose in mice, providing evidence that this conjugate is a promising agent for treating patients with a wide range of epithelial cancers.

All human cancer cell lines used in this study were purchased from the American Type Culture Collection. These include Calu-3 (non–small cell lung carcinoma), SK-MES-1 (squamous cell lung carcinoma), COLO 205 (colonic adenocarcinoma), Capan-1 and BxPC-3 (pancreatic adenocarcinomas), and PC-3 (prostatic adenocarcinomas). Humanized RS7 IgG and control humanized anti-CD20 (hA20 IgG, veltuzumab) and anti-CD22 (hLL2 IgG, epratuzumab) antibodies were prepared at Immunomedics, Inc. Irinotecan (20 mg/mL) was obtained from Hospira, Inc.

Synthesis of CL2-SN-38 has been described previously (22). Its conjugation to hRS7 IgG and serum stability were performed as described (22, 23). Preparations of CL2A-SN-38 (M.W. 1480) and its hRS7 conjugate, and stability, binding, and cytotoxicity studies, were conducted as described previously (22), and are presented in the Supplemental Data. Cell lysates were prepared and immunoblotting for p21^Waf1/Cip, p53, and PARP (poly-ADP-ribose polymerase) was done as described in Supplemental Data. Concentrations, timing, and primary antibodies are shown in the figure legends.

For all animal studies, the doses of SN-38 immunoconjugates and irinotecan are shown in SN-38 equivalents. Based on a mean SN-38/IgG substitution ratio of 6, a dose of 500 μg ADC to a 20-g mouse (25 mg/kg) contains 0.4 mg/kg of SN-38. Irinotecan doses are likewise shown as SN-38 equivalents (i.e., 40 mg irinotecan/kg is equivalent to 24 mg/kg of SN-38).

NCr female athymic nude (nu/nu) mice, 4 to 8 weeks old, and male Swiss-Webster mice, 10 weeks old, were purchased from Taconic Farms. All animal studies were approved by the Center for Molecular Medicine and Immunology's Institutional Animal Care and Use Committee (IACUC). Tolerability studies were performed in Cynomolgus monkeys (Macaca fascicularis; 2.5–4 kg male and female) by SNBL USA, Ltd. after approval by SNBL USA's IACUC.

Animals were implanted subcutaneously with different human cancer cell lines as described in the Supplementary data. Tumor volume (TV) was determined by measurements in 2 dimensions using calipers, with volumes defined as: L × w²/2, where L is the longest dimension of the tumor and w is the shortest. Tumors ranged in size between 0.10 and 0.47 cm³ when therapy began. Treatment regimens, dosages, and number of animals in each experiment are described in the Results. The lyophilized hRS7-CL2A-SN-38 and control ADC were reconstituted and diluted as required in sterile saline. All reagents were administered intraperitoneally (0.1 mL), except irinotecan, which was administered intravenously. The dosing regimen was influenced by our prior investigations, where the ADC was given every 4 days or twice weekly for varying lengths of time (22, 23). This dosing frequency reflected a consideration of the conjugate's serum half-life in vitro, to allow a more continuous exposure to the ADC.

Growth curves are shown as percent change in initial TV over time. Statistical analysis of tumor growth was based on area under the curve (AUC). Profiles of individual tumor growth were obtained through linear-curve modeling. An f-test was employed to determine equality of variance between groups before statistical analysis of growth curves. A 2-tailed t-test was used to assess statistical significance between the various treatment groups and controls, except for the saline control, where a 1-tailed t-test was used (significance at P ≤ 0.05). Statistical comparisons of AUC were performed only up to the time that the first animal within a group was euthanized due to progression.

¹¹¹In-radiolabeled hRS7-CL2A-SN-38 and hRS7 IgG were injected into nude mice bearing s.c. SK-MES-1 tumors (∼0.3 cm³). One group was injected intravenously with 20 μCi (250-μg protein) of ¹¹¹In-hRS7-CL2A-SN-38, whereas another group received 20 μCi (250-μg protein) of ¹¹¹In-hRS7 IgG. At various timepoints mice (5 per timepoint) were anesthetized, bled via intracardiac puncture, and then euthanized. Tumors and various tissues were removed, weighed, and counted by γ scintillation to determine the percentage injected dose per gram tissue (% ID/g). A third group was injected with 250 μg of unlabeled hRS7-CL2A-SN-38 3 days before the administration of ¹¹¹In-hRS7-CL2A-SN-38 and likewise necropsied. A 2-tailed t-test was used to compare hRS7-CL2A-SN-38 and hRS7 IgG uptake after determining equality of variance using the f-test. Pharmacokinetic analysis on blood clearance was performed using WinNonLin software (Parsight Corp.).

Details of these tolerability studies are given in the Supplementary data. Briefly, mice were sorted into 4 groups each to receive 2-mL i.p. injections of either a sodium acetate buffer control or 3 different doses of hRS7-CL2A-SN-38 (4, 8, or 12 mg/kg of SN-38) on days 0 and 3 followed by blood and serum collection, as described in Results. Cynomolgus monkeys (3 male and 3 female; 2.5–4.0 kg) were administered 2 different doses of hRS7-CL2A-SN-38. Dosages, times, and number of monkeys bled for evaluation of possible hematologic toxicities and serum chemistries are described in the Results.

Two different linkages were used to conjugate SN-38 to hRS7 IgG. The first is termed CL2-SN-38 and has been described previously (22, 23). A minor change was made to the synthesis of our CL2 linker in that the phenylalanine moiety was removed (Fig. 1A). This change simplified the synthesis, but did not affect the conjugation outcome (e.g., both CL2-SN-38 and CL2A-SN-38 incorporated ∼6 SN-38 per IgG molecule). Side-by-side comparisons found no significant differences in serum stability, antigen binding, or in vitro cytotoxicity.

To confirm that the change in the SN-38 linker from CL2 to CL2A did not impact in vivo potency, hRS7-CL2A and hRS7-CL2-SN-38 were compared in mice bearing COLO 205 or Capan-1 tumors (Fig. 1B and C, respectively), using 0.4 mg or 0.2 mg/kg SN-38 twice weekly × 4 weeks, respectively, and with starting tumors of 0.25 cm³ size in both studies. Both the hRS7-CL2A and CL2-SN-38 conjugates significantly inhibited tumor growth compared to untreated (AUC_14daysP < 0.002 vs. saline in COLO 205 model; AUC_21daysP < 0.001 vs. saline in Capan-1 model), and a nontargeting anti-CD20 control ADC, hA20-CL2A-SN-38 (AUC_14daysP < 0.003 in COLO-205 model; AUC_35days: P < 0.002 in Capan-1 model). At the end of the study (day 140) in the Capan-1 model, 50% of the mice treated with hRS7-CL2A-SN-38 and 40% of the hRS7-CL2-SN-38 mice were tumor-free, whereas only 20% of the hA20-ADC-treated animals had no visible sign of disease. Importantly, there were no differences in efficacy between the 2 specific conjugates in both the tumor models.

In vitro cytotoxicity studies demonstrated that hRS7-CL2A-SN-38 had IC₅₀ values in the nmol/L range against several different solid tumor lines (Table 1). The IC₅₀ with free SN-38 was lower than the conjugate in all cell lines. Although there was no correlation between Trop-2 expression and sensitivity to hRS7-CL2A-SN-38, the IC₅₀ ratio of the ADC versus free SN-38 was lower in the higher Trop-2-expressing cells, most likely reflecting the enhanced ability to internalize the drug when more antigen is present.

SN-38 is known to activate several signaling pathways in cells, leading to apoptosis (34–37). Our initial studies examined the expression of 2 proteins involved in early signaling events (p21^Waf1/Cip1 and p53) and 1 late apoptotic event [cleavage of poly-ADP-ribose polymerase (PARP)] in vitro (Fig. 2). In BxPC-3 (Fig. 2A), SN-38 led to a 20-fold increase in p21^Waf1/Cip1 expression, whereas hRS7-CL2A-SN-38 resulted in only a 10-fold increase, a finding consistent with the higher activity with free SN-38 in this cell line (Table 1). However, hRS7-CL2A-SN-38 increased p21^Waf1/Cip1 expression in Calu-3 more than 2-fold over free SN-38 (Fig. 2B).

A greater disparity between hRS7-CL2A-SN-38- and free SN-38-mediated signaling events was observed in p53 expression. In both BxPC-3 and Calu-3, upregulation of p53 with free SN-38 was not evident until 48 hours, whereas hRS7-CL2A-SN-38 upregulated p53 within 24 hours. In addition, p53 expression in cells exposed to the ADC was higher in both cell lines compared to SN-38. Interestingly, although hRS7 IgG had no appreciable effect on p21^Waf1/Cip1 expression, it did induce the upregulation of p53 in both BxPC-3 and Calu-3, but only after a 48-hour exposure. In terms of later apoptotic events, cleavage of PARP was evident in both cell lines when incubated with either SN-38 or the conjugate (Fig. 2C). The presence of the cleaved PARP was higher at 24 hours in BxPC-3, which correlates with high expression of p21 and its lower IC₅₀. The higher degree of cleavage with free SN-38 over the ADC was consistent with the cytotoxicity findings.

Because Trop-2 is widely expressed in several human carcinomas, studies were performed in several different human cancer models, which started with an evaluation of the hRS7-CL2-SN-38 linkage, but later, conjugates with the CL2A-linkage were used. Calu-3–bearing nude mice given 0.04 mg SN-38/kg of the hRS7-CL2-SN-38 every 4 days × 4 had a significantly improved response compared to animals administered the equivalent amount of hLL2-CL2-SN-38 (TV = 0.14 ± 0.22 cm³ vs. 0.80 ± 0.91 cm³, respectively; AUC_42daysP < 0.026; Fig. 3A). A dose–response was observed when the dose was increased to 0.4 mg/kg SN-38. At this higher dose level, all mice given the specific hRS7 conjugate were “cured” within 28 days, and remained tumor-free until the end of the study on day 147, whereas tumors regrew in animals treated with the irrelevant ADC (specific vs. irrelevant AUC_98days: P = 0.05). In mice receiving the mixture of hRS7 IgG and SN-38, tumors progressed >4.5-fold by day 56 (TV = 1.10 ± 0.88 cm³; AUC_56daysP < 0.006 vs. hRS7-CL2-SN-38).

Efficacy also was examined in human colonic (COLO 205) and pancreatic (Capan-1) tumor xenografts. In COLO 205 tumor-bearing animals, (Fig. 3B), hRS7-CL2-SN-38 (0.4 mg/kg, q4dx8) prevented tumor growth over the 28-day treatment period with significantly smaller tumors compared to control anti-CD20 ADC (hA20-CL2-SN-38), or hRS7 IgG (TV = 0.16 ± 0.09 cm³, 1.19 ± 0.59 cm³, and 1.77 ± 0.93 cm³, respectively; AUC_28daysP < 0.016). The MTD of irinotecan (24 mg SN-38/kg, q2dx5) was as effective as hRS7-CL2-SN-38, because mouse serum can more efficiently convert irinotecan to SN-38 (38–41) than human serum, but the SN-38 dose in irinotecan (2,400 μg cumulative) was 37.5-fold greater than with the conjugate (64 μg total). Animals bearing Capan-1 showed no significant response to irinotecan alone when given at an SN-38-dose equivalent to the hRS7-CL2-SN-38 conjugate (e.g., on day 35, average tumor size was 0.04 ± 0.05 cm³ in animals given 0.4 mg SN-38/kg hRS7-SN-38 vs. 1.78 ± 0.62 cm³ in irinotecan-treated animals given 0.4 mg/kg SN-38; AUC_day35P < 0.001; Fig. 3C). When the irinotecan dose was increased 10-fold to 4 mg/kg SN-38, the response improved, but still was not as significant as the conjugate at the 0.4 mg/kg SN-38 dose level (TV = 0.17 ± 0.18 cm³ vs. 1.69 ± 0.47 cm³, AUC_day49P < 0.001). An equal dose of nontargeting hA20-CL2-SN-38 also had a significant antitumor effect as compared to irinotecan-treated animals, but the specific hRS7 conjugate was significantly better than the irrelevant ADC (TV = 0.17 ± 0.18 cm³ vs. 0.80 ± 0.68 cm³, AUC_day49P < 0.018).

Studies with the hRS7-CL2A-SN-38 ADC were then extended to 2 other models of human epithelial cancers. In mice bearing BxPC-3 human pancreatic tumors (Fig. 3D), hRS7-CL2A-SN-38 again significantly inhibited tumor growth in comparison to control mice treated with saline or an equivalent amount of nontargeting hA20-CL2A-SN-38 (TV = 0.24 ± 0.11 cm³ vs. 1.17 ± 0.45 cm³ and 1.05 ± 0.73 cm³, respectively; AUC_day21P < 0.001), or irinotecan given at a 10-fold higher SN-38 equivalent dose (TV = 0.27 ± 0.18 cm³ vs. 0.90 ± 0.62 cm³, respectively; AUC_day25P < 0.004). Interestingly, in mice bearing SK-MES-1 human squamous cell lung tumors treated with 0.4 mg/kg of the ADC (Fig. 3E), tumor growth inhibition was superior to saline or unconjugated hRS7 IgG (TV = 0.36 ± 0.25 cm³ vs. 1.02 ± 0.70 cm³ and 1.30 ± 1.08 cm³, respectively; AUC_{28 days}, P < 0.043), but nontargeting hA20-CL2A-SN-38 or the MTD of irinotecan provided the same antitumor effects as the specific hRS7-SN-38 conjugate.

In all murine studies, the hRS7-SN-38 ADC was well tolerated in terms of body weight loss (not shown).

The biodistributions of hRS7-CL2A-SN-38 or unconjugated hRS7 IgG were compared in mice bearing SK-MES-1 human squamous cell lung carcinoma xenografts (Supplementary Table S1), using the respective ¹¹¹In-labeled substrates. A pharmacokinetic analysis was performed to determine the clearance of hRS7-CL2A-SN-38 relative to unconjugated hRS7 (Fig. 4A). The ADC cleared faster than the equivalent amount of unconjugated hRS7, with the ADC exhibiting ∼40% shorter half-life and mean residence time. Nonetheless, this had a minimal impact on tumor uptake (Fig. 4B). Although there were significant differences at the 24- and 48-hour timepoints, by 72 hours (peak uptake) the amounts of both agents in the tumor were similar. Among the normal tissues, hepatic (Fig. 4C) and splenic (Fig. 4D) differences were the most striking. At 24 hours postinjection, there was >2-fold more hRS7-CL2A-SN-38 in the liver than hRS7 IgG. Conversely, in the spleen there was 3-fold more parental hRS7 IgG present at peak uptake (48-hour timepoint) than hRS7-CL2A-SN-38. Uptake and clearance in the rest of the tissues generally reflected differences in the blood concentration.

Because twice-weekly doses were given for therapy, tumor uptake in a group of animals that first received a predose of 0.2 mg/kg (250 μg protein) of the hRS7 ADC 3 days before the injection of the ¹¹¹In-labeled antibody was examined. Tumor uptake of ¹¹¹In-hRS7-CL2A-SN-38 in predosed mice was substantially reduced at every timepoint in comparison to animals that did not receive the predose (e.g., at 72 hours, predosed tumor uptake was 12.5% ± 3.8% ID/g vs. 25.4% ± 8.1% ID/g in animals not given the predose; P = 0.0123; Fig. 4E). Predosing had no appreciable impact on blood clearance or tissue uptake (Supplementary Table S2). These studies suggest that in some tumor models, tumor accretion of the specific antibody can be reduced by the preceding dose(s), which likely explains why the specificity of a therapeutic response could be diminished with increasing ADC doses and why further dose escalation is not indicated.

Swiss-Webster mice tolerated 2 doses over 3 days, each of 4, 8, and 12 mg SN-38/kg of the hRS7-CL2A-SN-38, with minimal transient weight loss (Supplementary Fig. S2). No hematopoietic toxicity occurred and serum chemistries only revealed elevated aspartate transaminase (AST) and alanine transaminase (ALT; Fig. 5). Seven days after treatment, AST rose above normal levels (>298 U/L) in all 3 treatment groups (Fig. 5A), with the largest proportion of mice being in the 2 × 8 mg/kg group. However, by 15 days posttreatment, most animals were within the normal range. ALT levels were also above the normal range (>77 U/L) within 7 days of treatment (Fig. 5B) and with evidence of normalization by Day 15. Livers from all these mice did not show histologic evidence of tissue damage (not shown). In terms of renal function, only glucose and chloride levels were somewhat elevated in the treated groups. At 2 × 8 mg/kg, 5 of 7 mice had slightly elevated glucose levels (range of 273–320 mg/dL, upper end of normal 263 mg/dL) that returned to normal by 15 days postinjection. Similarly, chloride levels were slightly elevated, ranging from 116 to 127 mmol/L (upper end of normal range 115 mmol/L) in the 2 highest dosage groups (57% in the 2 × 8 mg/kg group and 100% of the mice in the 2 × 12 mg/kg group), and remained elevated out to 15 days postinjection. This also could be indicative of gastrointestinal toxicity, because most chloride is obtained through absorption by the gut; however, at termination, there was no histologic evidence of tissue damage in any organ system examined (not shown).

Because mice do not express Trop-2 identified by hRS7, a more suitable model was required to determine the potential of the hRS7 conjugate for clinical use. Immunohistology studies revealed binding in multiple tissues in both humans and Cynomolgus monkeys (breast, eye, gastrointestinal tract, kidney, lung, ovary, fallopian tube, pancreas, parathyroid, prostate, salivary gland, skin, thymus, thyroid, tonsil, ureter, urinary bladder, and uterus; not shown). Based on this cross-reactivity, a tolerability study was performed in monkeys.

The group receiving 2 × 0.96 mg SN-38/kg of hRS7-CL2A-SN-38 had no significant clinical events following the infusion and through the termination of the study. Weight loss did not exceed 7.3% and returned to acclimation weights by day 15. Transient decreases were noted in most of the blood count data (neutrophil and platelet data shown in Fig. 5C and D), but values did not fall below normal ranges. No abnormal values were found in the serum chemistries. Histopathology of the animals necropsied on day 11 (8 days after last injection) showed microscopic changes in hematopoietic organs (thymus, mandibular and mesenteric lymph nodes, spleen, and bone marrow), gastrointestinal organs (stomach, duodenum, jejunum, ileum, cecum, colon, and rectum), female reproductive organs (ovary, uterus, and vagina), and at the injection site. These changes ranged from minimal to moderate and were fully reversed at the end of the recovery period (day 32) in all tissues, except in the thymus and gastrointestinal tract, which were trending towards full recovery at this later timepoint.

At the 2 × 1.92 mg SN-38/kg dose level of the conjugate, there was 1 death arising from gastrointestinal complications and bone marrow suppression, and other animals within this group showed similar, but more severe adverse events than the 2 × 0.96 mg/kg group. These data indicate that dose-limiting toxicities were identical to that of irinotecan; namely, intestinal and hematologic. Thus, the MTD for hRS7-CL2A-SN-38 lies between 2 × 0.96 and 1.92 mg SN-38/kg, which represents a human equivalent dose of 2 × 0.3 to 0.6 mg/kg SN-38.

Trop-2 is a protein expressed on many epithelial tumors, including lung, breast, colorectal, pancreas, prostate, and ovarian cancers, making it a potentially important target for delivering cytotoxic agents (7–9, 11, 13). The RS7 antibody internalizes when bound to Trop-2 (18), which enables direct intracellular delivery of cytotoxics.

Conjugation of chemotherapeutic drugs to antibodies has been explored for over 30 years (42, 43). Because a substantial portion of an ADC is not processed by the tumor, but by normal tissues, there is a risk that these agents will be too toxic to normal organ systems before reaching the therapeutic level in tumors. As with any therapeutic, the therapeutic window is a key factor determining the potential of an ADC, and thus rather than examining “ultratoxic” drugs, we chose SN-38 as the drug component of the Trop-2-targeted ADC.

SN-38 is a potent topoisomerase-I inhibitor, with IC₅₀ values in the nanomolar range in several cell lines. It is the active form of the prodrug, irinotecan, that is used for the treatment of colorectal cancer, and which also has activity in lung, breast, and brain cancers. We reasoned that a directly targeted SN-38, in the form of an ADC, would be a significantly improved therapeutic over CPT-11, by overcoming the latter's low and patient-variable bioconversion to active SN-38 (26).

The Phe-Lys peptide inserted in the original CL2 derivative allowed for possible cleavage via cathepsin B (44). In an effort to simplify the synthetic process, in CL2A, phenylalanine was eliminated, and thus the cathepsin B cleavage site was removed. Interestingly, this product had a better-defined chromatographic profile compared to the broad profile obtained with CL2 (not shown), but more importantly, this change had no impact on the conjugate's binding, stability, or potency in side-by-side testing. These data suggest that SN-38 in CL2 was released from the conjugate primarily by the cleavage at the pH-sensitive benzyl carbonate bond to SN-38's lactone ring and not the cathepsin B cleavage site (Fig. 1A).

In vitro cytotoxicity of hRS7 ADC against a range of solid tumor cell lines consistently had IC₅₀ values in the nmol/L range. However, cells exposed to free SN-38 demonstrated a lower IC₅₀ value compared to the ADC. This disparity between free and conjugated SN-38 was also reported for ENZ-2208 (28, 29) and NK012 (27). ENZ-2208 utilizes a branched PEG to link about 3.5 to 4 molecules of SN-38 per PEG, whereas NK012 is a micelle nanoparticle containing 20% SN-38 by weight. With our ADC, this disparity (i.e., ratio of potency with free vs. conjugated SN-38) decreased as the Trop-2 expression levels increased in the tumor cells, suggesting an advantage to targeted delivery of the drug. In terms of in vitro serum stability, both the CL2- and CL2A-SN-38 forms of hRS7-SN-38 yielded a t/_½ of ∼20 hours, which is in contrast to the short t/_½ of 12.3 minutes reported for ENZ-2208 (29), but similar to the 57% release of SN-38 from NK012 under physiological conditions after 24 hours (27).

Treatment of tumor-bearing mice with hRS7-SN-38 (either with CL2-SN-38 or CL2A-SN-38) significantly inhibited tumor growth in 5 different tumor models. In 4 of them, tumor regressions were observed, and in the case of Calu-3, all mice receiving the highest dose of hRS7-SN-38 were tumor-free at the conclusion of study. Unlike in humans, irinotecan is very efficiently converted to SN-38 by a plasma esterase in mice, with a greater than 50% conversion rate (38), and yielding higher efficacy in mice than in humans (39–41). When irinotecan was administered at 10-fold higher or equivalent SN-38 levels, hRS7-SN-38 was significantly better in controlling tumor growth. Only when irinotecan was administered at its MTD of 24 mg/kg q2dx5 (37.5-fold more SN-38) did it equal the effectiveness of hRS7-SN-38. In patients, we would expect this advantage to favor hRS7-CL2A-SN-38 even more, because the bioconversion of irinotecan would be substantially lower.

We also showed in some antigen-expressing cell lines, such as SK-MES-1, that using an antigen-binding ADC does not guarantee better therapeutic responses than a nonbinding, irrelevant conjugate. This is not an unusual or unexpected finding. Indeed, the nonbinding SN-38 conjugates mentioned earlier enhance therapeutic activity when compared to irinotecan, and so an irrelevant IgG-SN-38 conjugate is expected to have some activity. This is related to the fact that tumors have immature, leaky vessels that allow the passage of macromolecules better than normal tissues (45). With our conjugate, 50% of the SN-38 will be released in ∼13 hours when the pH is lowered to a level mimicking lysosomal levels (e.g., pH 5.3 at 37°C; data not shown), whereas at the neutral pH of serum, the release rate is reduced nearly 2-fold. If an irrelevant conjugate enters an acidic tumor microenvironment, it is expected to release some SN-38 locally. Other factors, such as tumor physiology and innate sensitivities to the drug, will also play a role in defining this “baseline” activity. However, a specific conjugate with a longer residence time should have enhanced potency over this baseline response as long as there is ample antigen to capture the specific antibody. Biodistribution studies in the SK-MES-1 model also showed that if tumor antigen becomes saturated as a consequence of successive dosing, tumor uptake of the specific conjugate is reduced, which yields therapeutic results similar to that found with an irrelevant conjugate.

Although it is challenging to make direct comparisons between our ADC and the published reports of other SN-38 delivery agents, some general observations can be made. In our therapy studies, the highest individual dose was 0.4 mg/kg of SN-38. In the Calu-3 model, only 4 injections were given for a total cumulative dose of 1.6 mg/kg SN-38 or 32 μg SN-38 in a 20 g mouse. Multiple studies with ENZ-2208 were done using its MTD of 10 mg/kg × 5 (28, 31), and preclinical studies with NK012 involved its MTD of 30 mg/kg × 3 (27). Thus, significant antitumor effects were obtained with hRS7-SN-38 at 30-fold and 55-fold less SN-38 equivalents than the reported doses in ENZ-2208 and NK012, respectively. Even with 10-fold less hRS7 ADC (0.04 mg/kg), significant antitumor effects were observed, whereas lower doses of ENZ-2208 were not presented, and when the NK012 dose was lowered 4-fold to 7.5 mg/kg, efficacy was lost (27). Normal mice showed no acute toxicity with a cumulative dose over 1 week of 24 mg/kg SN-38 (1,500 mg/kg of the conjugate), indicating that the MTD was higher. Thus, tumor-bearing animals were effectively treated with 7.5- to 15-fold lower amounts of SN-38 equivalents.

As a topoisomerase-I inhibitor, SN-38 induces significant damage to a cell's DNA, with upregulation of p53 and p21^WAF1/Cip1 resulting in caspase activation and cleavage of PARP (34–37). When we exposed BxPC-3 and Calu-3 cells to our ADC, both p53 and p21^WAF1/Cip1 were upregulated above basal levels. In addition, PARP cleavage was also evident in both cell lines, confirming an apoptotic event in these cells. Of interest was the higher upregulation of p21^WAF1/Cip1 in BxPC-3 and Calu-3 relative to p53 by both free SN-38 and our hRS7-SN-38. This may be indicative of the mutational status of p53 in these 2 cell lines (46–48) and the use of a p53-independent pathway for p21^WAF1/Cip1-mediated apoptosis (49).

An interesting observation was the early upregulation of p53 in both BxPC-3 and Calu-3 at 24 hours mediated by the hRS7-ADC relative to free SN-38. Even the naked hRS7 IgG could upregulate p53 in these cell lines, although only after a 48-hour exposure. Trop-2 overexpression and cross-linking by antibodies has been linked to several MAPK-related signaling events (11), as well as intracellular calcium release (5). While binding of hRS7 was not sufficient to induce apoptosis in BxPC-3 and Calu-3, as evidenced by the lack of PARP cleavage, it may be enough to prime a cell, such that the inclusion of SN-38 conjugated to hRS7 may lead to a greater effect on tumor growth inhibition. Studies are currently underway to understand which pathways are involved with hRS7-delivery of SN-38 and how they may differ from free SN-38, and what effect p53 status may play in this signaling.

Biodistribution studies revealed the hRS7-CL2A-SN-38 had similar tumor uptake as the parental hRS7 IgG, but cleared substantially faster with 2-fold higher hepatic uptake, which may be due to the hydrophobicity of SN-38. With the ADC being cleared through the liver, hepatic and gastrointestinal toxicities were expected to be dose limiting. Although mice had evidence of increased hepatic transaminases, gastrointestinal toxicity was mild at best, with only transient loss in weight and no abnormalities noted upon histopathologic examination. Interestingly, no hematological toxicity was noted. However, monkeys showed an identical toxicity profile as expected for irinotecan, with gastrointestinal and hematological toxicity being dose-limiting.

Because Trop-2 recognized by hRS7 is not expressed in mice, it was critically important to perform toxicity studies in monkeys that have a similar tissue expression of Trop-2 as humans. Monkeys tolerated 0.96 mg/kg/dose (∼12 mg/m²) with mild and reversible toxicity, which extrapolates to a human dose of ∼0.3 mg/kg/dose (∼11 mg/m²). In a Phase I clinical trial of NK012, patients with solid tumors tolerated 28 mg/m² of SN-38 every 3 weeks with Grade 4 neutropenia as dose-limiting toxicity (DLT; ref. 30). Similarly, Phase I clinical trials with ENZ-2208 revealed dose-limiting febrile neutropenia, with a recommendation to administer 10 mg/m² every 3 weeks or 16 mg/m² if patients were administered G-CSF (50, 51). Because monkeys tolerated a cumulative human equivalent dose of 22 mg/m², it is possible that even though hRS7 binds to a number of normal tissues, the MTD for a single treatment of the hRS7 ADC could be similar to that of the other nontargeting SN-38 agents. Indeed, the specificity of the anti–Trop-2 antibody did not appear to play a role in defining the DLT, because the toxicity profile was similar to that of irinotecan. More importantly, if antitumor activity can be achieved in humans as in mice that responded with human equivalent dose of just at 0.03 mg SN-38 equivalents/kg/dose, then significant antitumor responses could be realized clinically.

In conclusion, toxicology studies in monkeys, combined with in vivo human cancer xenograft models in mice, have indicated that this ADC targeting Trop-2 is an effective therapeutic in several tumors of different epithelial origin, supporting future clinical testing.

T.M. Cardillo, S.V. Govindan, P. Trisal, and D.M. Goldenberg, employment and other financial relationships with Immunomedics, Inc., including patent inventions. R.M. Sharkey has no financial disclosures.

We thank Dr. Sung-Ju Moon, Fatma Tat, and Agatha Sheerin for contributions to synthetic and conjugation chemistries, Anju Nair, Maria Zalath, Lou Osorio, and Ashraf Gomaa for their assistance with the animal studies, and Roberto Arrojo for technical assistance in performing the in vitro studies.

The study was supported in part by a grant from the National Cancer Institute of the NIH (CA114802-02; PI: S.V. Govindan).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.