Paclitaxel Enhances Therapeutic Efficacy of the F8-IL2 Immunocytokine to EDA-Fibronectin–Positive Metastatic Human Melanoma Xenografts Free

Melanoma is the most common nonhematopoietic cancer in young adults, and the incidence of the disease is increasing at an alarming pace worldwide (1). In a meta-analysis of 42 phase II cooperative group trials, median survival in metastatic stage IV melanoma was only 6.2 months, with a mean 1-year overall survival (OS) rate of 25.5% regardless of treatment regimen (2). Despite the recent success of targeted therapies against the BRAF protein—led by vemurafenib (PLX 4032; ref. 3)—metastatic melanoma remains a solid malignancy that deserves therapeutic options.

In the arena of chemotherapy dacarbazine is considered the benchmark treatment for advanced melanoma, despite a response rate of less than 13% to 20% (4). Paclitaxel is a reasonable second-line therapy that has been evaluated in numerous trials as monotherapy (5, 6) or as part of a combination with other therapies (7). In metastatic melanoma, single-agent paclitaxel has antitumor effects comparable with those reported for dacarbazine and other single-agent chemotherapies, with a response rate between 0% to 33% (6).

Immunotherapy is a well-recognized therapeutic tool for patients with advanced melanoma. Interleukin-2 (IL2) is a U.S. Food and Drug Administration (FDA)-approved agent for the treatment of metastatic melanoma, whereas IFN-α is a therapy option for adjuvant treatment of patients with resected melanoma. A meta-analysis reviewed 18 studies of chemotherapy combined with IL2 and IFN-α: although some of these studies reported durable tumor responses or increased progression-free survival (PFS), no regimen to date has improved OS (8, 9). Moreover, these chemoimmunotherapy combinations are generally associated with considerable toxicity.

Very recently ipilimumab, a monoclonal antibody (mAb), which blocks cytotoxic T-lymphocyte–associated antigen 4 (CTLA-4) to potentiate T-cell antitumor response, was shown to improve OS in patients with previously treated metastatic melanoma (10). The approval of ipilimumab by FDA has renewed interest in immunotherapy as a legitimate therapeutic approach for melanoma (11).

The selective delivery of bioactive agents (i.e., cytotoxic drugs, radionuclides, or immunostimulatory cytokines) to the tumor site is a promising avenue for the development of novel anticancer therapies. Vascular and tumor stroma–associated markers are particularly attractive targets for antibody-based delivery of therapeutic agents in view of their selective and accessible expression in solid tumors (12).

Tumor progression is generally accompanied by angiogenesis together with the reorganization of the extracellular matrix (ECM). In this regard, the de novo synthesis of oncofetal isoforms of fibronectin (Fn) has been extensively described during tissue remodeling (13). Fn variants are generated by alternative splicing of the ED-A, ED-B, and IIICS domains, as well as by de novo glycosylation. Recently the Fn isoform containing EDA has been shown to be associated to vascular and stromal structures in a wide variety of cancer types. Furthermore, Natali and colleagues described an abundant expression of EDA-Fn in primary melanomas and their metastases (14). A high-affinity human mAb directed to EDA-Fn, F8, has been shown to display tumor-targeting selectivity in biodistribution experiments (15). Moreover we have recently reported that this antibody fused to human IL2 (F8-IL2) improves the therapeutic performance of sunitinib, an inhibitor of angiogenesis, in a xenograft model of human kidney cancer (16).

In this study, we show that human metastatic melanoma lesions express high levels of perivascular and stromal EDA-Fn, which is detectable with the recombinant antibody F8, and that the immunocytokine F8-IL2 administrated in combination with paclitaxel significantly inhibits [60%–80% complete regressions (CR)] the growth of a human xenografted melanoma expressing a high level of EDA-Fn in nude mice. Furthermore F8-IL2 added to paclitaxel significantly reduced metastatic nodules to the lung of mice bearing metastatic melanoma. In vivo optical and dynamic contrast enhanced-MRI (DCE-MRI) and immunohistochemical data indicate that the ability of paclitaxel to potentiate the antitumor activity of F8-IL2 is associated with its capability to modify certain tumor vessel functionalities related to vascular perfusion and permeability and to promote host immune reaction. Our findings endorse the use of chemoimmunotherapy in the treatment of metastatic melanoma and suggest that paclitaxel is endowed with properties affecting the tumor stroma that deserve attention for the design of novel combination treatments.

The A375M is a metastatic variant derived from the A375 human melanoma (American Type Culture Collection; ref. 17); this cell line was characterized by short-tandem repeat profiling (AmpFlSTR Identifiler Plus PCR Amplification kit; Applied Biosystems) and compared with DNA fingerprinting databases. WM1552, WM115, and WM983A cell lines, derived from patient melanomas, were kindly provided by Dr. Meenhard Herlyn (Wistar Institute, Philadelphia, PA; ref. 18); the WM1552/5 is a tumorigenic variant derived from WM1552 (19) that underwent DNA fingerprinting analysis (as above) to confirm the identity with the parental line.

Stocks of cell lines were stored frozen in liquid nitrogen and kept in culture for no more than 8 weeks before injection. Cell culture conditions are detailed in Supplementary Methods.

Paclitaxel (Indena S.p.A.), dacarbazine (Sanofi Aventis), recombinant human IL2 (IL2, Proleukin; Novartis), immunocytokine F8-IL2 (16), the isotype control KSF-IL2 (16), (SIP)F8-Alexa Fluor 750 (15), and the anti-asialo GM1 antiserum (Wako Chemical GmbH) were prepared and delivered as described in Supplementary Methods.

Specimens of melanomas were from patients at diagnosis; ethical approval was obtained from the Local Research Ethics Committee of the Friedrich Schiller University, Jena, Germany. Xenograft specimens were from human melanoma growing intradermal in nude mice (20). The spatial distribution and the relation of EDA-Fn to vascular structures were shown with a double fluorescence labeling procedure for F8 reactive-EDA-Fn and CD31 (21), as described in Supplementary Methods.

Six- to 8-week-old female NCr-nu/nu mice (Harlan) were used. Procedures involving animals and their care were conducted in conformity with institutional guidelines that comply with national and international laws and policies (22, 23).

WM1552/5 (2 × 10⁶ cells) was transplanted subcutaneously in the flank of nude mice which were then randomized for treatment (N = 10–12 per group) when tumors reached approximately 100 to 150 mg. Mice were treated with paclitaxel (20 mg/kg i.v.), dacarbazine (160 mg/kg i.p.), F8-IL2 (20 μg/mouse i.v.), KSF-IL2 (20 μg/mouse i.v.), or unconjugated recombinant IL2 (6.6 μg/mouse i.v) as single agent or in combination regimens. For the depletion of natural killer (NK) cells, anti-asialo GM1 antiserum (200 μL/1:10 dilution i.v.) was administered along the therapy. Schedule of treatments are detailed in the Results. Animal management and data collection were carried out with the help of the Study Director 1.8 software (Studylog System, Inc.). Results were plotted as the mean tumor weight against days after tumor randomization and the efficacy of treatment was expressed as tumor weight and tumor growth delay (T − C = median time to reach 500 mg of treated tumor − median time to reach 500 mg of control tumor; ref. 22). CR was defined by absence of tumor for at least 3 months. Toxicity was monitored by recording body weight loss.

A375M (10 × 10⁵ cells) was injected intravenously in the tail vein of nude mice which were then randomized for treatment (N = 10 per arm) 7 days later. Mice were treated with paclitaxel, F8-IL2 (same dose as above) as single agent or in combination regimens as detailed in Results. Mice were killed and lung harvested after 6 weeks of treatment. Metastases were counted blinded by 2 independent investigators and expressed as number and size of tumor nodules in the lung (17). The procedures for subcutaneous tumor and metastasis evaluation are detailed in Supplementary Methods.

Uptake of (SIP)F8-Alexa Fluor 750 in mice bearing WM1552/5 tumors was analyzed with the eXplore Optix imaging system. DCE-MRI was done after injection of the B22956/1 contrast agent with a BioSpec 70/30 AVIII system (24). Data analysis is in Supplementary Methods.

Analysis was carried out on WM1552/5 tumors from the different treatment regimens harvested 24 hours after the end of the third cycle (N = 4/5 tumors per group). For vessel perfusion analysis, mice were given Hoechst 33342 (40 mg/kg i.v; Sigma) 1 minute before sacrificing. Tumor sections were CD31 stained for immunofluorescence microscopy (22) and immunostained with the antibody against Asialo-GM1 for the detection of tumor-infiltrating NK cells. Evaluation analysis is described in Supplementary Methods.

Tumor specimens obtained from patients with metastatic melanoma were evaluated for EDA-Fn expression using the human recombinant SIP format antibody F8. Melanomas in general showed an abundant deposition of F8 accessible EDA-Fn antigen in tumor vessels, as well as in the tumor interstitium or in the stroma. Figure 1A and B show representative lymph node metastases of a malignant melanoma with a polygonal epitheloid/partially spindle shaped cell appearance and abundant stroma induction (Fig. 1A) and of a malignant melanoma with the typically medullary growth pattern (Fig. 1B). Both are characterized by a strong vascular and stromal positivity for EDA-Fn. The expression of EDA-Fn was studied in a panel of human melanoma–derived xenografts, transplanted orthotopically in the derma of nude mice. Interestingly, the 4 human melanoma xenografts were characterized by a highly diffuse EDA-Fn expression that fully replicated that of the patient lesion (Fig. 1C–F). The WM1552/5 and the A375M metastatic melanoma were chosen to evaluate the therapeutic activity of the F8-IL2 immunocytokine in preclinical models.

F8-IL2 alone and in combination regimens was tested on human tumor xenografts. Treatments were done for 5 cycles with chemotherapy administered 24 hours before the immunoconjugate (Treatment schedule in Fig. 2A).

Figures 2B and C show the therapeutic effects of F8-IL2 in combination with paclitaxel or dacarbazine on WM1552/5 melanoma growing subcutaneously in nude mice. F8-IL2 and IL2 administered as single agents did not significantly affect tumor growth. Paclitaxel treatment caused a marginal, though significant, delay in growth compared with controls (T-C = 8); this delay was not improved by the addition of IL2 (T-C = 10). Only the addition of F8-IL2 to paclitaxel yielded a clear anti-melanoma activity and induced the complete eradication of WM1552/5 tumors (Fig. 2B). Nine of 11 (82%) mice were tumor free, as confirmed by histologic analysis at necropsy. The injection of F8 antibody plus, but not conjugated to, IL2 had no effect on tumor growth (data not shown).

By contrast, in a similar trial carried out with dacarbazine in combination with F8-IL2 (Fig. 2C), no significant therapeutic benefit was observed by combining the 2 treatments, despite the ability of dacarbazine to cause a significant, albeit limited, growth delay as single agent (T-C = 4 and T-C = 6, for dacarbazine alone and in combination, respectively). The administration of IL2 conjugated to KSF that recognizes an unrelated antigen (KSF-IL2), in combination with paclitaxel, did not show advantage compared with paclitaxel alone (Fig. 2D).

Noteworthy was the finding that the therapeutic performance of the combination therapies was not associated with any significant loss of body weight, indicating that the combination therapy regimens were well tolerated.

The influence of paclitaxel on the delivery of F8-IL2 was investigated in vivo by optical imaging of (SIP)F8-Alexa Fluor 750 tumor uptake in nude mice bearing subcutaneous WM1552/5 and pretreated with paclitaxel, dacarbazine or vehicle. Figure 3A shows representative images of vehicle-, dacarbazine-, and paclitaxel-treated mice, the latter presenting a distinguishably higher fluorescence signal, which correspond to a stronger uptake of (SIP)F8-Alexa Fluor 750 in the tumor, compared with vehicle- or dacarbazine-treated mice. To evaluate the uptake of (SIP)F8-Alexa Fluor 750 in tumors according to the different tumor weights (vehicle, mean = 172 mg; paclitaxel, mean = 91 mg; dacarbazine, mean = 136 mg; N = 4 per group), the fluorescence signals were calculated as the ratio between photon counts and tumor weights (Fig. 3B). Signals in the tumors of mice treated with paclitaxel was significantly higher compared with vehicle- or dacarbazine-treated animals, for both acquisitions done at 6 and 24 hours after (SIP)F8-Alexa Fluor 750 injection.

To elucidate whether the paclitaxel-induced increase of F8 antibody uptake could be attributable to an altered tumor physiology (25), perivascular Hoechst 33342 perfusion was measured in tumors from animals treated with paclitaxel 24 hours before (Fig. 3C). The area perfused by Hoechst 33342 was significantly higher in paclitaxel-treated tumors compared with controls (Fig. 3D). The injection of F8-IL2 or IL2 alone did not modify the perfusion of Hoechst 33342 in tumors (data not shown).

To obtain a more detailed analysis of tumor vessel functionality, paclitaxel- and vehicle-treated nude mice bearing WM1552/5 melanoma xenografts (N = 4 mice per group) were analyzed by DCE-MRI 24 hours after paclitaxel administration (Fig. 3E–G). To prevent perfusion and permeability from being influenced by the size of lesions, paclitaxel- and vehicle-treated tumors were weight-matched (mean = 603 mg and 503 mg, respectively).

The kinetics of the distribution of the contrast agent (B22956/1) in the tumor core and rim were different. Tumor rim, a 1-mm thick band starting from the tumor boundary, is known to be hypervascularized whereas the core is often characterized by the presence of necrotic areas (Fig. 3E I).

After paclitaxel treatment a significant increase in the fractional plasma volume (fPV), an MRI-based measure of tumor perfusion, and transfer coefficient (kTrans), an MRI-derived quantity related to the extravasation of the contrast agent, was observed both in tumor rim and core, suggesting that the treatment with paclitaxel was able to enhance the overall delivery of the albumin-binding contrast agent to the pathologic tissue (Fig. 3E–G; ref. 26).

Altogether, these data suggested that paclitaxel induced transient functional modifications in tumors vessels, which in turn could promote F8-IL2 delivery and tumor response.

To evaluate the impact of the findings above on drug scheduling and tumor response, additional trials comparing different sequences of drug administration were carried out (Fig. 4). The time interval between treatments was prolonged to 7 days to allow clearance of circulating F8-IL2. In this setting paclitaxel as single agent only moderately affected the growth of WM1552/5 human melanoma xenografts (T-C = 7). In the combination arm, the weekly schedule of F8-IL2 given 24 hours after paclitaxel (paclitaxel → F8-IL2) significantly inhibited WM1552/5 melanoma growth, with 5 of 11 (45.5%) CR, thus confirming the findings above (Fig. 4A). In the same study, F8-IL2 administered 4 hours after paclitaxel (paclitaxel + F8-IL2) showed a comparable therapeutic efficacy with 6 of 11 (54.5%) CR. Cured mice were still tumor-free after 4 months, when the experiment was terminated and mice necropsied. Noncured tumors, after a significant growth delay (T-C = 29; 6 tumors for paclitaxel → F8-IL2 and T-C = 30; 5 tumors for paclitaxel + F8-IL2), resumed their growth at the same rate as control-untreated tumors. By contrast, the efficacy of the combination was reduced with the inverted sequence of drug administration (F8-IL2 → paclitaxel; paclitaxel administered 48 hours after F8-IL2), but not significantly different from paclitaxel alone (Fig. 4B). With no mice being tumor-free at the end of the study, no CRs were observed in this trial.

The contribution of host immune cells to the therapeutic effects of F8-IL2 combined with paclitaxel was analyzed in WM1552/5 melanoma sections harvested from mice 24 hours after the third cycle of treatments. F8-IL2 administered after paclitaxel led to a massive infiltration of NK cells in WM1552/5 tumors that was significantly higher than in lesions treated with paclitaxel or F8-IL2 as single agents (Fig. 5A and B), whereas the inverted sequence with F8-IL2 administered after paclitaxel did not enhance NK cell infiltration (Fig. 5C and D). Accordingly, IL2 had no effect on NK cell infiltration, and F8-IL2 alone caused only a marginal recruitment. Notably, a significant increase of NK cells was seen in the paclitaxel-only treated tumors (Fig. 5A–D), suggesting that paclitaxel is endowed with immunomodulating properties (27). The host infiltration of controls at randomization (day 0, mean tumor weight = 120 mg) and on the day of testing (mean tumor weight = 500 mg) was not different.

To examine the implication of NK cells on the growth of WM1552/5, nude mice undergoing therapy with paclitaxel → F8-IL2 were depleted of NK cells with anti-asialo GM1. Figure 5E shows that in mice receiving anti-asialo GM1, paclitaxel → F8-IL2 was significantly less efficacious in inhibiting the growth of WM1552/5 xenografts and not significantly different from paclitaxel alone. Anti-asialo GM1 given to vehicle-treated tumor did not influence tumor growth. These findings point to the contribution of the host response to the therapeutic benefit of the combination.

The therapeutic efficacy of F8-IL2 immunocytokine in combination with paclitaxel was assessed on lung tumor nodules formed by the A375M melanoma. Treatment started 7 days after tumor injection, when tumor foci were established in the lung of nude mice (17). According to the findings above paclitaxel was given 24 hours before F8-IL2 and therapy repeated weekly for 6 cycles. As shown in the representative lungs of Fig. 6A, F8-IL2 did not affect metastasis, whereas paclitaxel-based chemotherapy inhibited the growth of tumor nodules (tumor burden, Fig. 6C), but not the overall number of colonies in the lung (colony number, Fig. 6B). Only the combination of paclitaxel with F8-IL2 significantly reduced both number and tumor burden of lung colonies (Fig. 6B and C and Supplementary Table S1).

Our study shows in mouse xenograft models of human melanoma that paclitaxel has a synergistic antitumor effect with the immunocytokine F8-IL2 by favoring its tumor uptake and potentiating its antitumor efficacy. This treatment modality induced CR of the WM1552/5 melanoma transplanted subcutis in nude mice and inhibited metastasis to the lung from the A375M.

The therapeutic performance of the F8-IL2-paclitaxel seems to be contingent on the pattern of positivity for EDA-Fn: in responsive melanoma, this Fn isoform was highly expressed in vessel walls and in the tumor interstitial space (Fig. 1). At variance the same combination was moderately active in a model of ovarian cancer in which EDA-Fn expression was restricted to the perivascular space only (Supplementary Results and Supplementary Fig. S1). Noteworthy, our findings are in agreement with the results of Natali and colleagues, showing EDA-Fn expression associated with melanoma metastatic lesions (14). Altogether these findings suggest that vascular targeting based therapy should be adapted to the individual features of patient's tumor; in the case of melanoma, this tumor type is putatively a fitting candidate for immunoconjugated based therapy using the F8 antibody as a vehicle.

Our data also show that the therapeutic efficacy of F8-IL2 is significantly improved by combining certain chemotherapeutic agents—here we show paclitaxel—but not others, such as dacarbazine. The increased uptake of fluorescent F8 antibody in the tumors of mice pretreated with paclitaxel, but not with dacarbazine, supports its use in this combination setting. Because Hoechst 33342 staining used as an indicator of vascular perfusion (28) was increased in the tumors of mice treated with paclitaxel, we hypothesized an increase of tumor perfusion. Surprisingly, although IL2 had previously been shown to increase vascular permeability in patient healthy tissue and in tumor-bearing mice (29, 30), a similar effect was not observed in our studies when the fusion protein F8-IL2 was used as single agent. The resistance of tumor tissue to IL2-induced increase vascular permeability could be expression of the angiogenic phenotype of tumor endothelial cells (more resistant to IL2 direct toxic effect) or to the different cytokine milieu of tumor tissue.

Subsequent DCE-MRI studies on mice pretreated with paclitaxel showed a significant increase in tumor perfusion (fPV) and in vessel permeability (kTrans), suggesting that these modifications improve distribution of the immunocytokine.

Of further note was our use as contrast agent of gadocoletic acid trisodium salt (B22956/1), which belongs to the class of albumin binder agents. Given that the fractional binding of the gadocoletate ion to animal and human serum albumin is estimated at around 90% (31), we can conclude that in our tumor model paclitaxel is ultimately able to induce an increase in B22956/1-labeled albumin extravasations (represented by the increase of kTrans values) partially through its ability to increase fPV in the leaky tumor vessels. The molecular weights of the F8 antibody in SIP format (80 kDa; imaging study) and in scFv-IL2 format (86 kDa for the noncovalent homodimer used for the therapy study) are comparable with the one of albumin. It is reasonable to conclude that one of the therapeutically relevant actions of paclitaxel consist in mediating an increased vascular permeability at the tumor site. The increased vascular permeability could be attributable to the effect of paclitaxel on tumor angiogenesis (32, 33). However at the dose/schedule used in this study we did not find significant changes in vessel number, area, and diameter (Supplementary Fig. S2), so increased perfusion/permeability is presumably not caused by a direct effect on tumor vascularization. Instead our findings are in line with results from other Authors. Jain and colleagues showed that paclitaxel is able to reduce tumor interstitial fluid pressure (IFP) and decompress tumor blood vessels. The observation that these effects are no more evident in taxane-resistant tumors supports the hypothesis that the solid stress generated by neoplastic cell proliferation determines the increase in tumor vascular resistance (through the compression of tumor vessels) and in IFP (34). Thus, a reduction in tumor cell density through the use of a conventional cytotoxic drug ultimately decompresses blood vessels and, hence, reduces microvascular resistance and IFP and increases tumor perfusion (34). However, the theory of solid stress in tumors only partially explains the biologic mechanisms by which taxanes induce changes in vessel parameters. Indeed, another study showed that paclitaxel and docetaxel lowered IFP and significantly increased albumin extravasation regardless of their cytotoxic activity, suggesting that these effects could be taxane specific and related to pharmacodynamics of these drugs (35).

Elevated IFP is the major physiologic barrier to the delivery of macromolecules. It reduces the minimal physiologic extravasation of albumin that normally accounts for interstitial oncotic pressure (36) and could explain the inadequate delivery and therapeutic efficacy of F8-IL2 when administered as monotherapy. With the coadministration of paclitaxel, the antibody is likely able to reach the tumor at the right plasma concentration, to extravasate and accumulate in the perivascular and interstitial space of the tumor.

The translational potential of these findings is substantiated by clinical studies. In breast cancer patients treated with neoadjuvant chemotherapy, paclitaxel decreased IFP and increased oxygenation, whereas doxorubicin did not produce any significant change (25); this finding suggests that at least these tumors would be better treated first with paclitaxel to reduce IFP and increase pO₂ to improve the delivery of subsequent therapy, in particular of large molecules such as antibodies. Further studies are needed to clarify the mechanism of this sequencing.

These findings, suggesting that tumors would be better treated with paclitaxel first to improve subsequent drug delivery, have prompted us to evaluate the sequencing of paclitaxel and F8-IL2 to maximize tumor response. We found that paclitaxel administered 4 or 24 hours before F8-IL2 was more efficacious than F8-IL2 first followed by paclitaxel. In the last sequence paclitaxel was administered 48 hours after F8-IL2, a time sufficient for the clearance of the antibody from circulation. Here we used a recombinant antibody in scFv format to develop an immunocytokine with adequate tumor targeting and rapid clearance (16, 37). Further investigations, however, will be necessary to ascertain whether paclitaxel can also increase the tumor uptake of conventional IgG-based therapeutics.

In previous studies with IL2-based immunocytokines, the antitumor activity was traced principally to NK cells (16, 38). Here we found that F8-IL2 combined with paclitaxel induced a massive infiltration of NK cells in WM1552/5 tumors that was significantly greater than what occurred in xenografts treated with either of the 2 agents singly. We suggest that paclitaxel-caused increased permeability leads to the accumulation of F8-IL2 in the interstitium of the tumor (where abundant EDA-Fn can be found), resulting in a local activation and mitogenic activity for NK cells. The hypothesis that NK cells mediate the therapeutic activity of the immunocytokine with paclitaxel was reinforced by the observation that NK cell depletion in tumor-bearing mice abolishes the therapeutic gain associated to the combined use of F8-IL2 with paclitaxel. Furthermore, an increase of F4-80⁺ macrophages in tumors from mice receiving F8-IL2/paclitaxel (Supplementary Fig. S3) was also observed, thus calling for a more complex implication of the host in the therapeutic activity of the immunocytokine.

Notably, a significant increase in Asialo-GM1⁺ NK cells was also observed in tumors treated with paclitaxel alone, suggesting that the taxane possesses immunomodulating properties. These findings speak for a role of paclitaxel in boosting this response to F8-IL2. Indeed, there is an emerging body of data confirming that paclitaxel, like other related taxanes, displays immunomodulatory properties that endorse its therapeutic application beyond tumor chemotherapy (27, 39).

We are aware of the possible limits of studies done in immunodeficient nude mice, in which most of the immunomodulating effects were underestimated and restricted to innate immunity (nude mice cannot generate mature T lymphocytes; ref. 40). Indeed, in some tumor models, T cells have also been shown to contribute to the action of IL2-based immunocytokines (41). Nevertheless, in a number of cancer models based on the grafting of murine tumors in syngeneic mouse strains, the therapeutic activity of IL2-based immunocytokines was comparable when immunocompetent mice or immunocompromised mice (which still had NK cells) were used (38).

Attempts to treat melanoma patients with high dose IL2 have been conducted for more than 2 decades (42). However, the modest benefits achieved by treatment regimens including this cytokine must be weighed against the serious side effects. In our study in melanoma-bearing mice the targeted delivery of IL2 to the tumor allowed repeated and well-tolerated administrations of F8-IL2; more importantly, the administration in combination with paclitaxel did not seem to cause any cumulative toxicity. Paralleling our findings 2 different IL2-based immunocytokines (L19-IL2 and F16-IL2; refs. 37, 38) are currently being investigated in phase II clinical trials for different indications and in combinations with cytotoxic agents (43). A large variation in response to IL2-based immunocytokine treatment, in combination with dacarbazine, has been observed in a recent phase IIa trial in patients with metastatic melanoma, ranging from patients who did not respond to treatment to a set of patients (29%) who enjoyed a RECIST-confirmed partial or complete response (43). It is still unclear whether these large interpatient differences are due to a variability in the uptake of the immunocytokine at site of disease, or rather to the immunologic features of individual patients (e.g., different tumor-associated antigens presented by the tumor, different cytokine environment, or different activity of T and NK cells).

In conclusion, these preclinical results point to a significant benefit resulting from the combination of the immunocytokine F8-IL2 with paclitaxel for the treatment of metastatic melanoma, a tumor that historically has been resistant to chemotherapy. Paclitaxel, when administered first, seems to enhance the antibody-targeted delivery of IL2 to tumors that selectively express EDA-Fn, thereby improving the therapeutic index of the cytokine. We found that paclitaxel alone was able to affect the growth of metastatic lesions, but only its combination with F8-IL2 significantly reduced tumor nodules in the lung of nude mice. Should our present findings be confirmed by further investigations, the EDA-Fn expression pattern in lesions and the scheduling of drug administration will prove to be critical factors to take into account with this new biopharmaceutical agent. As the control of visceral metastases remains a main challenge in the treatment of melanoma patients (44), the therapeutic modalities and results reported herein should have an impact on the design of immunocytokine-based clinical trials in patients with melanoma and other malignancies, in combination with chemotherapeutic agents.

D. Neri is cofounder and shareholder of Philochem, as well as owner of the F8 antibody. The other authors declared no potential conflicts of interest.

Conception and design: M. Moschetta, F. Pretto, H. Kosmehl, M.R. Bani, D. Neri, and R. Giavazzi.

Writing, review, and/or revision of the manuscript: M. Moschetta, F. Pretto, M.R. Bani, D. Neri, and R. Giavazzi.

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Pretto, A. Berndt, K. Galler, P. Richter, A. Bassi, P. Oliva, and E. Micotti.

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, and computational analysis): F. Pretto, A. Berndt, K. Galler, P. Richter, A. Bassi, G. Valbusa, D. Neri, and R. Giavazzi.

Development of methodology: P. Oliva and E. Micotti.

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Schwager, M. Kaspar, and E. Trachsel.

Production of F8-IL2 and KSF-IL2: K. Schwager.

Study supervision: M.R. Bani and R. Giavazzi.

Development of reagents for the execution of the therapy study: D. Neri.

The authors thank Antonietta Silini and Valentina Scarlato for technical assistance.

The research work was supported by the 7th EU Framework Program (ADAMANT HEALTH-F2-2008-201342 to R. Giavazzi, D. Neri, and E. Trachsel), the Italian Association for Cancer Research (R. Giavazzi) and Fondazione CARIPLO (no. 2008-2264 to R. Giavazzi).

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.