Downloaded from
`
`http://theoncologist.alphamedpress.org/
`
` by guest on April 17, 2018
`
`This material is protected by U.S. Copyright law.
`Unauthorized reproduction is prohibited.
`For reprints contact: Reprints@AlphaMedPress.com
`
`Sarcomas
`
`TNF-α in Cancer Treatment:
`Molecular Insights, Antitumor Effects, and Clinical Utility
`
`Remco van Horssen, Timo L. M. ten Hagen, Alexander M. M. Eggermont
`
`Department of Surgical Oncology, Erasmus MC–Daniel den Hoed Cancer Center, Rotterdam, The Netherlands
`
`Key Words. Cancer • TNF-α • TNFR-1 • Tumor vasculature • Isolated limb perfusion
`
`Learning Objectives
`After completing this course, the reader will be able to:
` 1. Discuss the role of TNF-a in cancer survival and apoptosis.
` 2. Describe the mechanism of chemotherapy potentiation by TNF-a.
` 3. Explain the selective targeting of tumor vasculature by TNF-a.
` 4. Discuss TNFR-1 and TNFR-2 signaling.
`
`CMECME
`
`Access and take the CME test online and receive 1 AMA PRA category 1 credit at CME.TheOncologist.com
`
`Abstract
`Tumor necrosis factor alpha (TNF-α), isolated 30 years
`ago, is a multifunctional cytokine playing a key role in
`apoptosis and cell survival as well as in inflammation
`and immunity. Although named for its antitumor prop-
`erties, TNF has been implicated in a wide spectrum of
`other diseases. The current use of TNF in cancer is in the
`regional treatment of locally advanced soft tissue sarco-
`mas and metastatic melanomas and other irresectable
`tumors of any histology to avoid amputation of the limb.
`It has been demonstrated in the isolated limb perfusion
`setting that TNF-α acts synergistically with cytostatic
`drugs. The interaction of TNF-α with TNF receptor 1
`and receptor 2 (TNFR-1, TNFR-2) activates several
`signal transduction pathways, leading to the diverse
`functions of TNF-α. The signaling molecules of TNFR-
`1 have been elucidated quite well, but regulation of the
`signaling remains unclear. Besides these molecular
`
`insights, laboratory experiments in the past decade have
`shed light upon TNF-α action during tumor treatment.
`Besides extravasation of erythrocytes and lymphocytes,
`leading to hemorrhagic necrosis, TNF-α targets the
`tumor-associated vasculature (TAV) by inducing hyper-
`permeability and destruction of the vascular lining. This
`results in an immediate effect of selective accumulation
`of cytostatic drugs inside the tumor and a late effect of
`destruction of the tumor vasculature. In this review,
`covering TNF-α from the molecule to the clinic, we pro-
`vide an overview of the use of TNF-α in cancer starting
`with molecular insights into TNFR-1 signaling and cel-
`lular mechanisms of the antitumor activities of TNF-α
`and ending with clinical response. In addition, possible
`factors modulating TNF-α actions are discussed. The
`Oncologist 2006;11:397–408
`
`Introduction
`cytokine involved in apoptosis, cell survival, inflammation,
`Tumor necrosis factor alpha (TNF-α) is a multifunctional
`and immunity acting via two receptors [1, 2]. Currently it
`Correspondence: Alexander M. M. Eggermont, M.D., Ph.D., Erasmus MC–Daniel den Hoed Cancer Center, Department of Surgical
`Oncology, 301 Groene Hilledijk, 3075 EA Rotterdam, The Netherlands. Telephone: 31-0-10-439-1911; Fax: 31-0-10-439-1011; e-mail:
`a.m.m.eggermont@erasmusmc.nl Received August 26, 2005; accepted for publication February 10, 2006. ©AlphaMed Press 1083-
`7159/2006/$20.00/0
`The Oncologist 2006;11:397–408 www.TheOncologist.com
`
`IPR2018-00685
`Celgene Ex. 2028, Page 1
`
`

`

`Downloaded from
`
`http://theoncologist.alphamedpress.org/
`
` by guest on April 17, 2018
`
`398
`
`TNF-α in Cancer Treatment
`
`is used in cancer treatment in the isolated limb perfusion
`(ILP) setting for soft tissue sarcoma (STS), irresectable
`tumors of various histological types, and melanoma in-
`transit metastases confined to the limb [3]. TNF-α was iso-
`lated in 1975 from the serum of mice treated with bacterial
`endotoxin as the active component of “Coley’s toxin” and
`was shown to induce hemorrhagic necrosis of mice tumors
`[4, 5]. It was almost a century ago that William Coley, a
`surgeon from New York, observed high fever and tumor
`necrosis in some cancer patients treated with his bacterial
`filtrate (“Coley’s mixed toxins”) [6]. A decade after its iso-
`lation, TNF-α was also characterized as “cachectin” and
`as T-lymphocyte differentiation factor [7, 8]. In 1984, the
`human TNF-α gene was cloned [9, 10], and a range of clini-
`cal experiments were set up, leading to a license from the
`European Agency for the Evaluation of Medicinal Products
`(EMEA) for the treatment of limb-threatening STS in an
`isolated perfusion setting [11].
`
`TNF-α and TNF Receptor 1 Signaling
`TNF-α is a 17-kDa protein consisting of 157 amino acids
`that is a homotrimer in solution. In humans, the gene is
`mapped to chromosome 6 [12]. Its bioactivity is mainly
`regulated by soluble TNF-α–binding receptors. TNF-α
`is mainly produced by activated macrophages, T lympho-
`cytes, and natural killer (NK) cells. Lower expression is
`known for a variety of other cells, including fibroblasts,
`smooth muscle cells, and tumor cells. In cells, TNF-α is
`synthesized as pro-TNF (26 kDa), which is membrane-
`bound and is released upon cleavage of its pro domain by
`TNF-converting enzyme (TACE) [13].
`As mentioned above, TNF-α acts via two distinct recep-
`tors [14]. Although the affinity for TNF receptor 2 (TNFR-
`2) is five times higher than that for TNFR-1 [15], the latter
`initiates the majority of the biological activities of TNF-α.
`TNFR-1 (p60) is expressed on all cell types, while TNFR-
`2 (p80) expression is mainly confined to immune cells [16].
`The major difference between the two receptors is the death
`domain (DD) of TNFR-1 that is absent in TNFR-2. For this
`reason, TNFR-1 is an important member of the death recep-
`tor family that shares the capability of inducing apoptotic
`cell death [17]. Besides this apoptotic signaling, TNFR-1
`is widely studied because it is a dual role receptor: next to
`induction of apoptosis, it also has the ability to transduce
`cell survival signals. Although signaling pathways are well
`defined nowadays, the life-death signaling regulation is still
`poorly understood [18, 19]. The TNFR-1 signaling pathways
`are depicted in Figure 1. Upon binding of the homotrimer
`TNF-α, TNFR-1 trimerizes, and silencer of death domain
`(SODD) protein is released [20]. TNFR-associated death
`domain (TRADD) binds to the DD of TNFR-1 and recruits
`
`the adaptor proteins receptor interacting protein (RIP),
`TNFR-associated factor 2 (TRAF-2), and Fas-associated
`death domain (FADD) [21]. In turn, these adaptor proteins
`recruit key molecules that are responsible for further intra-
`cellular signaling. When TNFR-1 signals apoptosis, FADD
`binds pro-caspase-8, which is subsequently activated. This
`activation initiates a protease cascade leading to apoptosis,
`also involving the mitochondria and with caspases as key
`regulators [22]. The ultimate event in this apoptotic signal-
`ing is the activation of endonucleases, like EndoG, resulting
`in DNA fragmentation. Alternatively, when TNFR-1 signals
`survival, TRAF-2 is recruited to the complex, which inhib-
`its apoptosis via cytoplasmic inhibitor of apoptosis protein
`(cIAP). The binding of TRAF-2 initiates a pathway of phos-
`phorylation steps resulting in the activation of cFos/cJun
`transcription factors via mitogen-activated protein kinase
`(MAPK) and cJun N-terminal kinase (JNK) [23]. The major
`signaling event of TRAF-2 and RIP is the widely studied
`activation of nuclear factor kappa B (NF-κB) transcription
`factor via NF-κB–inducing kinase (NIK) and the inhibitor
`of κB kinase (IKK) complex [24]. Both the NF-κB and cFos/
`cJun transcription factors induce transcription of antiapop-
`totic, proliferative, immunomodulatory, and inflammatory
`genes. NF-κB is the major survival factor in preventing TNF-
`α–induced apoptosis, and inhibition of this transcription fac-
`tor may improve the efficacy of apoptosis-inducing cancer
`therapies [25]. NF-κB activation in many human malignan-
`cies is aberrant or constitutive, and its role in the regulation of
`the apoptosis–proliferation balance in tumor cells indicates
`its role in oncogenesis [26, 27]. For further details on the dual
`signaling of TNFR-1, see Figure 1.
`
`Implications for Cellular Mechanisms
`Underlying TNF-α Effects During Solid
`Tumor Treatment
`It is widely known that TNF-α induces hemorrhagic necro-
`sis in a certain set of tumor types. To investigate the underly-
`ing mechanisms of TNF-α action during ILP of solid tumors
`in humans, we set up perfusion models in rats and reported
`that hemorrhagic necrosis was much greater in tumors
`treated with TNF-α and chemotherapeutic drugs [28]. In
`addition, we showed a synergistic antitumor effect of the
`combination treatment with TNF-α and chemotherapeutic
`drugs [29]. In contrast, TNF-α alone induced only a mild
`central necrosis, and there was no objective tumor response
`observed. The same rat models also revealed that the addi-
`tion of TNF-α improved the accumulation of chemothera-
`peutic drugs selectively in the tumor up to three- to sixfold.
`The augmented uptake of melphalan added to the molecular
`properties of this small molecule (distribution by gradient
`instead of convection) resulted in intratumoral concentra-
`TheOncologist®
`
`IPR2018-00685
`Celgene Ex. 2028, Page 2
`
`

`

`van Horssen, ten Hagen, Eggermont
`
`
`
`
`
`
`
`399
`
`Downloaded from
`
`http://theoncologist.alphamedpress.org/
`
` by guest on April 17, 2018
`
`Figure 1. Tumor necrosis factor receptor 1 (TNFR-1) signaling pathway. Tumor necrosis factor alpha (TNF-α) activates both
`survival and proliferation pathways along with apoptotic pathways via TNFR-1. The separate pathways are well defined, while the
`survival-death balance regulation remains unclear. Abbreviations: APAF-1, apoptosis protein activating factor 1; Bcl-2, B-cell
`lymphoma 2; Bid, Bak, Bax, and Bcl-XL, mitochondrial proteins of the Bcl-2 family; CAD, caspase-activated DNAse; Caspase-
`3/8/9, cysteine aspartase (apoptotic protease) 3/8/9; Cdc37, co-chaperon of HSP90; cIAP, cytoplasmic inhibitor of apoptosis; cFos/
`cJun, transcription factors; DD, death domain; EndoG, mitochondrial DNAse; FADD, Fas-associated DD; HSP90, heat shock
`protein 90; I-CAD, inhibitor of CAD; IκB, inhibitor of NF- κB; IKKα/β, IκB kinase; JNK, cJun n-terminal kinase; MEKK1, mito-
`gen-activated protein kinase/extracellular signal–related kinase kinase 1; MKK3/7, MAPK kinase 3/7; NEMO, NF-κB essential
`modulator; NF-κB, nuclear factor kappa B transcription factor; NIK, NF-κB inducing kinase; p38MAPK, p38 mitogen-activated
`protein kinase; RIP, receptor interacting protein; SODD, silencer of DD; sTNFR-1, soluble TNFR-1; TNF-α, tumor necrosis factor
`alpha; TNFR-1, TNF receptor 1; TRADD, TNF receptor-associated DD; TRAF-2, TNF receptor-associated factor-2.
`
`tions close to the 50% inhibitory concentration (IC50) in STS
`cells in vitro [30, 31]. These levels result in tumor cell kill
`in the ILP setting, and melphalan can distribute within the
`well-perfused parts of the tumor even though the intratu-
`moral pressure is high. This selective uptake of melphalan
`by the tumor was also observed when other vasoactive drugs
`were used in the ILP setting (see below). It is important to
`note that the cell lines we used were not sensitive to TNF-α
`in vitro, which is in accordance with other reports describing
`a lack of effect of TNF-α and no synergism with cytotoxic
`drugs in cell lines [32, 33]. Next to these ILP data, studies in
`mice and rats showed that a systemic low dose of TNF-α aug-
`ments the antitumor activity of pegylated liposomal doxoru-
`bicin [34, 35]. These observations are comprehensible clues
`
`that mechanisms underlying the TNF-α effect during solid
`tumor treatment cannot be caused by a direct cytotoxic or
`cytostatic effect of TNF-α toward the tumor cells. It was sug-
`gested that, rather than tumor cells themselves, cells of the
`tumor stroma may be responsible for the observed antitumor
`effect of TNF-α in patients. This hypothesis was confirmed
`by data from mice experiments revealing that TNF-α had a
`cytotoxic effect on tumor vasculature [36].
`
`Angiogenesis and Tumor-Associated
`Vasculature
`Angiogenesis, the formation of new blood vessels from pre-
`existing ones, has become a major field of research, mainly
`in cancer [37]. Angiogenesis is essential for a tumor to pro-
`
`www.TheOncologist.com
`
`IPR2018-00685
`Celgene Ex. 2028, Page 3
`
`

`

`Downloaded from
`
`http://theoncologist.alphamedpress.org/
`
` by guest on April 17, 2018
`
`400
`
`TNF-α in Cancer Treatment
`
`vide the tumor cells with oxygen and essential nutrients for
`growth and to metastasize hematogenically [38]. A growing
`tumor activates surrounding vessels by secreting angiogenic
`factors, thereby changing the dormant tumor phenotype
`toward an angiogenic one, the so-called “angiogenic switch”
`[39]. Activated endothelial cells have to migrate toward the
`tumor along a newly formed matrix, the components of which
`are synthesized by themselves, tumor cells, and other cells
`such as macrophages and fibroblasts [40]. Figure 2 shows
`schematically the process of tumor angiogenesis, which can
`be divided into four different stages. A small, dormant tumor
`(stage 1) can, depending on the nature of the tumor and its
`microenvironment, make the angiogenic switch to ensure
`exponential growth. The tumor secretes growth factors to
`activate endothelial cells of surrounding vessels (stage 2).
`Upon activation, these endothelial cells start to migrate and
`proliferate toward the tumor. Only one endothelial cell starts
`an angiogenic sprout and develops into an endothelial tip cell
`migrating along the extracellular matrix (ECM) and guid-
`ing the following so-called stalk endothelial cells (stage 3)
`[41]. Finally, the growing tumor is connected to the vascu-
`lature (stage 4). In addition to growth and proliferation, the
`tumor can metastasize. Malignant tumor cells, by invasion
`of the vessels, ECM degradation, attachment, and homing
`to target sites can form distal metastases [42]. The process
`of tumor angiogenesis results in a tumor-associated vascula-
`ture (TAV) that is rather chaotic, both in structure and func-
`tion. In comparison with normal vessels, tumor vessels have
`a noncontinuous endothelium, an enlarged basal membrane,
`
`and an aberrant pericyte coverage [43]. Frequently in tumors,
`the vascular hierarchy of arterioles, capillaries, and venules
`is absent, resulting in loosely associated pericytes [44]. From
`animal experiments, it is known that pericytes are present in
`small tumors and more abundant in large tumors [45]. The
`contribution of pericytes to (anti)-angiogenic therapies is
`currently an attractive focus of research. On one hand, these
`characteristics impair tumor blood flow, delivery of oxygen,
`and therapeutics to the tumor cells and vessel functionality,
`but on the other hand, these differences may be used as a tar-
`get. The solid tumors treated by ILP with TNF-α have a mas-
`sive vascular structure consisting of vessels with a pheno-
`type specific to tumor vessels, although detailed study needs
`to clarify the exact contribution of the TAV to the observed
`antitumor responses.
`
`Activity of TNF-α in Solid Tumors:
`Hypothetical Mechanism
`The vascular differences mentioned above are depicted in
`Figure 3A. These differences are responsible for a more leaky
`vasculature in the tumor, with average intraendothelial gaps
`of 400 nm, depending on the tumor type [46]. Blood cells such
`as lymphocytes and monocytes easily adhere and extrava-
`sate into the tumor. We speculate that the endothelial cells
`of the tumor vessels, compared with normal vessels, have an
`upregulation of TNFR-1 on their membranes, which may be
`dependent on TNFR-1–upregulating factors produced by
`vessel-surrounding cells like tumor cells and macrophages.
`This upregulation, along with the specific architecture of the
`
`Figure 2. The sequential steps during tumor angiogenesis. The dormant tumor in stage 1 starts to secrete angiogenic growth factors
`(GF) after its “angiogenic switch”, which is accomplished by an imbalance in pro- and antiangiogenic factors. These GFs activate
`endothelial cells of surrounding vessels, and these cells start to migrate (stage 2) and proliferate toward the tumor. An endothelial tip
`cell (TC) is guiding this sprouting process (stage 3). In stage 4, the novel sprout has formed a lumen and the tumor is connected to the
`vasculature, thereby ensuring its growth and enabling it to metastasize hematogenically. Abbreviation: ECM, extracellular matrix.
`
`TheOncologist®
`
`IPR2018-00685
`Celgene Ex. 2028, Page 4
`
`

`

`van Horssen, ten Hagen, Eggermont
`
`
`
`
`
`
`
`401
`
`Downloaded from
`
`http://theoncologist.alphamedpress.org/
`
` by guest on April 17, 2018
`
`Figure 3. Differences between healthy and tumor endothelium and a proposed mechanism of the tumor necrosis factor alpha
`(TNF-α) effect. (A): Schematic representation of a vessel with healthy endothelial lining (upper) and tumor endothelial lining
`(lower). Healthy endothelium is a continuous lining of endothelial cells, covered with pericytes and with a thin basement mem-
`brane. The permeability is low, and extravasation is tightly organized. In contrast, tumor endothelium does not consist of a con-
`tinuous lining of endothelial cells; it lacks pericyte coverage, and the basement membrane is thickened. This phenotype results
`in greater permeability of the tumor vessel (small arrows). We hypothesize that tumor endothelium exhibits a higher TNFR-1
`expression level because of TNFR-1–upregulating factors produced by vessel-surrounding cells. (B): Upon TNF-α treatment
`with isolated limb perfusion, the healthy endothelium stays intact because it is TNF-α insensitive by its lack of TNFR-1 expres-
`sion on the membrane. Tumor endothelium binds TNF-α, which affects the endothelial phenotype and induces apoptosis in some
`endothelial cells. These two processes result in an enormous induction of vessel permeability (big arrows). As a result, the chemo-
`therapy drug is well distributed throughout the tumor, and a strong extravasation of erythrocytes results in massive hemorrhagic
`tumor necrosis. The selectively targeted tumor vessels are no longer functional and regress.
`
`endothelial lining, defines the tumor vessels as a specific tar-
`get for TNF-α treatment (Fig. 3B). When TNF-α is adminis-
`tered via ILP to treat solid tumors, it binds soluble receptors,
`and because of the high dosage, TNFR-1 receptors on tumor
`endothelial cells become occupied. Healthy endothelium, in
`contrast, also binds TNF-α; however, because of a lower num-
`ber of membrane-bound TNFR-1 receptors (most TNFR-1 is
`stored in the golgi apparatus [47]), there is no toxicity. We
`propose that this TNF-α to TNFR-1 binding results in hyper-
`permeability of the tumor vessels, and erythrocytes and other
`blood cells extravasate. The strong extravasation of erythro-
`cytes results in massive hemorrhagic necrosis of the tumor.
`As a result of the direct cytotoxicity of high-dose TNF-α to
`endothelial cells, some of these cells undergo apoptosis, and
`this process strongly enhances the induced hyperpermeabil-
`
`ity (Fig. 3B). Several studies have shown that a lower dose of
`TNF-α results in comparable responses [48, 49], suggesting
`that a lower dose still may induce these antivascular effects.
`The healthy vessels, however, stay intact; no apoptosis and
`no extravasation occurs. The observed synergistic activity of
`TNF-α and chemotherapeutic drugs is a consequence of this
`double-induced hyperpermeability. This hyperpermeability
`throughout the tumor facilitates the augmented accumula-
`tion and distribution of the drug in the tumor, resulting in bet-
`ter exposure of the tumor cells to the cytostatic agent [30].
`This double-induced hyperpermeability, along with the dual
`targeting—the TAV (by TNF-α) and the tumor cells (by the
`chemotherapy drug)—is one explanation for the observed
`synergistic response of tumors to TNF-α and chemotherapy
`that results in high response rates in patients.
`
`www.TheOncologist.com
`
`IPR2018-00685
`Celgene Ex. 2028, Page 5
`
`

`

`Downloaded from
`
`http://theoncologist.alphamedpress.org/
`
` by guest on April 17, 2018
`
`402
`
`TNF-α in Cancer Treatment
`
`Clinical Efficacy of TNF-α–Based Isolated
`Limb Perfusion
`The use of TNF-α in the ILP setting was pioneered by Lien-
`ard et al. [50]. In 19 melanoma patients and a few STS cases,
`impressive and very rapid responses were observed. This
`observation was followed by multicenter trials in patients
`
`with locally advanced STS and melanoma. In Table 1, we
`present an overview of the multicenter trials in Europe
`that led to the approval of TNF-α by the EMEA in 1998
`for its application in ILP for the treatment of high-grade
`(2–3) STS. In these multicenter trials, an overall response
`rate of 76% and a median limb salvage rate of 82% were
`
`Table 1. Principle studies on tumor necrosis factor (TNF)–based isolated limb perfusion (ILP) for irresectable soft tissue sarcomas
`ILP Studies
`Pts
`CR
`PR
`NC/PD
`Limb Salvage
`(n)
`Drugs
`(%)
`(%)
`(%)
`(%)
`Pivotal Multicenter Studies
`TNF + IFN + M
`TNF + IFN + M
`
`20
`59
`
`TNF ± IFN + M
`
`TNF ± IFN + M
`
`195
`
`270
`196
`
`55 a
`18 a
`36 b
`18 a
`29 b
`28 c
`17 c
`
`40 a
`64 a
`51 b
`57 a
`53 b
`48 c
`48 b
`
`5 a
`18 a
`13 b
`25 a
`18 b
`24 c
`35 b
`
`Reference
`
`90
`84
`
`Eggermont et al. [51]
`Eggermont et al. [52]
`
`82
`
`76
`71 d
`
`Eggermont et al. [53]
`
`Eggermont et al. [54]
`
`Single-center studies (>20 pts)
`TNF ± IFN + M
`TNF ± IFN + M
`TNF + Dox
`TNF ± IFN + M
`TNF ± IFN + M
`TNF ± IFN + M
`TNF ± IFN + M j
`TNF ± IFN + M i
`TNF ± IFN + M k
`TNF ± IFN + M g
`TNF + Dox g
`
`TNF ± IFN + M
`
`TNF + M g
`
`35
`34
`20
`22
`55
`49
`29
`64
`29
`37
`21
`
`217
`
`100
`
`85
`85
`85
`77
`84
`58
`76
`82
`65
`97
`71
`
`87
`
`87
`
`82
`
`Gutman et al. [55]
`Olieman et al. [56]
`Rossi et al. [57]
`Lejeune et al. [58]
`Hohenberger et al. [59]
`Noorda et al. [60]
`van Etten et al. [61]
`Grünhagen et al. [62]
`Lans et al. [63]
`Grünhagen et al. [64]
`Rossi et al. [65]
`
`Grünhagen et al. [66]
`
`Bonvalot et al. [49]
`
`Grünhagen et al. [67]
`
`37 b
`9 b
`54 b
`35 b
`6 b
`59 b
`26 e
`10 e
`64 e
`18 b
`18 b
`64 b
`-
`-
`-
`8 b
`37 b
`55 b
`18 b
`38 b
`38 b
`13 b
`45 b
`42 b
`30 b
`50 b
`20 b
`16 b
`68 b
`16 b
`38 a
`57 a
`5 a
`55 b
`10 b
`35 b
`31 a
`51 a
`18 a
`26 b
`25 b
`49 b
`34 f
`17 f
`49 f
`35 f
`43 f
`22 f
`26 b
`50 b
`24 b
`240
`TNF ± IFN + M g
`a Objective clinical response rate by World Health Organization criteria.
`b CR, clinical CR or 100% necrosis; PR, clinical PR or >50%–90% necrosis.
`c CR only recognized by the European Medicines Agency when histopathology showed 100% necrosis.
`d Independent committee recognized 196 patients as pure amputation candidates.
`e No clinical response data; CR >90%; PR, radiological and/or hisopathological >50% necrosis.
`f CR/PR, loss of vasculature on ultrasound MRI; lower panel CR, >90% necrosis on histopathology.
`g Low-dose TNF of 1 mg or various doses of TNF-α (0.5–4 mg); Grünhagen et al. (1–4 mg), soft tissue sarcomas patients.
`h Patients with metastatic disease.
`i Patients with multiple tumors in extremity.
`j Patients >75 years old.
`k Patients with recurrent sarcomas in 60–70 Gray irradiated fields.
`Abbreviations: CR, complete remission; IFN, interferon gamma; M, melphalan; NC, no change; PD, progressive disease; PR,
`partial remission; Pts, patients.
`
`TheOncologist®
`
`IPR2018-00685
`Celgene Ex. 2028, Page 6
`
`

`

`Downloaded from
`
`http://theoncologist.alphamedpress.org/
`
` by guest on April 17, 2018
`
`van Horssen, ten Hagen, Eggermont
`
`
`
`
`
`
`
`403
`
`observed. Moreover, this table lists the largest single-center
`studies in STS that confirm the results of the multicenter
`experience [49, 51–67]. We observe strikingly consistent
`major response rates, with a median of 76% (range, 58%–
`91%), and with a median limb salvage rate of 84% (range,
`58%–97%). TNF-α–based ILP now is performed in 35
`cancer centers in Europe with national referral patterns
`for limb salvage. ILP with melphalan alone for melanoma
`in-transit metastases is reported in the literature to result
`in about a 50% complete response (CR) rate and an 80%
`overall response rate [68]. The introduction of TNF-α in
`this setting was reported to increase CR rates to 70%–90%
`and overall response rates to 95%–100%. These results are
`summarized in Table 2 [50, 69–78]. Early on, however, it
`was observed that ILP with TNF-α plus melphalan (TM-
`ILP) was especially effective against bulky tumors such as
`STSs, in which ILP with melphalan alone [79] or doxoru-
`bicin alone [79, 80] fails. It should also be noted, of course,
`that both drugs have no activity against melanoma in the
`systemic setting and that melphalan has no activity against
`STS in the systemic setting. TNF-α–based ILP with mel-
`phalan or doxorubicin results in similar tumor response
`rates, but because of less locoregional toxicity, melphalan
`is preferred over doxorubicin in the ILP setting [65–67]. In
`our own series of 50 ILPs in patients with bulky melanoma
`in-transit metastases, the CR rate was still 58% [73], identi-
`cal to the CR rate that was seen in an interim analysis of
`a randomized trial by Fraker et al. [74], in which TNF-α–
`
`based ILP was shown to be of significant benefit in patients
`with a high tumor load, increasing the CR rate from 19% for
`M-ILP to 58% for TM-ILP. Apart from bulky melanoma, a
`further indication for TNF-α–based ILP is response failure
`to a prior ILP because excellent response rates have been
`reported in this situation [76, 77]. Similarly, high response
`rates have been reported for TNF-α–based ILP for nonmel-
`anoma locally advanced skin cancers [78]. Because TNF-α
`acts primarily on the tumor vasculature, these observations
`make sense, and the propensity to respond to a TNF-α–
`based ILP is assumed to depend more on tumor vasculature
`than on the histologic type of the tumor.
`Response of STS to TNF-α–based ILP is shown in Fig-
`ure 4A. Magnetic resonance imaging of a patient with high-
`grade (6–7) leiomyosarcoma in the upper leg shows clear
`dark tumor masses with high gadolinium uptake before
`ILP. Five weeks after ILP, all tumor masses are gadolin-
`ium-negative. Along the distal femur, only small tumor
`remnants are visible, but at the proximal femur, a large but
`gadolinium-negative tumor mass without signs of regres-
`sion is visible. All lesions were resected and found to be
`100% necrotic. Thus the response was classified as a histo-
`pathologic CR.
`Targeting by TNF-α of the tumor vasculature is
`revealed in patients by angiographies before and after ILP.
`The TAV is selectively destroyed by TNF-α–based ILP;
`the TAV is gone while normal vessels of the limbs are still
`intact after ILP (Fig. 4B). TNF-α targets the vasculature
`
`Table 2. Tumor necrosis factor (TNF)–based isolated limb perfusion (ILP) in melanoma and nonmelanoma skin cancer patients.
`TTLP
`Median (Mo) Reference
`8+
`Lienard et al. [50]
`18+
`Lejeune et al. [69]
`26
`Eggermont et al. [70]
`ns
`Fraker et al. [71]
`14
`Lienard et al. [72]
`14
`Lienard et al. [72]
`16+
`Grünhagen et al. [73]
`ns
`Fraker et al. [74]
`ns
`Rossi et al. [75]
`8
`Grünhagen et al. [73]
`6
`Bartlett et al. [76]
`14
`Grünhagen et al. [77]
`20+
`Olieman et al. [78]
`
`CR (%)
`89
`90
`88
`76
`78
`69
`69
`59
`70
`58
`65
`75
`60
`
`PR (%)
`11
`10
`12
`16
`22
`22
`26
`16
`25
`34
`29
`25
`27
`
`Overall RR
`100 a
`100 a
`100 a
`92 a
`100 a
`91
`95
`75
`95
`83
`94
`100
`87
`
`Nonmelanoma skin cancers
`a TNF + IFN
`Abbreviations: CR, complete remission; Overall RR, overall (CR + PR) response rate; PR, partial remission rate; Pts, patients;
`TTLP, time to local progression; ns, not specified.
`
`ILP
`All melanoma patients
`
`Bulky melanoma only
`
`Repeat ILP for melanoma
`
`Pts (n)
`19
`44
`58
`26
`32
`32
`100
`39
`20
`50
`17
`26
`15
`
`www.TheOncologist.com
`
`IPR2018-00685
`Celgene Ex. 2028, Page 7
`
`

`

`404
`
`TNF-α in Cancer Treatment
`
`Downloaded from
`
`http://theoncologist.alphamedpress.org/
`
` by guest on April 17, 2018
`
`vascular TNF-α effects are achieved by the high concen-
`tration reached during ILP. At these high levels, TNF-α
`activity is antivascular and antiangiogenic, while at lower
`concentration TNF-α is known to promote angiogenesis
`[83]. In addition to direct TAV-mediated effects, TNF-α
`reduces blood flow in tumors in a dose-dependent fashion
`[84]. This set of antivascular TNF-α effects was recently
`confirmed by experiments revealing that tumor response
`to TNF-α correlates with the degree of tumor vascularity
`[85]. Along with this dual role in angiogenesis, TNF-α is
`also known for its dual role in cancer treatment, anti-TNF-
`α therapy is also used for several types of cancer. The anti-
`neoplastic and tumor-promoting effects of TNF-α are dis-
`cussed in a recent review [86].
`
`Approaches to Modulate TNF-α Action in
`Cancer Treatment
`High response rates in the ILP setting do not avoid the need
`to search for factors that modulate the TNF-α effect in solid
`tumors. In addition to the possible application of TNF-α in
`other settings (e.g., systemic treatment) and for other tumors
`types, nonresponding patients in the ILP setting may also
`benefit from TNF-α sensitizers. Some of these approaches
`are mentioned below. An obvious target is inhibition of the
`NF-κB survival pathway. Inactivation of NF-κB is known
`to sensitize several tumors to TNF-α [87]. NF-κB can be
`blocked in several ways: overexpression of its inhibitor IκB
`and selective NF-κB inhibitors have been shown to increase
`TNF-α–induced apoptosis of tumor cells [88, 89]. One
`such inhibitor, bortezomib, has entered the clinical arena as
`a combination therapy with chemotoxic drugs for prostate
`cancer and myeloma [90].
`Nitric oxide (NO) is involved in survival of TNF-α–
`treated cells through NF-κB–induced expression of induc-
`ible NO synthase (iNOS) [91, 92]. We have previously
`shown that inhibition of NOS by the addition of L-NAME
`(NG-nitro-L-arginine methyl ester) during TNF-α–based
`ILP resulted in an increased tumor response in rats bear-
`ing STSs [93]. These observations were confirmed by a
`recent study showing that NOS inhibition in endothelial
`cells reduces their sensitivity to TNF-α in vitro, leading to
`the hypothesis that tumor vessels exhibit a higher level of
`NOS, which might explain their higher TNF-α sensitivity
`[94]. These studies justify further evaluation of NOS inhi-
`bition in tumors of patients treated by ILP to stimulate the
`anti-TAV activities of TNF-α.
`Apoptosis induced by TNF-α is also associated with
`the generation of reactive oxygen species (ROS). It has
`been shown that the key survival factor NF-κB induces
`ROS-neutralizing enzymes like superoxide dismutase
`[95]. Induction of ROS production or an inhibition of the
`TheOncologist®
`
`Figure 4. Antitumor and antivascular effects of tumor necro-
`sis factor alpha (TNF-α) upon isolated limb perfusion (ILP)
`treatment of sarcoma and melanoma patients. (A): Magnetic
`resonance imaging of a patient with a high-grade leiomyosar-
`coma in the upper leg showing the tumor mass before treatment
`on the left. Five weeks after ILP with TNF-α and melphalan,
`there is no gadolinium uptake in the tumor remnants (right).
`Tumor remnants were resected, and all were necrotic. (B):
`Angiographies of two patients with rapidly growing sarcomas
`in the leg before (left) and after (right) ILP with TNF-α and
`melphalan. Angiographies clearly show the well-developed
`tumor vasculature before ILP, which is selectively destroyed
`after treatment while the normal vessels are still present and
`intact. Both patients were classified as complete responders.
`(C): Endothelial lining of tumor vessels is destroyed. Endo-
`thelial staining (CD31) of vessels in normal skin (left) and
`melanoma (right) in biopsies of a melanoma patient taken after
`ILP. Vessels in the normal skin show a continuous endothelial
`lining (arrow), while in the melanoma-associated vessel, this
`lining is disrupted and the endothelial cells detach from the
`basement membrane (arrowheads). The elastic fibers (stars,
`stained aspecific) in the thickened basement membrane are
`also visible in the melanoma vessel. Scale bar, 50 μm.
`
`of tumors with completely different histologies, but as the
`TAV is well developed in all these tumors, combination
`therapy in the ILP setting is very effective for the specific
`tumors treated. Synergistic and high response rates are
`achieved in sarcomas consisting of a broad range of sub-
`types, as well as in melanomas. At the histopathological
`level, massive hemorrhagic necrosis is observed inside
`melanomas treated with ILP [81], an effect likely caused
`by TNF-α–induced coagulation and extravasation of
`erythrocytes [82]. In accordance with the angiographies
`of STS, the vascular lining of m