Systemic cancer therapy: Evolution over the last 60 years

Authors


Abstract

The 1940s marked the beginning of an era of important discoveries that contributed to modern concepts underlying the current practice of cancer chemotherapy, such as the log kill hypothesis reported by Skipper, the Norton-Simon hypothesis, and the Goldie-Coldman hypothesis. The early success of nitrogen mustards and antifolates in the treatment of hematologic malignancies paved the way for drug discovery platforms, which resulted in the generation of more drugs that nonetheless predominantly are genotoxic. The turn of the new millennium marked a new phase in the evolution of cancer chemotherapy. Scientific progress in the preceding 60 years elucidated the important ideas behind tumor microenvironment and ‘targeted’ therapy that had their inception in the late 19th century. Breakthroughs in molecular biology have paved the way for the development of novel agents that modulate the dysregulated molecular pathways implicated in carcinogenesis. The key approaches and evidence pertinent to the clinical development of these novel agents are presented in this review. Cancer 2008;113(7 suppl):1857–87. © 2008 American Cancer Society.

Cancer chemotherapy or ‘chemical therapy’ in the narrowest sense is used to refer to genotoxic drugs that were developed in the 1940s for the treatment of cancer. Breakthroughs in molecular biology and the widespread adoption of bioinformatics and high-throughput ‘-omic’ (eg, genomic, transcriptomic, proteomic, etc) approaches have shed some light into the search for distinctive molecular signatures pertinent to ‘rational’ therapy against targets that are relevant in various disease processes such as cancer. In this review, we present a concise summary of systemic cancer therapy over the last 60 years focusing in detail on the au courant paradigm in cancer therapeutics that continues to evolve in the new millennium.

Brief History

The concept of treating cancer with drugs goes back more than 500 years when preparations of mercury, silver, and zinc were used. The first documented case of cancer chemotherapy, however, was by Lissauer in 1865.1 He gave Fowler solution (potassium arsenite) to a patient with leukemia with ‘beneficial effects.’ A precursor of current targeted cancer therapy is another 19th century discovery that laid the groundwork for antiestrogen therapies in breast cancer when Schinzinger first proposed surgical oophorectomy in 1889 as a treatment for breast cancer.2 Historic credit for this intervention has been accorded subsequently to Beatson, who first reported in 1896 the complete remission of a patient with advanced breast cancer after bilateral oophorectomy.3

The modern era of chemotherapy started when the war gas sulfur mustard was used in patients in 1931,4 first topically and then by direct intratumoral injection, but was believed to be too toxic for human use. The second-generation mustard, nitrogen mustard, was tested subsequently by Gilman et al, first in mice and then in a patient with non-Hodgkin lymphoma in the 1940s, extrapolating from the autopsy findings of lymphoid hypoplasia and myelosuppression in soldiers who had been exposed to the sulfur mustard gas during the First World War.5 Further investigation of the mustard agents defined the mechanism of action based on alkylation of DNA, thus spurring the subsequent development of newer generation alkylating agents, such as cyclophosphamide and chlorambucil. Closely after along the trail of discovery were the antimetabolites, with the antifolates aminopterin and methotrexate representing the first agents in this class of drugs to show therapeutic efficacy by inducing remission in children with acute lymphoblastic leukemia (ALL) based on the observation that folate supplementation accelerated the proliferation of leukemic cells in children with ALL.6, 7

From these beginnings arose the enthusiasm in promoting drug discovery for cancer. In 1950, anthracyclines that possessed antibacterial activity were isolated from a soil streptomycete, although their clinical use was limited because of toxicity.8 Screening of soil isolates subsequently was pursued formally in the search for anticancer compounds derived from antimicrobial antibiotics. In the ensuing 2 decades, testing of natural products from plant and marine sources led to the discovery of agents such as the vinca alkaloids, the taxanes, and camptothecins.9 Then, it was demonstrated that cisplatin had antitumor activity in the 1960s, a discovery that arose from a series of experiments prompted by a serendipitous finding that electrical fields were inhibiting the cell division of Escherichia coli.10

In conjunction with drug discovery were investigations on tumor growth kinetics, which constitute the theoretic framework of cancer chemotherapy. Skipper et al demonstrated the exponential (logarithmic) growth of tumor cells and, conversely, proposed the log kill hypothesis that drug therapy will eliminate a constant proportion rather than a constant number of tumor cells regardless of the initial size of the tumor.11 If sufficient drug is administered, then a cure is attained when less than 1 tumor cell remains. Those investigators also provided the hypothetical groundwork for high-dose chemotherapy (with bone marrow/stem cell transplantation as rescue) in the cure of refractory malignancies based on the dose-response antitumor effect of conventional cytotoxic chemotherapeutic drugs. Indeed, clinical aspects underlying the principles of chemotherapy, such as the intermittent use of drugs, the use of agents with nonoverlapping toxicities, and the practice of administering the maximum tolerated dose of a specific anticancer drug, rest on this dose-response relation.

Although the work is seminal, the log kill model described by Skipper et al is applicable largely only to malignancies in which the proliferating fraction of cells constitutes the bulk or entirety of the malignant process, such as leukemias. For solid tumors, experimental data support the Gompertzian model of sigmoidal tumor growth. This relies on the Norton-Simon hypothesis that log kill will be greatest when tumor burden is at a minimum, as in micrometastatic states.12 However, because tumor growth likewise is fastest in this setting, tumor eradication is difficult and is achieved best by minimizing the time interval between drug therapies, thus giving rise to dose-dense regimens. In the North American Intergroup factorial trial design (Cancer and Leukemia Group B trial 9741), the concept of dose-dense adjuvant chemotherapy was tested in patients with lymph node-positive breast cancer. In keeping with the prediction based on the Norton-Simon hypothesis, the dose-dense schedule proved superior to the standard schedule.13

Another core concept is the importance of combination chemotherapy using agents with different mechanisms of action and resistance. This is based in part on the hypothesis of Goldie and Coldman that drug resistance can arise from spontaneous mutations and will occur inevitably with cell proliferation and that these mutations occur at a rate of 1 of 105 cells. If 1 gram of tumor contains 109 cells, then 104 clones that are resistant to any given drug will be present. However, resistance to 2 drugs would be observed in fewer than 1 cell in 1010 (10 g) tumor.14 Combination chemotherapy became an established practice when more patients were cured successfully of lymphomas and leukemias using regimens that consisted of multiple drugs, whereas most remissions previously were short-lived after single-agent therapies and were limited to few tumors (eg, choriocarcinomas, Burkitt lymphoma). Conversely, the use of alternating regimens in theory would be a superior strategy over a sequential use of different regimens based on this hypothesis if all potential combination of drugs were used early upfront to minimize the emergence of drug resistance. However, the converse was demonstrated in the study reported by Bonadonna et al comparing alternating and sequential dose schedules of adjuvant cyclophosphamide, methotrexate, and fluorouracil (CMF) given simultaneously in combination with doxorubicin (A) for patients with high-risk stage II breast cancer. A sequential regimen consisting of 4 courses of doxorubicin followed by 8 courses of CMF (A × 4→CMF × 4) resulted in better long-term disease-free and overall survival (OS) compared with an alternating regimen of 2 courses of CMF followed by 1 course of doxorubicin, repeated for 4 cycles ([CMF2→A] × 4).15, 16 It is interesting to note that, in some fashion, that the sequential approach produced superior results is predicted by the Norton-Simon hypothesis, because 4 consecutive administrations of doxorubicin, which is believed to be more effective than CMF, take less time and thus are more ‘dose-dense’ than an alternating approach.

The 21st Century

Signaling networks as targets

The past decade marked the beginning of a transitional era in cancer therapy from DNA/mitosis-based mechanisms to strategies manipulating the dysregulated molecular pathways that characterize the malignant phenotype. Indeed, this ‘targeted’ approach is not an entirely novel concept and has its roots in the hormonal interventions for the treatment of breast, prostate, and neuroendocrine cancers. Moreover, the seminal concept of the contribution of the tumor stroma and microenvironment to carcinogenesis first proposed in the late 19th century finally has gained widespread acceptance; thus, therapeutic interventions to manipulate processes such as angiogenesis, cell adhesion, and epithelial to mesenchymal transition are being studied.

Because there exist complex, nonlinear interactions with multiple redundant, hierarchical intersections and feedback loops occurring among the key targets in signaling pathways, targeted agents commonly affect multiple aspects of cancer biology simultaneously. Thus, it is not surprising to find that the drugs that modulate signal transduction affect not only the primary biologic pathway of their respective targets, whether those targets include the induction of cell cycle arrest, antiangiogenesis, or apoptosis, but they often yield a combination of effects. A comprehensive discussion of each relevant signaling pathway is beyond the scope of this review; therefore, we refer interested readers to reviews detailing the basic biology of the various signaling networks. Figure 1 is a simplified diagram encapsulating the key ‘nodes’ in the signaling networks against which several drugs already are in clinical use or in late stages of clinical development. Table 1 summarizes the findings in the pivotal trials of the novel agents discussed in this review.

Figure 1.

This diagram encapsulates the key ‘nodes’ in the signaling networks against which several drugs already are in clinical use or in late stages of clinical development. VEGF indicates vascular endothelial growth factor; EGFR, epidermal growth factor receptor; VEGFR, vascular endothelial growth factor receptor; PI3K, phosphoinositide 3-kinase; mTOR, mammalian target of rapamycin; Akt, protein kinase B; Abl, proto-oncogene tyrosine-protein kinase ABL; Bad, Bcl-2 antagonist of cell death; TSC1, TSC2, tuberous sclerosis 1 and 2; p70S6K, ribosomal protein S6 kinase 1; STAT, signal transducer and activator of transcription; 4EBP1, eukaryotic initiation factor 4E-binding protein 1; elF4E, eukaryotic initiation factor 4E; IKK, IκB kinase; IκBα, inhibitor of Nf-κB; NF-κB, nuclear factor κB; MEK1/2, mitogen-activated protein kinases 1 and 2; ERK1/2, extracellular signal regulated kinases 1 and 2; PDGF, platelet-derived growth factor; FGF, fibroblast growth factor; HDAC, histone deacetylase; TGF, transforming growth factor; HGF, hepatocyte growth factor; DNMT, DNA methyltransferase.

Table 1. Randomized Trials of Selected Novel Agents in the Treatment of Various Malignancies
Drug/Target and Cancer TypePhase/No.Comparison ArmsPrimary Endpoint(s)Secondary Endpoint(s)Cross-over, %Reference
  1. EGFR indicates epidermal growth factor receptor; SCCHN, squamous cell carcinoma of the head and neck; RT, radiotherapy; OS, overall survival; PFS, progression-free survival; RR, relative risk; 5-FU, 5-fluorouracil; CRC, colorectal carcinoma; IFL, irinotecan, 5-FU, plus leucovorin; TTP, time to progression; ORR; overall response rate; HR; hazard ratio; TK, tyrosine kinase; NSCLC, nonsmall cell lung carcinoma; adeno, adenocarcinoma; TTF; time to treatment failure; PC, paclitaxel plus carboplatin; NS, nonsignificant; GC; gemcitabine plus cisplatin; AC, doxorubicin plus cyclophosphamide; RCC, renal cell carcinoma; EGFR+, EGFR-positive; VEGF, vascular endothelial growth factor; IFN-α, interferon α; RTK, receptor tyrosine kinase; FOLFOX/FOLFOX4, oxaliplatin, leucovorin, plus 5-FU; bFOL, oxaliplatin, weekly 5-FU, plus low-dose leucovorin; CapOx, capecitabine plus oxaliplatin; G, gemcitabine; HCC, hepatocellular carcinoma; PDGF, platelet-derived growth factor; GIST, gastrointestinal stromal tumors; mTOR, mammalian target of rapamycin; DNMT, DNA methyltransferases; MDS, myelodysplastic syndromes; CMML, chronic myelomonocytic leukemia; AML, acute myeloid leukemia; CML, chronic myeloid leukemia; DP, disease progression; CCyR, complete cytogenetic response; MCyR, major cytogenetic response; CLL, chronic lymphocytic leukemia; FC, fludarabine plus cyclophosphamide; FT, farnesyl transferase; ACT, doxorubicin, cyclophosphamide, plus trastuzumab; DFS, disease-free survival; FEC, 5-FU, epirubicin, plus cyclophosphamide.

Treatment for locally advanced or metastatic disease
 Cetuximab/EGFR      
  SCCHNPhase 3First-lineLocoregional controlOS Bonner 200645
 213RT14.9 mo29.3 mo  
 211RT plus cetuximab24.4 mo49 mo  
  P.005.03  
 Phase 3First-linePFSOS/RR22Burtness 200542
 80Cisplatin2.7 mo8 mo/26%  
 57Cisplatin plus weekly cetuximab4.2 mo9.2 mo/10%  
  P.78.21/.03  
 Phase 3First-lineOS  Vermorken 200743
 220Platinum/5-FU7.4 mo   
 222Platinum/5-FU plus cetuximab10.1 mo   
  P<.0362   
  CRCPhase 2Irinotecan failureRROS/TTP50Cunningham 200440
 111Cetuximab10.8%6.9 mo/4.1 mo  
 218Cetuximab plus IFL22.9%8.6 mo/1.5 mo  
  P.007.48/.001  
 Phase 3EGFR-expressingOS6-mo PFS/RR7Jonker 2007225
 285Best supportive care6.1 mo3%/0%  
 287Best supportive care plus cetuximab4.6 mo15%/8%  
  P.005<.001/<.001  
  PancreasPhase 3First-lineOSPFS/ORR Philip 200747
 369Gemcitabine6 mo3 mo/14%  
 366Gemctabine plus cetuximab6.5 mo3.5 mo/12%  
  P.14.58  
 Panitumumab/EGFR      
  CRCPhase 3Previously treatedPFSOS/RR76Van Cutsem 200748
 232Best supportive care7.3 wk—/0%  
 231Best supportive care plus panitumumab8 wk1.0/10%  
  P<.0001.81/<.0001  
 Gefitinib/EGFR-TK      
  NSCLCPhase 3Previously treatedSurvival:All/adenoTTF/RR3Thatcher 200550
 563Placebo5.1 mo/5.4 mo2.6 mo/1.3%  
 1129Gefitinib5.6 mo/6.3 mo3 mo/8%  
  P.087/.089.0006/<.0001  
 Phase 3First-lineOSTTP/RR Herbst 2004226
 345PC plus placebo9.9 mo5 mo/28.7%  
 345PC plus gefitinib 250 mg9.8 mo5.3 mo/30.4%  
 347PC plus gefitinib 500 mg8.7 mo4.6 mo/30%  
  P.64.0562/NS  
 Phase 3First-lineOSTTP/RR Giaccone 2004227
 363GC plus placebo10.9 mo6 mo/47.2%  
 365GC plus gefitinib 250 mg9.9 mo5.8 mo/51.2%  
 365GC plus gefitinib 500 mg9.9 mo5.5 mo/50.3%  
  P.46.76/NS  
 Erlotinib/EGFR-TK      
  NSCLCPhase 3Previously treatedOSPFS/RR Shepherd 200551
 243Placebo4.7 mo1.8 mo/<1%  
 488Erlotinib6.7 mo2.2 mo/8.9%  
  P<.001<.001/<.001  
 Phase 3First-lineOSTTP/RR Herbst 2005228
 540PC plus placebo10.5 mo4.9 mo/19.3%  
 539PC plus erlotinib10.6 mo5.1 mo/21.5%  
  P.95.36/.36  
 Phase 3First-lineOSTTP Gatzemeier 2007229
 579GC plus placebo44.1 wk24.6 wk  
 580GC plus erlotinib43 wk23.7 wk  
  P.49.74  
  PancreasPhase 3First-lineOSPFS/RR Moore 2007230
 284Gemcitabine plus placebo5.91 mo3.55 mo/8%  
 285Gemcitabine plus erlotinib6.24 mo3.75 mo/8.6%  
  P.038.004/—  
 Trastuzumab/HER2      
  BreastPhase 3First-lineTTPOS/RR:66Slamon 200156
 234Chemotherapy (AC or paclitaxel)4.6 mo20.3 mo/32%  
 235Chemotherapy plus trastuzumab7.4 mo25.1 mo/50%  
  P<.001.046/<.001  
 Phase 2First-lineRROS/TTP57Marty 2005231
 94Docetaxel34%22.7 mo/6.1 mo  
 92Docetaxel plus trastuzumab1%31.2 mo/11.7 mo  
  P.0002.0325/.0001  
 Lapatinib/EGFR/HER2 TK      
  BreastPhase 3Previously treatedTTPRRYesGeyer 200666
 161Capecitabine plus placebo8.4 mo14%  
 163Capecitabine plus lapatinib4.4 mo22%  
  P<.001.09  
  RCCPhase 3Cytokine failureTTP (all/EGFR+)OS (all/EGFR+) Ravaud 200670
 207Hormone therapy15.4 wk/10.9 wk43.1 wk/37.9 wk  
 209Lapatinib15.3 wk/15.1 wk46.9 wk/46 wk  
  P.60/.06.21/.02  
 Bevacizumab/VEGF      
  BreastPhase 3Previously treatedPFSOS/RR Miller 2005232
 230Capecitabine4.17 mo14.5 mo/9.1%  
 232Capecitabine plus bevacizumab4.86 mo15.1 mo/19.8%  
  P.98—/.001  
 Phase 3First-linePFSOS/RR Miller 2007218
 326Paclitaxel5.9 mo25.2 mo/21.2%  
 347Paclitaxel plus bevacizumab11.8 mo26.7 mo/36.9%  
  P<.001.16/<.001  
  CRCPhase 3First-lineOSPFS/RR Hurwitz 200477
 411IFL plus placebo15.6 mo6.2 mo/34.8%  
 402IFL plus bevacizumab20.3 mo10.6 mo/44.8%  
  P.001<.001/.004  
 Phase 3First-lineTTP18-mo Survival/RR Hochster 200678
 49FOLFOX8.7 mo53%/41%  
 71FOLFOX plus bevacizumab9.9 mo63%/52%  
 50bFOL6.9 mo50%/20%  
 70bFOL plus bevacizumab8.3 mo63%/39%  
 48CapeOx5.9 mo49%/27%  
 72CapeOx plus bevacizumab10.3 mo68%/46%  
 Phase 3Irinotecan failureOSPFS/RR Giantonio 200779
 291A) FOLFOX410.8 mo4.7 mo/8.6%  
 286B) FOLFOX4 plus bevacizumab12.9 mo7.3 mo/22.7%  
 243C) bevacizumab10.2 mo2.7 mo/3.3%  
  P.0011(A vs B)<.0001 (A vs B)  
  NSCLCPhase 3First-lineOSPFS/RR Sandler 200680
 444PC10.3 mo4.5 mo/15%  
 434PC plus bevacizumab12.3 mo6.2 mo/35%  
  P.003<.001/<.001  
 Phase 3First lineOSRR/PFSYesEscudier 200781
  RCC322IFN-α plus placebo19.8 mo13%/5.4 mo  
 327IFN- α plus bevacizumabNot reached31%/10.2 mo  
  P.067<.0001/.0001  
  PancreasPhase 3First-lineOSPFS/RR Kindler 200785
 300G plus placebo6.0 mo4.3 mo/11.3%  
 302G plus bevacizumab5.7 mo4.8 mo/13.1%  
  P.4.99  
 Vatalanib/VEGF RTK      
  CRCPhase 3First-linePFSRR Hecht 2005233
 583FOLFOX plus placebo7.6 mo42%  
 585FOLFOX plus vatalanib7.7 mo46%  
  P.118   
 Phase 3Irinotecan failuresOSPFS/RR Koehne 2006234
 429FOLFOX plus placebo11.8 mo4.1 mo/17.5%  
 426FOLFOX plus vatalanib12.1 mo5.5 mo/18.5%  
  P.511.026/NS  
 Sorafenib/VEGF RTK      
  HCCPhase 3First-lineOSTTP Llovet 2007235
 303Placebo7.9 mo2.8 mo  
 299Sorafenib10.7 mo5.5 mo  
  P.00058.000007  
  RCCPhase 3Previously treatedOSPFS/RR Escudier 200788
 452Placebo15.9 mo2.8 mo/2%48 
 451Sorafenib19.3 mo5.5 mo/11%  
  P.02<.001/<.001  
 Sunitinib/VEGF RTK      
  RCCPhase 3First-linePFSRRYesMotzer 200787
 375IFN-α5 mo6%  
 375Sunitinib11 mo31%  
  P<.001<.001  
 Sunitinib/c-kit/PDGF RTK      
  GISTPhase 3Imatinib resistance or intoleranceTTPOS/RR Demetri 200691
 105Placebo6.4 wk/56.9%—/0%56 
 207Sunitinib27.3 wk/79.4%—/7%  
  P<.0001.007 [HR, 0.49]  
 Temsirolimus/mTOR      
  RCCPhase 3First-line, high-riskOSPFS/RR Hudes 2007140
 207IFN7.3 mo1.9 mo/4.8%  
 209Temsirolimus10.9 mo (P<.008)3.8 mo/8.6%  
 210IFN plus temsirolimus8.4 mo (P=.70)3.7 mo/8.1%  
 Azacytidine/DNMT      
  MDS, CMMLPhase 3 RR/time to AML or deathOS Silverman 2002236
 92Best supportive care0%/12 mo14 mo53 
 99Azacytidine23%/21 mo20 mo  
  P<.0001/.007.1  
 Decitabine      
 MDS, CMMLPhase 3 RR/time to AML or deathOSYesKantarjian 2006237
 81Best supportive care0%/7.8 mo14.9 mo  
 89Decitabine17%/12.1 mo14 mo  
  P<.001/<.16.636  
 Bortezomib/Proteasome      
  Multiple myelomaPhase 3 TTPOS (1-y)/RR44Richardson 2005238
 336Dexamethasone3.49 mo66%/18%  
 333Bortezomib6.22 mo80%/38%  
  P<.001.003/<.001  
 Imatinib/Bcr/abl      
  CMLPhase 3First-line18-mo DP18-mo CCyR O'Brien 2003239
 553IFN-α plus cytarabine73.5%14.5%57.5 
 553Imatinib92.1%76.2%2 
  P <.001  
 Imatinib/c-kit/PDGF RTK      
  GISTPhase 3First-linePFSOSYesVerweij 2004240
 345Imatinib 400 mg 2×d18 mo55 mo38.6 
 349Imatinib 400 mg 1×d20 mo51 mo  
  P.13.83  
 Dasatinib/Src/abl kinase      
  CMLPhase 2Imatinib-resistantMCyR at wk 12CCyR at wk 12/TTF Kantarjian 2007241
 101Dasatinib36%22%/Not reached15 
 49High-dose imatinib29%4%/3.5 mo80 
  P.40.041/.001  
 Oblimersen/Bcl-2      
  CLLPhase 3Previously treatedRRTTP/OS O'Brien 2007173
 121FC7%9 mo/32.9 mo  
 120FC plus oblimersen17%6.1 mo/33.8 mo  
  P−.025.83  
 Tipifarnib/FT      
  CRCPhase 3Previously treatedOSPFS/RR Rao 2004116
 133Best supportive care185 d80 d/0%  
 235Best supportive care plus tipifarnib174 d81 d/1%  
  P.376.088  
  PancreasPhase 3First lineOSPFS/RR Van Cutsem 2004117
 347Gemcitabine plus placebo182 d109 d/8%  
 341Gemcitabine plus tipifarnib193 d112 d/6%  
  P.75.72  
Adjuvant setting      
 Trastuzumab/HER2      
  BreastPhase 3 DFS (4-y/3 y)OS (4-y/3-y) Romond 200557
 1679ACT chemotherapy67.1%/75.4%86.6%/91.7%  
 1672ACT chemotherapy plus trastuzumab (1 y)85.3%/87.1%91.4%/94.3%  
  P<.0001.015  
 Phase 3 DFS (2-y)OS (2-y) Piccart-Gebhart 200558
 1693Observation77.4%95.1%  
 1694Trastuzumab (1 yr)85.8%96%  
  P<.0001NS  
 Phase 3 DFS (2-y/3-y)*OS (2-y/3-y)* Joensuu 200659
 116Docetaxel vs vinorelbine-FEC83.6%/77.6%95.7%/89.7%  
 1169 Weekly trastuzumab plus docetaxel vs vinorelbine)-FEC91.3%/89.3%97.4%/96.3%  
  P.01.07  

Approaches to Target Modulation

In discussing novel agents that have been introduced into the clinic over the last decade, it is important to distinguish drug targets from the methods and technologies that have been used to inhibit these targets. There are 3 broad categories of technologies used in the rational design of new drugs that modulate target protein interactions in tumors: 1) immunologic, 2) nucleic acids, 3) and small molecules. Table 2 lists the context-dependent advantages and disadvantages with the clinical application of each approach. The immunologic approach currently is dominated by the use of monoclonal antibodies (MoAbs), typically to disrupt ligand-receptor interactions. A major obstacle encountered in the early phases of development included the development of neutralizing antibodies and hypersensitivity reactions because of the murine derivation (-momab) of the initial compounds. These problems essentially are circumvented with techniques to reduce immunogenicity through chimeric (-ximab), humanized (-zumab), or fully human MoAbs (-mumab).

Table 2. Advantages and Disadvantages With Various Approaches in Target Modulation
ApproachAdvantagesDisadvantages
  1. BBB indicates blood-brain barrier; ADCC, antibody-dependent cell-mediated cytotoxicity; CDC, complement-mediated cytotoxicity.

Monoclonal antibodiesTarget specificityLarge molecular weight (poor tissue penetration, eg, BBB)
 High binding affinityComplex structure
 Induction of ADCC, CDC, apoptosisLong half-life (toxicities)
 Immunoconjugation (eg, radionuclides, toxins) for diagnostic and therapeutic purposesImmunogenicity
 Target receptor internalization/down-regulationHypersensitivity reactions
 Long half-life (reduced frequency of dosing)Requires parenteral administration
  Applicable only to extracellularly expressed domains of target proteins
Small moleculesPleiotropic targets (heterogeneity of tumors and signal pathway crosstalk)Schedule of administration (increased frequency)
 Tissue penetration and deliveryPleiotropic targets (more side effects)
 Availability of oral formulation 
Nucleic acidTarget specificityProinflammatory effects (sequence-dependent)
 High binding affinityOff-target effects (cross-hybridization)
 Stoichiometric efficiencySchedule of administration (increased frequency)
 Immune stimulation (sequence-dependent)Requires special delivery systems (susceptibility to degradation, inability to cross cellular membranes)
 Escort aptamers for delivering radionuclides, toxins, etc 

Surface-expressed tumor-associated antigens have been exploited therapeutically using MoAb strategy. However, it is important to recognize that many of the purported tumor-derived proteins are in fact lineage-specific differentiation antigens that are observed in normal tissues.17 Therefore, targeting such markers is complicated by effects on tumor cells as well as normal cells. Moreover, tumor cells are predisposed toward dedifferentiation and thus, in fact, may express less surface receptors, such as the expression of CD20 in B cells.18 Nonetheless, this approach has been successful in the treatment of hematopoietic tumors, as exemplified by rituximab, a chimeric monoclonal anti-CD20 immunoglobulin G1 (IgG1) used in the treatment of B-lymphoproliferative malignancies. Other examples are listed in Table 3.

Table 3. Monoclonal Antibodies in Clinical Use or Development
DrugTargetDisorder
  1. CD indicates cluster of differentiation; CLL, chronic lymphocytic leukemia; Calicheamicin; antitumor antibiotic isolated from soil bacteria Micromonospora; ALCL, anaplastic large cell lymphoma; MM, multiple myeloma; SCLC, small cell lung cancer; HGF, hormone growth factor; VEGF, vascular endothelial growth factor; CRC, colorectal carcinoma; NSCLC, nonsmall cell lung carcinoma; EGFR, epidermal growth factor receptor; CEA, carcinoembryonic antigen; HGS, Human Genome Sciences; ETR1, ethylene response 1; TRAIL-R1, tumor necrosis factor-related apoptosis-inducing ligand receptor 1; NHL, non-Hodgkin lymphoma; TR2J, TRAIL-2R monoclonal antibody; IMMU, Immunomedics, Inc.; AFP, α-fetoprotein; gastrointestinal; CTLA4, cytotoxic T lymphocyte-associated antigen 4; RCC, renal cell carcinoma; CanAg, cancer antigen; Maytansinoid. fungal toxin.

SiplizumabCD2T-cell lymphoma/leukemia
ZanolimumabCD4Cutaneous T-cell lymphoma
Rituximab, ofatumumab (HuMax-CD20); radioimmunoconjugates (Y90-ibritumomab tiuxetan, 131I-tositomumab)CD20Malignant B-cell disorders
GaliximabCD80Malignant B-cell disorders (eg, recurrent disease previously treated with rituximab)
EpratuzumabCD22Malignant B-cell disorders
LumiliximabCD23CLL
Gemtuzumab ozogamicin (Calicheamicin conjugated)CD33Myeloid leukemias
AlemtuzumabCD52T and B cell
MDX-060CD30Hodgkin, ALCL
SGN-40CD40Malignant B-cell disorders, MM
HuMax-CD38CD38Myeloma
HuN901-DM1 (Maytansinoid conjugated)CD56Myeloma, SCLC
AMG 102HGFMultiple tumor types
BevacizumabVEGFCRC, NSCLC
Cetuximab, Panitumumab (ABX-EGF), zalutumumab (HuMax-EGFR), matuzumab (EMD7200)EGFRNSCLC, head and neck, CRC
Trastuzumab, pertuzumabHER2Breast
Y-90 labetuzumabCEACEA-expressing tumors
MapatumumabTRAIL-R1NHL
HGS-ETR1TRAIL-R1Multiple tumor types
HGS-ETR2, HGS-TR2JTRAIL-R2Multiple tumor types
Integrins Melanoma, pancreas, renal cell
 Volociximab (M200)Integrin α5β1 
 AbegrinIntegrin αvβ3 
 CNTO 95Integrins αvβ3 and αvβ5 
HPAM4 (IMMU 107)MUC1Multiple tumor types; pancreas
Y-90 hAFP (IMMU 105) (Ytrrium 90-labeled)AFPHepatocellular, germ cell
MDX-070, MLN2704Prostate-specific membrane antigenProstate
RAV12N-linked carbohydrate antigenGI adenocarcinoma (gastric, stomach, pancreas)
Ipilimumab (MDX-101), TicilimumabCTLA4Melanoma, RCC
Cantuzumab mertansine (Maytansinoid conjugated)CanAgCRC, pancreatic, GI, NSCLC
Bivatuzumab mertansine (Maytansinoid conjugated)CD44v6Head and neck, breast

The high degree of target specificity of antibodies counterbalances inherent undesirable pharmaceutical properties, such as structural complexity, large molecular weight, physiologic barriers to intratumoral uptake, and heterogeneous antibody distribution within tumors, enough to make them one of the success stories in novel drug development for solid tumors, as exemplified by cetuximab and trastuzumab.19 Moreover, antibody-dependent cell-mediated cytotoxicity (ADCC), complement-mediated cytotoxicity, and induction of apoptosis are nontarget mechanisms that can contribute to the antitumor efficacy of MoAbs.20, 21 In addition, MoAbs can be labeled readily or conjugated with other biologic agents for diagnostic and therapeutic purposes. However, because of the challenges associated with achieving satisfactory tissue distribution and intratumoral concentrations, MoAb strategies most likely will not be as effective against bulky lesions, as demonstrated by the clinical experience with alemtuzumab in chronic lymphocytic leukemia (CLL), in which patients with leukemic-phase disease respond better than patients with bulky adenopathy.22, 23 This also is applicable to radioimmunotherapy, although the ability to effect additional nontarget events, such as the radiobiologic bystander effect, confers an additional advantage against tumor cells that otherwise would have escaped immunotargeting alone.

Antisense oligonucleotides (ASO) and aptamers constitute the major technologies in cancer therapeutics that use the nucleic acid approach. An ASO is a single-stranded, chemically modified, DNA-like molecule 17 to 22 nucleotides in length that can selectively inhibit posttranscriptional gene expression through complementary nucleic acid hybridization, thus sterically hindering the translation of a specific gene messenger RNA (mRNA) and/or, through recruitment of RNase H, to cleave the target mRNA.24 Subcategories of the nucleic acid technology include the catalytic sequence-specific RNA and DNA enzymes (ribozymes and DNAzymes, respectively), which cleave target RNA substrates, and short-interefering RNA molecules, which degrade target mRNA through a multicomponent enzyme complex.25 Because multiple copies of a protein are produced by each mRNA molecule, targeting mRNA rather than the protein itself is potentially a more efficient approach to modulate protein function by altering its levels, particularly of targets that cannot be manipulated directly by MoAbs or small molecule inhibitors, such as transcription factors. However, there remain several technical obstacles to the successful clinical application of this technology in terms of sequence design and delivery (eg, a platform providing efficiency of cell uptake and metabolic stability against degradation by ubiquitous nucleases), including the stimulation of nontarget effects, perhaps the most significant of which is the unintended stimulation of proinflammatory and immune responses presumably through Toll-like receptor activation,26 although this last finding has opened up new avenues for modulation of the immune system. More recently, despite the theoretical advantages of RNA interference (RNAi) technology, reports of fatal hepatotoxicity in mice have led to safety concerns for therapeutic applications of RNAi.27 A lesson learned from this as well as from the disappointing clinical results using antisense technologies against targets that otherwise have demonstrated their relevance using other approaches, such as the vascular endothelial growth factor receptor (VEGFR) mRNA ribozyme Angiozyme, is that clinical failure of a new drug does not necessarily mean that the target is not ideal for cancer therapeutics.

Whereas ASO and RNAi strategies alter target protein levels, aptamers, which are synthetic, short nucleic acid ligands made from RNA and DNA generated by in vitro systematic evolution of ligands by exponential enrichment technology that can fold into stable, 3-dimensional structures, possess binding affinities and specificities that are comparable to or even better than MoAbs28 and, thus, may interact at the ligand-receptor interface as well. Similarly, aptamers may be engineered to act as ‘escorts’ delivering radionuclides, etc, for diagnostic and therapeutic purposes.29 Moreover, the small size of aptamers facilitates tissue penetration. On the basis of lessons learned from ASO, most aptamers currently possess structural features to enhance nuclease resistance. Another important feature is that they essentially are nonimmunogenic even when administered in excess of therapeutic doses. Nevertheless, for intracellular targets, effective intracellular delivery remains a significant challenge in common with other nucleic acid technologies.

The clinical success of small molecule inhibitors parallels that of MoAbs in cancer therapy. The ability to modulate intracellular targets confers a distinct advantage over MoAbs, for instance, in situations in which the small molecule inhibitors can block the catalytic activity of mutant receptor kinases that lack an extracellular domain and are active constitutively in a ligand-independent fashion, eg, the epidermal growth factor receptor (EGFR) mutant EGFRvIII in glioblastomas.30, 31 These versatile molecules are well suited to alter target intracellular or extracellular proteins either qualitatively or quantitatively. Target protein activity can be affected either directly, such as inhibitors of matrix metalloproteinases, or indirectly by affecting posttranslational modification of the target protein, such as farnesylation of ras. Conversely, target protein levels may be altered by several means, such as through epigenetic regulation (eg, inhibition of histone deacetylases [HDACs]) or through degradation pathways either directly by modulating proteasomal activity (eg, bortezomib) or indirectly through interference with chaperone functions (eg, 17-allylamino demethoxygeldanamlycin) of heat shock protein 90 (HSP90), which then lead to degradation of target HSP90 client proteins by the ubiquitin-dependent proteasome pathway.

Imatinib was the prototype small molecule inhibitor of the tyrosine kinase (TK) domains of c-kit, platelet-derived growth factor (PDFR) receptor (PDGFR), and c-abl. Druggable molecular targets that are amenable to modulation by drugs (as defined by the Lipinski rule of 5) are gene products largely dominated by protein kinases, and 518 of those kinases have been described to date in the human kinome.32 Although the concept of ‘druggability’ is distinct and separate from the validity of a chosen drug target, both are necessary considerations in drug development. Because protein kinases are involved in a variety of intracellular signaling processes crucial for biologic functions, such as apoptosis, angiogenesis, and cell cycle regulation, enthusiasm continues for the development of protein kinase inhibitors as cancer therapy.

Although initial designs for small molecules were focused on achieving high target specificity based on the rationale of minimizing off-target side effects, subsequent experience revealed that ‘promiscuous’ drugs, such as adenosine triphosphate (ATP)-mimetic drugs, with pleiotropic targets arising from close homology between related protein kinases, in fact, possess a favorable efficacy profile over a wider spectrum of malignant disorders. This is not entirely surprising because of the well known, inherent redundancy and overlap in biologic response pathways and because the ‘mono-oncogenic’ model of carcinogenesis is restricted to only a few examples. This pleiotropism in fact has led to several serendipitous clinical findings, such as the case of sorafenib, which ostensibly was developed as a C-raf serine threonine kinase inhibitor, but its antitumor activity in clinical testing subsequently was determined to be more consistent with its activity as an inhibitor of the VEGFR TK. The ability to inhibit multiple targets also confers an advantage over MoAbs, eg, when receptor heterodimerization occurs as exemplified by ErbB receptors. Because there are nonoverlapping features between the use of MoAbs and small molecule inhibitors, the combination of both against a common drug target potentially may exhibit synergistic activity.33

Ligand-Receptor Interface

EGFR-mediated pathway

Mirroring the drug developmental path after the discovery of the Philadelphia chromosome in chronic myeloid leukemia (CML), the discovery by Cohen et al in the 1950s of epidermal growth factor (EGF) and subsequently, the establishment of the pathogenic role in neoplastic diseases of the EGF signaling pathway catalyzed efforts to develop EGFR-directed therapies. Because it is the most studied, we select the EGFR TK system as a prototype to demonstrate the highly regulated nature of signaling networks. It is characterized by an hourglass or bow-tie architecture,34 whereby multiple inputs (13 growth factor ligands with conserved EGF domains) received by a permutation of dimeric receptors arising from 4 distinctive receptors of the erBb family (EGFR, HER‒2, HER‒3, and HER‒4), are funneled through second messengers and transcription factors that amplify the signals through multiple pathways, such as Ras-mitogen-activated protein kinase, phosphoinositide 3-kinase (PI3K)-Akt-mammalian target of rapamycin (mTOR), toward a wide base of myriad biologic processes. A key feature integral to the understanding of the therapeutic manipulation of the EGFR pathway is that EGFR, in its typical unligated form, is in an autoinhibited ‘closed’ conformation.35, 36 Growth factors function to unfurl the receptor into an ‘open’ conformation that can dimerize through the exposed dimerization arm at domain II of the extracellular region.37 Another important finding to recognize is that HER‒2, despite its lack of ligand-binding activity and autonomous function, is the preferred heterodimeric partner of the 3 other receptors.38 This is because of the structural nature of HER‒2, which constitutively assumes a receptive, ‘open’ conformational structure with an extended domain II arm that mimics the ligand-bound state of EGFR. These heterodimers induce greater mitogenic effects because of higher ligand affinity and evasion of receptor degradation, in contrast to ligand-induced receptor endocytosis with subsequent endosomal degradation of EGFR dimers.39

Cetuximab is a chimeric MoAb that avidly competes with and sterically blocks access of growth factors to a key ligand-binding region of EGFR. It is approved for treatment, either alone or in combination with irinotecan, in patients with irinotecan-refractory colorectal cancers (CRCs).40 In patients with recurrent or metastatic head and neck cancers, cetuximab is approved as a single agent after failure of prior platinum-based therapy.41 It increases first-line response rates and, in combination with cisplatin, survival was not improved; however, the study was not powered adequately, and crossover was allowed.42, 43 A larger study subsequently demonstrated a survival advantage43 On the basis of preclinical evidence that demonstrated synergistic activity with ionizing radiation and a subsequent phase 3 trial that demonstrated a survival benefit, currently, cetuximab also is approved in combination with radiotherapy as first-line therapy for patients with locoregionally advanced squamous cell carcinoma of the head and neck.44, 45

Cetuximab currently is undergoing further evaluation in combination with chemotherapy given in the adjuvant setting to patients with CRC. First-line combination with systemic chemotherapy and bevacizumab therapy also is being tested in ongoing randomized phase 3 trials. Because of favorable activity in early phase 2 studies in nonsmall cell lung cancer (NSCLC) and pancreatic cancer, several randomized phase 3 trials were conducted. An open-label phase 3 study that included >600 patients from the US and Canada tested the combination of cetuximab plus a taxane and carboplatin in the first-line treatment for metastatic NSCLC. This trial did not meet its primary endpoint of progression-free survival (PFS), as assessed by an independent radiology review committee (IRRC). However, key secondary endpoints of the study, including response rate as assessed by the IRRC and PFS as assessed by clinical investigators, were statistically significant and favored the cetuximab-containing arm.46 Similarly, analysis of the Southwest Oncology Group phase 3 randomized study that enrolled >700 patients demonstrated no significant difference in outcomes when cetuximab was added to gemcitabine compared with gemcitabine alone in the first-line treatment of advanced pancreatic cancer.47 Data from phase 3 trials investigating first-line treatment of cetuximab with cisplatin/vinorelbine in advanced EGFR-expressing NSCLC, as well as irinotecan and docetaxel with or without cetuximab in patients with metastatic pancreatic adenocarcinoma, are eagerly awaited.

Whereas nontarget immunologic effects (eg, ADCC) may be mediated by cetuximab because of its IgG1 isotype, the IgG2 isotype used in the fully humanized panitumumab precludes immunologic mechanisms contributing to its efficacy. It is more potent and does not require prophylaxis against hypersensitivity reaction; and, because of its longer half-life, it may be given on a biweekly schedule. It also has shown single-agent activity in metastatic CRC after failure of standard chemotherapy compared with best supportive care, although the crossover design likely accounted for the lack of an OS advantage.48 However, the phase 3 Panitumumab Advanced Colorectal Cancer Evaluation trial in first-line metastatic CRC testing the combination of oxaliplatin, fluorouracil, and leucovorin (FOLFOX) plus bevacizumab with or without panitumumab was discontinued in March 2007, because a preplanned interim efficacy analysis revealed PFS in favor of the control arm.49

Gefitinib is the first EGFR TK inhibitor (TKI) that has received accelerated approval from the U.S. Food and Drug Administration (FDA) based on encouraging phase 2 data results, with the stipulation that postmarketing trials be conducted, including the Iressa Survival Evaluation in Lung Cancer study, a randomized, placebo-controlled, phase 3 study that was designed to assess the survival advantage of gefitinib versus best supportive care in patients with advanced NSCLC who were refractory to or intolerant of their latest chemotherapy regimen.50 Preplanned subgroup analyses demonstrated heterogeneity in survival outcome, such that survival was better in the gefitinib group than in the placebo group among never-smokers and among patients of Asian origin in contrast to the similar outcomes between treatment and placebo among smokers or patients of non-Asian origin. These were similar to the findings with the phase 3 data from the National Cancer Institute of Canada Clinical Trials Group Study BR.21 using erlotinib, another oral EGFR TKI.51 Unlike erlotinib, however, primary survival endpoints were not met in the gefitinib trial. An explanation often repeated is that gefitinib dosing was suboptimal in contrast to the counterpart trial in erlotinib that used dosing at the maximum tolerated dose level. A phase 3 survival study (the Iressa NSCLC Trial Evaluating Response and Survival Against Taxotere or INTEREST study), however, recently demonstrated noninferiority of gefitinib compared with docetaxel with superior safety on the gefitinib arm.52, 53

Overexpression or gene amplification of Her-2/neu, which occurs in 20% to 30% of breast cancers, predicts for a poor clinical outcome and for resistance to chemotherapy and hormone therapy. Trastuzumab is an HER‒2 MoAb that exhibits single-agent activity, either as first-line54 or salvage therapy,55 and also has synergistic activity with chemotherapy in metastatic breast cancers that over express Her2/neu.56 More recently, enthusiasm for this drug was enhanced by the stunning survival outcomes confirmed by several phase 3 trials, B-31, N9831, and Herceptin Adjuvant (HERA), which incorporated trastuzumab maintenance for 1 to 2 years in the adjuvant treatment setting in breast cancer.57, 58 Even more compelling was the observation from the joint analysis that demonstrated the virtual elimination of the early peak in recurrences generally observed during the first 2 to 3 years after a breast cancer diagnosis.57 Although the risk of cardiotoxicity seems lower with the sequential administration of trastuzumab and chemotherapy, optimal timing of trastuzumab therapy (sequential vs concurrently with paclitaxel) has yet to be determined. An intriguing finding from the Finland Herceptin (FinHer) study, another randomized phase 3 trial that introduced a twist in the schedule of trastuzumab therapy, was that survival outcomes from the subset of patients with HER‒2/neu-overexpressing breast cancer were superimposable with those from the collective data reported by Romond et al despite a shorter course (9 weekly doses) of trastuzumab preceding administration of an anthracycline-containing regimen.57, 59 The early results from that study suggest that a shortened duration of trastuzumab therapy before anthracycline exposure may limit cardiotoxicity without compromising efficacy. Definitive answers regarding the timing and duration of trastuzumab therapy hopefully will be available with more mature data from these trials in the ensuing years.

However, not all Her2/neu-overexpressing breast cancers respond to trastuzumab, which most likely is a function of the weak inhibitory activity of trastuzumab against the formation of HER‒2 heterodimers, as in the kinase-defective HER‒3, which, in turn, activates the PI3K signaling pathway.60 Emerging crystallographic data revealed the structural mechanism that may explain this finding. Because trastuzumab binds to domain IV of the extracellular region, the dimerization domain II arm of Her2 is left exposed and poised for coupling.60, 61 Pertuzumab, conversely, binds to domain II and, thus, disrupts heterodimerization and subsequent signaling activity.60 Because trastuzumab and pertuzumab have nonoverlapping effects on receptor shedding and endocytosis, these 2 MoAbs in combination provide complementary action, as evident in the synergistic activity observed.62 Preliminary results from a phase 2 trial investigating this combination in 42 patients with HER‒2-positive metastatic breast cancer who progressed after trastuzumab therapy indicated promising results, including complete responses, which have led to an ongoing phase 3 trial of the combination in this patient population.63 Also of interest is that, although HER‒2 overexpression/gene amplification in ovarian/primary peritoneal cancer is low and trastuzumab has marginal value as monotherapy, early exploratory analysis of data from 130 patients enrolled in a randomized, double-blinded, placebo-controlled phase 2 trial of gemcitabine with or without pertuzumab for women with platinum-resistant ovarian, fallopian tube, or primary peritoneal cancer indicated improved overall PFS among the patients who received the combination.64 Although the data are immature, there are promising correlative data that may help identify the appropriate patient population for treatment, because patients with higher HER‒2 to HER‒3 gene expression ratios appeared to derive a significant PFS benefit from this combination therapy.

We can surmise from the discussion above of the EGFR family of receptors that simultaneous inhibition of EGFR and HER‒2 may be a suitable approach for drug therapy. Indeed, abundant preclinical evidence set the stage for the development of the dual TK inhibitor lapatinib (GW572016). Unlike erlotinib, which dissociates rapidly upon binding to the active conformation of EGFR, lapatinib has a slow dissociation rate after it is bound, and the lapatinib/EGFR complex is characterized as in an inactive-like conformation.65 Results from an international, randomized phase 3 study that compared capecitabine with or without lapatinib in patients who had disease progression after trastuzumab treatment were disclosed early after interim analyses showed that the primary time-to-progression (TTP) endpoint was exceeded favorably, as determined by independent reviewers. The median TTP was 8.4 months in the combination therapy group compared with 4.4 months in the monotherapy group.66 Because that study was halted early, and crossover was allowed subsequently, the actual survival benefit may have been obscured. Although there were no serious toxic or symptomatic cardiac events reported, definite conclusions regarding cardiotoxicity are premature at this juncture. An interesting observation, although limited by the small number of events, was that brain metastases developed in fewer women who received the lapatinib combination compared with women in the monotherapy group. A phase 2 study of lapatinib for brain metastases in patients with HER‒2-positive breast cancer, although it did not meet its primary endpoint of objective response (defined by Response Criteria in Solid Tumors), provided evidence that lapatinib induced a volumetric reduction of brain metastases.67 The latest results from the extension trial involving 49 patients in which capecitabine was added to lapatinib when brain metastases progressed on lapatinib monotherapy demonstrated that 37% of patients experienced a volumetric decrease ≥20%, and patients who received combination therapy experienced a volumetric reduction ≥50% in measurable brain metastases.68, 69 Lapatinib currently is being studied in several other randomized phase 3 breast cancer trials, including first-line therapy in combination with paclitaxel or antiangiogenic agents, such as pazopanib for metastatic disease; in the adjuvant setting with trastuzumab; and neoadjuvantly with chemotherapy or hormone therapy using letrozole. Conversely, there is a limited role for lapatinib in prostate cancer, gastric cancer, and renal cell cancer (RCC).70–72 Subset analyses from an RCC trial, however, demonstrated that patients who had tumors that overexpressed EGFR derived a benefit from lapatinib compared with hormone therapy. This finding will require further investigation.

VEGFR-mediated pathways

The tenets surrounding the role of the tumor microenvironment started from the ‘seed-and-soil’ hypothesis described by Paget in 1889 and are supported by evidence in recent times demonstrating the interdependent roles of cancer cells, neighboring stromal cells, tissue macrophages, and the intercellular matrix. The purification of VEGF-A a century later in 198973 and subsequent preclinical evidence that targeting this ligand74 can suppress angiogenesis and tumor growth in preclinical models led to the recognition of the crucial role played by underlying vasculature and angiogenesis in carcinogenesis. Similar to the EGF system, multiple signal inputs are mediated through 5 distinct receptors in the VEGFR system. Although VEGF165 is the isoform implicated most in angiogenesis and vascular permeability, other isoforms, such as VEGF121 and VEGF110, are potent permeability factors, and blocking these isoforms likely will contribute to greater efficacy. This is supported by the clinical findings that an Fab fragment that binds to all isoforms of VEGF-A potentially is more effective than pegaptanib, an aptamer against VEGF165, in age-related macular degeneration.75

Bevacizumab is a humanized MoAb against all isoforms of VEGF-A that is the first antiangiogenic agent to be approved for therapeutic use. Despite the concern that antiangiogenic agents will impede drug delivery by their very mechanism of action, in fact, preclinical evidence indicates that, with normalization of aberrant vessels that are often leaky, reduction in vascular permeability and intratumor pressure leads to improved drug delivery.76 It has shown improved survival in combination with standard therapy in patients with CRC, lung cancer, and (most recently) RCC.77–81 Early phase 2 studies in recurrent ovarian cancer, either as a single-agent or in combination with chemotherapy, have produced promising results tempered by a higher than expected incidence of bowel perforation.82–84 Nonetheless, the limited efficacy as monotherapy, the context-dependent clinical benefit with combination regimens (a survival benefit in either the first-line or second-line setting in CRC but a TTP benefit only in the first-line setting in breast cancer; no benefit in the first-line setting in pancreatic cancer85) the survival benefit in RCC observed in patients with favorable-risk and intermediate-risk features (Motzer criteria) raises more questions than answers. Multiple phase 3 trials incorporating bevacizumab in combination with various chemotherapy, antiestrogen, or novel agents (eg, imatinib) in the metastatic setting and in the adjuvant/neoadjuvant setting are ongoing or soon will be initiated.

Although the potential therapeutic use of anti-VEGF therapies against RCC first was demonstrated clinically using bevacizumab, based on the observation that mutations in the von Hippel-Lindau (VHL) gene that cause the over secretion of VEGF occur frequently in clear cell RCC, randomized phase 3 results demonstrating efficacy in RCC were reported first in multitargeted TKIs. Moreover, there is phase 2 clinical evidence that multitargeted agents may be effective in bevacizumab-resistant RCC.86 Sunitinib (SU11248), a compound that simultaneously targets several RTKs, including VEGFR, PDGFR, Kit, and FMS-like TK3 (FLT3), received approval for the treatment of advanced RCC after the confirmation of an improvement in PFS among patients with low- and intermediate-risk features (Motzer criteria) in a phase 3 study.87 Sorafenib (BAY43-9006), another multikinase inhibitor, was approved based on the clinical activity demonstrated in a phase 3 study in the treatment for RCC patients with low- and intermediate-risk features who received previous biologic therapy.88 It also was approved recently for the treatment of patients with unresectable hepatocellular carcinoma.235 Although it was developed originally as an inhibitor of the RAF serine/threonine kinases (RAF-1, wild-type BRAF, V600E BRAF mutant), subsequent studies revealed that it also inhibits VEGFR, PDGFR, Kit, and FLT3,89 which currently is believed to be the main mechanism by which sorafenib exerts its antitumor activity. It is also a potent inhibitor of the oncogenic RET kinase, and further studies may enlarge its repertoire of tumor activity to include RET-positive medullary thyroid tumors. Despite similarity in drug targets between sorafenib and sunitinib, the higher tumor response rate observed with sunitinib is attributed to its stronger binding affinity to the involved kinases compared with sorafenib.90 Moreover, sunitinib prolonged survival compared with placebo among patients with imatinib-resistant gastrointestinal stromal tumors (GISTs).91

A Combination of EGFR and VEGFR blockade has been investigated with variable results. Erlotinib does not improve the efficacy of bevacizumab in RCC,92 whereas the combination produced encouraging results in NSCLC.93 Whether TKIs with activity against both targets, such as ZD6474 (EGFR, VEGFR, RET) and AEE788 (EGFR, HER‒2, VEGFR2, c-Src) will exhibit a similar spectrum of clinical utility currently is being investigated.

Originally developed as a sedative, thalidomide had a tainted history in pharmacotherapy because of unexpected teratogenicity that arose from its use in pregnant women. Now, it is experiencing a ‘renaissance,’ because its antiangiogenic effect has been recognized as a valid platform for cancer therapy. Thalidomide down-regulates the expression of VEGF and basic fibroblast growth factor apart from suppressing the production of tumor necrosis factor-α (TNF-α) and modulating cytokine responses.94–96 However, its clinical benefit is limited by cumulative and dose-dependent toxicities of neuropathy, fatigue, constipation, and sedation that often preclude prolonged administration. Lenalidomide is an orally bioavailable thalidomide analog with 100-fold to 1000-fold greater potency than its parent compound in promoting T-cell activation, inhibiting neoangiogenesis, and suppressing the production of TNF-α and other inflammatory cytokines. It lacks the sedative properties of thalidomide but exhibits more myelosuppression. Lenalidomide has remarkable activity in the subtype of myelodysplastic syndrome (MDS) with an interstitial deletion of chromosome 5q.97 Its karyotype-dependent mechanism of action is supported by correlative biomarker studies, in which direct cytotoxicity predominates in del(5q) MDS, whereas proliferation arrest is observed predominantly in lenalidomide-induced responses in non-del(5q) MDS.98

Class III RTKs: PDGFR/c-kit/FLT3-mediated pathways

The Class III RTKs, such as PDGFR, c-kit, and FLT3, are characterized by the presence of 5 Ig-like domains in the extracellular region and by the separation of the catalytic TK domain in 2 parts by the insertion of a specific hydrophilic ‘interkinase’ sequence of variable length. The split TK feature separates this class from other RTKs, a notable characteristic that is shared by VEGFRs. Therefore, it was not surprising to find eventually that ATP mimetics that initially were designed to inhibit PDGFR also inhibited c-kit and FLT3 activity and, with more versatile multitargeted TKIs, inhibited the activity of VEGFR as well.

Stem cell factor (SCF) (also called steel factor or mast cell growth factor) is the ligand of the c-Kit TK receptor that is expressed in hematopoietic progenitor cells, normal mature mast cells, Cajal cells, melanocytes, and germ cells. Receptor dimerization activates multiple downstream substrate proteins, such as PI3K and src family kinases.99

Gain-of-function mutations of c-kit occur in several human neoplasms, including mastocytosis (>90%), GISTs (90%), sinonasal T-cell lymphomas (17%), seminomas/dysgerminomas (9%), and acute myelogenous leukemia (1%).100 There is an apparent association between the type of c-kit mutation and specific disease groups. In the majority of adult systemic mastocytosis, the most frequent mutation in the activation loop is D816V, whereas GISTs typically are negative for D816V and commonly exhibit activating mutations at the regulatory juxtamembrane region in exon 11 instead.101 The PDGFR/PDGF system includes 2 receptors (PDGFRα and PDGFRβ) and 4 ligands (PDGF-A, PDGF-B, PDGF-C, and PDGF-D). It is involved in angiogenesis and in the regulation of vascular permeability and interstitial fluid pressure. It can become active constitutively in a variety of ways. Cytogenetic abnormalities arising from reciprocal translocations or interstitial deletions in hematopoietic malignancies give rise to activated PDGFR fusion proteins.102 Moreover, gain-of-function mutations occur in positions analogous to those observed with c-kit in patients with GISTs. Although they are mutually exclusive, oncogenic mutations in PDGFRA and c-KIT result in the activation of common signaling pathways. FLT3 expression in the hematopoietic system, conversely, is restricted to CD34-positive stem/progenitor cells. FLT3 mutations are the most frequent genetic alteration reported in acute myeloid leukemia (AML) to date. Internal tandem duplication (FLT3/ITD) of the juxtamembrane domain-coding sequence and mutations of D835 within the activation loop of the kinase domain are the most common causes of constitutive activation of FLT3 encountered among patients with AML.103, 104 Imatinib is an inhibitor of several TKs that include c-KIT and PDGFR. It is effective in gastrointestinal stromal tumors (GIST) and dermatofibrosarcoma protuberans and is the treatment of choice for patients with chronic myelomonocytic leukemia (CMML) harboring the activating translocations involving the PDGFβ receptor locus on chromosome 5q33.105 Conversely, structural changes induced by the D816V mutation prevents imatinib binding and limits the use of imatinib in mastocytosis to cases in which the D816V mutation is absent or FIP1L1/PDGFRA gene rearrangements are present.

Cytotoxic T-lymphocyte–associated antigen 4-mediated pathway

Intratumoral T cells may modify tumor stroma or tumor cells in ways that modulate the metastatic potential of tumor cells. For example, the type, density, and location of immune cells within tumor samples were identified as better predictors of patient survival than the histopathologic methods currently used to stage CRC.106 Over the past decade, cytotoxic T-lymphocyte (CTL)–associated antigen 4 (CTLA4) has been established as a critical element in the homeostatic mechanism of immunosuppression and T-cell regulation. It is a transmembrane protein expressed on the surface of T lymphocytes and monocytes. For weak antigens, such as tumor-associated antigens, the activation of antigen-specific T-cell responses through interaction with T-cell receptors relies on nonantigen-specific signals generated when CD28 on the T-cell binds to B7 on antigen-presenting cells. Once this occurs, CTLA4 subsequently becomes expressed on the T-cell surface within 2 or 3 days of T-cell activation.107

CTLA4 exhibits a higher binding affinity to B7, out competing its prior interaction with CD28, thus abrogating the immune response,108 including decreased interleukin 2 (IL-2) and IL-2 receptor expression.109 Treatment leading to decreased CTLA4 expression in patients with melanoma also was correlated with improved outcomes.110 CTLA4 blockade also has demonstrated therapeutic efficacy in murine models of prostate cancer, colon cancer, fibrosarcoma, and melanoma.111

Trelimumab, a fully human IgG2 MoAb, and ipilimumab, a fully human IgG1 MoAb, are 2 of the CTLA4-blocking agents that are in advanced stages of clinical development. Both bind to the CTLA-4 molecule, resulting in the inhibition of B7-CTLA4–mediated down-regulation of T-cell activation. Promising results in melanoma patients in early clinical trials have led to the design of ongoing phase 3 clinical trials to confirm the role of CTLA4 blockade in the treatment of metastatic melanoma. Phase 2 trials in patients with prostate cancer, pancreatic cancer, and colon cancer also are ongoing.

Intracellular Signaling Interface

Ras pathway

The RAS gene superfamily encodes guanosine triphosphate hydrolases (GTPases), which transduce signals that regulate cellular proliferation, differentiation, migration, and survival.112 Activating Ras mutations are observed in approximately 30% of human cancers. Farnesyl transferase (FT) inhibitors (FTIs) were developed ostensibly to block the posttranslation modification of Ras, an essential step for membrane anchorage to allow the initiation of downstream target activation (for targets such as Raf, mitogen-activated protein kinase 1, and extracellular signal-regulated kinases). Aside from Ras, posttranslation farnesylation is necessary for the activation of numerous regulatory proteins, such as rho, lamin, centromere-associated protein, protein tyrosine phosphatases.113 The finding that FTIs exert antitumor activity regardless of ras mutation status and retention of ras oncogenic function attests to the relevance of proteins other than Ras in mediating FTI-induced inhibition of angiogenesis, cell adhesion, and survival.114 In addition, FTIs can rapidly trigger the generation of reactive oxygen species (ROS) independent of its effect on FT. The genotoxic stress arising from ROS initiates various DNA-damage responses that lead to apoptosis.115 Tipifarnib and lonafarnib are the most widely studied FTIs. However, with several negative phase 3 studies, the development of these agents has been discontinued.116, 117 Currently, it is unclear when second-generation agents will become available for testing.

Abl pathway

Imatinib, as discussed above, exhibits plurality in its effects. The targets of its inhibition include the nonreceptor TK abl, which, in oncogenic form as a fusion protein with bcr in CML, constituted its first validation as a targeted therapy. However, primary and secondary resistance to imatinib is common. To address this, second-generation bcr/abl drugs have been developed, such as dasatinib and nilotinib. Nilotinib exhibits increased affinity for wild-type bcr-abl by 20- to 30-fold compared with imatinib while retaining similar activity against PDGFR and c-kit.118 Dasatinib, conversely, binds to its targets with greater affinity than either imatinib or nilotinib, making it 300-fold more potent than imatinib against abl kinase and 20-fold more potent against wild-type c-kit.119 Unlike imatinib, dasatinib is not a substrate of multidrug P-glycoprotein efflux pump.120, 121 Both nilotinib and dasatinib have demonstrated impressive clinical activity in dose-escalation phase 1 studies involving patients with imatinib-resistant CML.122, 123 Unlike nilotinib, dasatinib has dual specificity for both abl and src family kinases and remains effective in both imatinib-resistant and nilotinib-resistant CML, eg, CML arising from the overexpression of src family kinases (eg, LYN).124 In contrast, complete response rates with both agents were <10% in patients with systemic mastocytosis despite promising preclinical data.125, 126

Proteasome pathway

The systematic regulation of protein synthesis and degradation is essential for normal cellular function. For example, the Cbl E3 ubiquitin ligase ubiquitylates active EGF receptors and other RTKs, which results in down-regulation of these transmembrane receptors as they are targeted for endocytosis and lysosomal degradation. The loss of Cbl binding observed in several oncogenic variants of EGFR, c-Met, and c-Kit leads to their accumulation at the plasma membrane and sustained signaling.127

The ubiquitin-proteasome pathway mediates protein degradation in a stepwise fashion. Proteins become tagged with ubiquitin polypeptides during ubiquitination, which targets them for degradation by the multienzyme proteolytic complex 26S proteasome. in an ATP-dependent manner.128 The substrates for proteasomal degradation include proteins that mediate various cellular functions, such as transcription, stress response, cell cycle regulation, oncogenesis, ribosome biogenesis, cellular differentiation, angiogenesis, and DNA repair, which are dysregulated in neoplastic cells.128 The differential increased susceptibility of malignant cells to proteasome inhibition compared with normal cells purportedly arises from a greater reliance of malignant cells on proteasomal degradation of a higher proportion of mutated, misfolded proteins, which threaten cell viability. Results of preclinical and clinical studies of proteasome inhibitors indicate that their antitumor effects are associated with increased expression of several cell-cycle–regulatory/inhibitory proteins (such as p27/ KIP1 and p21/WAF1), caspase activation, and blockade of the nuclear factor κB (NF-κB) proliferative pathway, which overcomes bcl-2-mediated resistance to apoptosis and results in the reversal of chemoresistance and radioresistence and the inhibition of angiogenesis and metastasis.129

The 26S proteasome contains proteolytic core 20S and 19S regulatory subunits. Bortezomib, a dipeptide boronate, potently binds to a single threonine in the chymotrypsin-like site of the 20S proteasome and dissociates slowly, reversibly inhibiting the proteasome.130 Recent studies demonstrated that selective inactivation of the chymotrypsin-like site is inadequate, inhibiting protein breakdown by only 10% to 30%.131 Newer inhibitors are being developed to inhibit both the caspase-like and tyrpsin-like proteolytic activities of the proteasome that may be effective in cells resistant to bortezomib.132 Bortezomib is the first proteasome inhibitor to enter clinical development and is approved for the treatment of multiple myeloma. It also has demonstrated remarkable clinical activity in mantle cell lymphoma, another disease entity in which cyclin D1 is overexpressed. Ongoing studies are investigating bortezomib in combination with chemoradiation in lung cancer with cytotoxic chemotherapy as well as targeted therapy, such as erlotinib.

PI3K-Akt-mTOR pathway

The PI3K-Akt-mTOR pathway is one of the best characterized pathways mediating oncogenic signaling inputs from various sources, such as receptor TKs (RTKs). Constitutive activation is demonstrated in malignancies, such as small cell lung cancer, resulting in cell proliferation, cell survival, and chemoresistance. Development of PI3K inhibitors and Akt inhibitors for clinical use has lagged behind that of mTOR inhibitors, which found their initial pharmacologic niche in transplantation medicine. Although immunosuppressive drug therapy using calcineurin inhibitors has been implicated in the increased incidence of and deaths from malignancies in solid organ transplantation recipients, maintenance immunosuppression with mTOR inhibitors (eg, sirolimus and everolimus) was associated with a significantly reduced risk of developing post‒transplantation malignancies.133 This finding, coupled with preclinical rationale, spurred the development of mTOR inhibitors for the treatment of malignancies.

mTOR, also referred to as sirolimus effector protein, FK506-binding protein (FKBP12)-rapamycin–associated protein, rapamycin and FKBP12 target, rapamycin-associated protein, or rapamycin target, is a highly conserved serine/threonine kinase that acts as the ‘hub’ to relay signals coming in from 2 major pathways. mTOR is activated by PI3K/Akt through growth factor stimulation, whereas it is inhibited by the serine/threonine kinase 11 (LKB1)/AMP-activated kinase pathway in response to nutrient status (ie, amino acids, ATP levels).134 mTOR exists in 2 complexes: the raptor-mTOR complex (mTORC1) is rapamycin sensitive, and the rictor-mTOR complex (MTORC2) is rapamycin resistant.135 This is important to remember, because the mTOR2 complex is responsible for the full activation of Akt135 and may be responsible in part for mTOR resistance. Indeed, rapamycin-induced feedback activation of Akt has been demonstrated preclinically.136, 137 Nonetheless, because activation of mTOR releases translational repression and promotes mRNA translation and protein synthesis, inhibition of mTORC1 reduces the levels of oncogenic proteins, such as c-Myc, hyopxia-inducible factor 1a (HIFa), VEGF, and cyclin D1.138 Moreover, prolonged treatment with rapamycin can potentially inactivate mTORC2 in some cases.139

Temsirolimus (cell cycle inhibitor 779) is the third drug to receive approval from the FDA, after sunitinib and sorafenib, for the treatment of RCC. This approval was based on results from a large phase 3 trial in patients with previously untreated, poor-prognosis, metastatic RCC. Those results demonstrated that temsirolimus alone improved OS compared with interferon α (IFN-α).140 The addition of temsirolimus to IFN did not improve survival compared with IFN alone, and the authors attributed that result to the increased toxicities arising from the combination therapy. Several features of this drug are noteworthy. Unlike sunitinib and sorafenib, which block VEGFR, representing the distal/receptor arm of the angiogenesis-promoting VHL/hypoxia-HIF axis, temsirolimus inhibits the VHL/HIF axis proximally by decreasing HIFa and VEGF levels through mTOR kinase inhibition.141 This is interesting, because the majority of RCCs are the clear cell type, which characteristically is associated with loss of function of the VHL gene and the up-regulation of HIF1/2α, leading to VEGF overproduction.142 It is also the only targeted agent to date that has demonstrated a statistically significant, albeit modest improvement in OS in poor-risk patients. It has demonstrated promising activity in patients with mantle cell lymphoma and sarcomas.

Transcriptional regulation pathways

Epigenetic pathway

Epigenetic changes refer to heritable changes in gene expression that do not arise from mutations or polymorphisms in the nucleotide sequence. The best described epigenetic changes refer to promoter region CpG dinucleotide hypermetylation and histone deacetylation, both of which result in transcription repression. DNA methyltransferases (DNMTs) catalyze the transfer of a methyl group to the cytosine residues in CpG dinucleotides, which tend to cluster in promoter regions of genes. DNMT1 is the most abundant and important in maintaining methylation of the CpG islands that result in transcriptional silencing of gene expression, tumor suppressor genes, and DNA repair genes in malignancies.143 5-Azacitidine and 5-aza-2-deoxycytidine (decitabine) are demethylating agents that originally were developed as cell-cycle, S-phase–specific nucleoside analogues and were administered in high doses as classic cytotoxic agents for the treatment of myeloid malignancies. Further investigation revealed that, even at low concentrations, when these agents are incorporated into DNA, they covalently bind and trap DNMT enzymes, depleting the intact enzymes and causing passive demethylation.144 Hypomethylation of p15 is a representative event that often occurs after treatment with these demethylating agents, and consequent increased expression appears to correlate with clinical response in patients with MDS, although this is not a consistent finding.145 Both agents are approved for the treatment of MDS. Other mechanisms of action include restoration of the expression of major histocompatibility Class I molecules and cancer testis antigens on tumor cells, rendering the tumor cells susceptible to CTL attack.146 Moreover, demethylation and re-expression of the tumor-suppressor gene that encodes death-associated protein kinase can restore the IFNγ-mediated apoptotic cell-death pathway in leukemia cells.147 Both agents require prolonged intravenous or subcutaneous administration; thus, research to develop oral DNMT inhibitors continues.

DNA methylation alone is insufficient to repress transcription. Acetylation and deacetylation of histones further regulate transcription events in eukaryotic cells, with acetylation promoting transcriptional activation because of relaxation of the chromatin structure. Histone acetyltransferases add acetyl groups, whereas HDACs remove acetyl groups from lysine residues.148 Classified into 4 major classes, there are multiple isoforms of HDACs, and 18 have been identified in humans to date that are not redundant in function.149 Class I and II HDACs share conserved catalytic residues, and it has been demonstrated that they are overexpressed and mutated in cancer. Class III HDACs differ from Class I and II HDACs in their catalytic site and are not inhibited by compounds like trichostatin A or vorinostat (suberoylanilide hydroxamic acid or SAHA).150 HDAC inhibitors act as transcriptional repressors by promoting chromatin condensation, altering a relatively small proportion of expressed genes (2-10%) in transformed cells. This paradoxically represses as many genes as are induced.150 The cyclin-dependent kinase inhibitors p21 and p27 are among the more common genes induced by HDAC inhibition, which results in cell cycle G1 and G2 arrests. Moreover, because HDACs have multiple nonhistone substrates, such as hormone receptors, chaperone proteins, signal-transduction mediators, cytoskeletal proteins, etc, inhibition of HDACs also can alter cellular processes by transcription-independent mechanisms.149 All of these actions can lead to cell differentiation, apoptosis and growth arrest. Inhibition of angiogenesis also occurs by reduced expression of HIF-1 and VEGF.151

Cell death induced by HDAC inhibitors, by both intrinsic and extrinsic apoptotic pathways and by autophagy, occurs preferentially in transformed cells, because normal cells are relatively resistant to HDAC inhibitor-induced cytotoxicity.152–154 Vorinostat is a pan-HDAC inhibitor and is approved for the treatment of cutaneous T-cell lymphoma (CTCL). It is currently undergoing phase 3 evaluation for treatment of mesothelioma. Clinical development of this class of drugs emphasize combinatorial strategies with both cytotoxic chemotherapeutic agents as well as other novel anticancer agents to circumvent mechanism of resistance, such as demethylating agents, inhibitors of NF-κB, and the bcl-2 antiapoptotic pathway.155

Retinoid pathway

Retinoids modulate gene expression by activating nuclear retinoic acid receptors (RARs) and retinoid X receptors (RXRs), which are transcription factors that act predominantly as RAR-RXR heterodimers affecting cell differentiation, proliferation, and apoptosis.156 In the absence of ligands, RAR-RXR heterodimers form multiprotein complexes with transcriptional corepressors, such as HDACs, to silence gene expression.157 All-trans retinoic acid (ATRA), an RAR-specific ligand, constitutes the backbone of therapy for acute promyelocytic leukemia (APL), which is characterized by the promyelocytic leukemia (PML)-RAR-α fusion gene resulting from the chromosomal translocation t(15;17). The oncogenic activity of the fusion protein arises from the disruption of PML function as transcription coactivator of the tumor suppressor p53. Moreover, PML-RAR-α exhibits a higher affinity for HDACs, which then ‘superrepresses’ RAR-α signaling, thus suppressing the ability of myelopoietic cells to differentiate into mature neutrophils.158

Bexarotene, a synthetic retinoid analogue, belongs to a class of drug known as rexinoids that specifically activates RXRs and has a lower affinity for the RARs. It is an oral noncytotoxic drug and is the first of its class to be approved for the treatment of refractory CTCL. It has been demonstrated that bexarotine induces apoptosis in various types of malignant cells through caspase-3 activation in association with the down-regulation of RXR-α and RAR-α.159 However, it failed to improve OS in 2 phase 3 trials when it was combined with first-line chemotherapy administered for advanced NSCLC.160, 161

Apoptotic Pathways

The prevalent dogma indicates that inactivation of apoptosis, 1 of various modes of cellular death that may occur with chemotherapy, is central not only to the development of cancer but also to treatment resistance.162 The 2 main pathways of apoptosis, the intrinsic and extrinsic pathways, analogous to the convergence on a common pathway observed with the dichotomous complement and coagulation cascades, ultimately result in the activation of effector caspases, a family of cysteine proteases, to execute the programmed cell death. The extrinsic pathway is triggered by the ligation of receptors of the TNF superfamily that contain functional cytoplasmic death domains (DD), such as Fas/apoptosis inducing ligand-like protein 1 (Apo1), death receptor 4 (DR4)/TNF-related apoptosis inducing ligand-like receptor 1 (TRAILR1)/Apo2, and DR5/TRAILR2, by specific ligands. The intrinsic pathway is mediated by mitochondria with the release of mitochondrial intermembrane proteins, such as cytochrome c, and is regulated in part by the bcl-2 family of proapoptotic and antiapoptotic proteins.

Arsenic trioxide (ATO) promotes caspase-dependent apoptosis, disruption of mitochondrial respiration, and depletion of cellular organic thiols. ATO can induce dose- and time-dependent apoptosis in activated endothelial cells and can inhibit VEGF production.163 In the Far East, arsenic was the mainstay for the treatment of leukemia since the 1700s. In the 18th century, Thomas Fowler developed a potassium bicarbonate-based solution of ATO called ‘Fowler solution’, a tonic that was prescribed for the treatment of many diseases, including cancer, in the 18th and 19th centuries.164

When ATO was used in different types of leukemia, consistent responses occurred only in APL. ATO degrades this fusion protein in a proteolytic pattern different from that of ATRA, thus negating the suppressive effect of the PML-RAR-α protein.165, 166 At low concentrations (0.1-0.5 μM), ATO induces differentiation of malignant promyelocytes through inactivation of the PML-RARα fusion protein; whereas, at high concentrations (0.5-2 μM), it triggers apoptosis of the promyelocytes and other cancer cells.167 It is important to understand that PML-RAR-α is not required for the apoptotic response of APL cells to ATO.168 Induction of apoptosis is related to the generation and accumulation of high intracellular levels of ROS,169 which affect sulfhydryl-rich proteins, including enzymes that catalyze protein tyrosine phosphorylation and other proteins crucial in cytokine-mediated signal-transduction pathways.170, 171

Antiapoptotic members of the Bcl-2 protein family, such as Bcl-2 and Bcl-xL, often are overexpressed in many human cancers and contribute to tumor development and treatment resistance. G3139 (oblimersen), or Genasense, is an 18-base antisense phosphorothioate oligonucleotide complementary to the bcl-2 mRNA. Its binding to the cognate sequence of bcl-2 mRNA results in inhibition of mRNA translation and RNase H-mediated mRNA degradation.172 Oblimersen down-regulated bcl-2 protein in a concentration- and time-dependent manner and demonstrated preclinical evidence of activity in CLL. A randomized, multicenter, phase 3 study comparing chemotherapy with fludarabine and cyclophosphamide with or without oblimersen among patients with relapsed or refractory CLL demonstrated superior survival in the subset of patients who achieved an objective tumor response from therapy, although there was no significant difference in the TTP or OS in the by intention-to-treat analysis.173, 174 These results were deemed insufficient for drug approval by the FDA.

Similarly tantalizing, results from a phase 3 trial of dacarbazine alone or in combination with oblimersen in 771 patients with advanced malignant melanoma, although it did not achieve a statistical difference in the primary endpoint of OS, demonstrated a significant increase in PFS (2.6 months vs 1.6 months; P < .001) and objective response rates (13.5% vs 7.5%; P = .007) favoring the treatment arm that contained oblimersen.175 Retrospective analysis of serum lactate dehydrogenase (LDH) levels at study entry suggested that the benefit from oblimersen was limited to patients with normal serum LDH levels. This spurred the implementation of an ongoing phase 3 randomized trial of dacarbazine with or without oblimersen in chemotherapy-naive patients with advanced melanoma and low-normal LDH levels (the AGENDA trial). A potential criticism against that study, in which future results turned out to be nonsignificant, was the lack of patient selection, which limited inclusion to patients with tumors that overexpressed bcl-2.

Oral small molecule inhibitors of BCL-2 modeled after the naturally occurring bcl-2 inhibitor, BH3-interacting domain death agonist (BID), are in early-phase clinical studies. These include AT-101 (gossypol), ABT-263, and ABT-737, which inhibit both bcl-2 and related apoptotic inhibitors bcl-X and bcl-W.

Toxicities

Although targeted therapies ideally should have a better therapeutic index than conventional cytotoxic chemotherapy because of the premise that the selected targets are expressed differentially in tumor cells compared with normal tissues, the finding that the various signaling networks discussed above are involved in the physiology and homeostasis of normal cells inevitably leads to mechanism-based class effects. Nonspecific symptoms of fatigue and gastrointestinal complaints are common. Oncologists, thus, ear encountering more complex issues related not only to drug-induced morbidities and mortalities but also to quality of life and long-term maintenance treatment using these agents.

Cardiovascular

Hypertension and proteinuria are now well established phenotypic manifestations attributable to endothelial dysfunction arising from manipulation of the VEGF signaling pathway. A severe clinical presentation related to endothelial dysfunction in malignant hypertension is posterior reversible encephalopathy syndrome or PRES, which has been reported with bevacizumab176, 177 and with small molecule inhibitors, such as vatalanib.178 Moreover, because normal functions of the endothelium encompass both anticoagulant and procoagulant activities critical for physiologic maintenance of normal hemostasis and VEGF has direct hemostatic effects on the endothelial cells, both thrombotic events (eg, deep vein thormobosin) and hemorrhagic events (eg, epistaxis) are increased with anti-VEGF therapy, as demonstrated in the early bevacizumab trials in CRC and lung cancer and in the thalidomide/lenalidomide trials in multiple myeloma. Syncopal episodes arising from bradyarrhythmias are unique to thalidomide because of its vagolytic effect.179

Heart rate corrected QT (QTc) interval prolongation is recognized increasingly as a potential sequelae with any pharmacologic agent. The most common side effects of ATO in patients with APL are electrocardiograph changes, especially a delayed cardiac repolarization manifested by prolongation of the QT interval and the QTc interval.180 In patients receiving HDACs and drugs with EGFR-inhibitory activity, dose-related QTc prolongation has been documented. QTc prolongation also was observed in patients who received an early-generation FTI, L-778,123.181 Although they mostly have been asymptomatic and without clinical sequelae, fatal arrhythmias have been reported.182 The mechanism for this toxicity is unknown, although drug interactions with selective serotonin reuptake inhibitors taken by some of these patients cannot be excluded. Other forms of cardiotoxicity have been reported with other agents, such as sunitinib, in which left ventricular dysfunction is described in 11% to 15% of patients treated, and imatinib, in which a murine model demonstrated that c-abl inhibition mediated the toxic cardiomyopathy that this drug induced.183 Fluid retention resulting in pleural effusions or peripheral edema are observed with dasatinib and imatinib therapy. It is believed that the this is caused by inhibition of the TKs responsible for capillary integrity, such as PDGFR-β, a receptor expressed in pericytes that regulate vascular permeability and interstitial fluid pressure.184

The best-established model of drug-induced cardiomyopathy associated with novel therapies is that induced by long-term trastuzumab. The manifestations are similar to the congestive heart failure observed with anthracycline therapy. A retrospective analysis of data collected from over 1000 women who participated in phase 2 and 3 trastuzumab trials indicated that the incidence of Class III or IV cardiac dysfunction was 2% for patients who were receiving first-line trastuzumab, 4% for patients who were receiving trastuzumab in the refractory setting, 2% for patients who were receiving concurrent paclitaxel plus trastuzumab, 1% for patients who were receiving paclitaxel alone, 16% for patients who were receiving concurrent doxorubicin and cyclophosphamide (AC) plus trastuzumab, and 4% for patients who were receiving AC alone.185 Prior or concomitant therapy with an anthracycline, age, and underlying cardiac disease were independent risk factors.

Immunologic

Many of the novel therapies selectively target the cells involved in immune system regulation. Targeting CTLA4 is rational, because CTLA4-expressing regulatory T cells restrain the immune system from attacking cancer cells. Conversely, because it serves a crucial role in immune tolerance, thereby dampening the immune response in general, the expected consequence of CTLA4 blockade is treatment-induced autoimmunity manifested in diverse fashion that includes enterocolitis, dermatitis, hypophysitis, uveitis, hepatitis.186–188 On the basis of preclinical testing, a way to reduce such autoimmune response is by the administration of a vaccine together with anti-CTLA4 therapy to direct immune responses toward target antigens. This strategy currently is being explored in clinical trials.

On the opposite extreme, an acquired immunodeficiency syndrome-like immunodeficiency state maybe induced by the administration of alemtuzumab, which causes a profound and long-lasting depletion of mature B- and T-lymphocytes, natural killer cells, and monocytes.189 This abnormality may persist for >1 year; thus, routine prophylaxis against important bacterial, viral, and fungal infections are included in the treatment of patients receiving alemtuzumab.190 Similarly, T-cell function may be affected by TKIs. Murine models also reveal that imatinib selectively impair the expansion of memory CTLs.191 Dasatinib as well as high doses of imatinib suppresses T-cell proliferation and activation, presumably through inhibition of the lymphocyte-specific protein TK LCK.192, 193 These effects are context-dependent (eg, local drug concentration, duration of drug exposure), because immunostimulatory natural killer-cell activation194, 195 has been observed in vivo and is believed to mediate the antitumor effects in patients with GISTs who have no KIT or PDGFRA mutation and who still respond to imatinib mesylate. It has been demonstrated that retinoids such as ATRA enhance IFN-γ and IL-12 production by peripheral blood mononuclear cells, induce cell-mediated cytotoxicity, and stimulate natural killer activity in vivo.196, 197

A reminder that there potentially could be serious, unanticipated toxicity from these ‘targeted’ agents is the recent events surrounding the development of immunomodulatory drugs. TGN1412, a humanized MoAb that binds to the CD28 receptor of T cells, possessed superagonistic properties to fully activate T cells without the need for additional antigen-receptor stimulation. In a highly sensationalized phase 1 trial involving healthy volunteers, 4 of 6 patients who received the drug suffered multiorgan failure arising from severe hypersensitivity reaction.198 This triggered the temporary suspension of clinical trials involving immunomodulatory drugs in the UK in early 2006.

A cloud also was cast over the development of natalizumab, a recombinant MoAb against α4 integrins used in multiple sclerosis, which was withdrawn from the market after several reports of progressive multifocal leukoencephalopathy. The risk of developing this complication was estimated at 1 in 1000 patients who received treatment for an average of 18 months.199 It is hypothesized that blockade of integrin function is restricted immune cell migration that otherwise would have restricted the reactivation of dormant JC polyomavirus. Whether other drugs that target integrins will encounter a similar toxicity is yet to be demonstrated.

Metabolic

Metabolic derangements mimicking the lipid and glucose abnormalities observed with insulin resistance in diabetic patients are not unexpected with agents affecting the insulin-like growth factor/Akt/mTOR pathway. Hyperlipidemia, and severe hypertriglyceridemia in particular, has been associated with drug-induced pancreatitis. Hyperlipidemia (hypercholesterolemia, hypertriglyceridemia) also is a well known complication of retinoid and rexinoid therapy. It is believed that this process is mediated by increased plasma apolipoprotein C-III (apoC-III) concentrations as a consequence of induction of hepatic apoC-III gene expression at the transcriptional level through an RXR-mediated mechanism.200 Reversible central hypothyroidism with profound suppression of serum thyroid-stimulating hormone levels also is a well documented and anticipated adverse event of rexinoids that is mediated primarily through RXR.201 Hypothyroidism itself can contribute to the mechanistic explanation for the hyperlipidemia observed with this class of drug. In contrast, subclinical primary hypothyroidism is the form observed in a small proportion of patients who receive agents with antiangiogenic properties, specifically sunitinib, thalidomide, lenalidomide, and cediranib.

Asymptomatic hyperamylasemia, hyperlipasemia, and hypophosphatemia have been observed in 30% to 45% of patients treated with sorafenib. Conversely, severe hypomagnesemia in 10% to 15% of patients has been reported with cetuximab.202 It has been postulated that interference with proteins involved in the active transport of magnesium in the nephron is a possible mechanism.202 However, this toxicity is not observed with EGFR TKI therapy; thus, the true explanation is unclear.

Pulmonary

Drug-induced interstitial lung disease (ILD) can be progressive and fatal and, although it is rare, has been attributed causally to several novel therapies. Reversible ILD has been reported with mTOR inhibitors and imatinib therapy. Gefitinib induces ILD changes that mimic the changes induced by conventional cytotoxic chemotherapy.203 Although EGF signaling most likely represents a potential mechanism of protection from lung injury by maintaining barrier integrity,204 it is possible that EGFR inhibition may impair the ability of pneumocytes to respond to lung injury. Pharmacogenetic markers of susceptibility may account for the higher incidence rates noted in East Asian patients, and particularly in Japanese patients, compared with the incidence rates among patients who were treated in North America. The apparently disproportionately increased risk of ILD among Japanese patients also was observed with leflunomide205 and bortezomib therapy.206 Among patients with NSCLC who receive bevacizumab, life-threatening hemoptysis may occur. This has been correlated with the presence of a cavitating central lesion or squamous histology, although the latter most likely is a confounding variable because of its frequent occurrence in central airways.

Hematologic

Prolonged circulation time of radioimmunoconjugated MoAbs results in radiation exposure of normal organs and tissues, particularly tissue retention in the reticuloendothelial system and kidneys. Consequently, myelosuppression is a well known toxicity, especially among patients with hematologic disorders, either because of direct bone marrow involvement or because of prior extensive exposure to myelotoxic therapies.207 Although lower molecular weight fragments reduce bone marrow exposure because of faster blood clearance, reduction in tumor uptake is a trade-off.

Dose-dependent myelosuppression may be observed with many targeted agents, such as FTIs, abl/src kinase inhibitors, mTOR inhibitors, sunitinib, bortezomib, and immunomodulatory drugs (IMiDs) (lenalidomide more than thalidomide). Severe neutropenia is observed with gemtuzumab, an expected sequelae of its lineage-specific mechanism of targeting myeloid cells. Cylic thrombocytopenia is a frequent complication from treatment with bortezomib and is more pronounced among patients with hematologic disorders, because of either bone marrow involvement or extensive prior chemotherapies.208 Thrombocytopenia is transient and usually decreases approximately 60% regardless of baseline parameters during the first 10 days of each cycle, recovering toward baseline during the rest period between cycles.209 It is believed that this arises from reversible inhibition of platelet budding from megakaryocytes that, in turn, is a tightly regulated process, which depends in part on the activity of NF-κB, a target of proteasome inhibition.210

Neurologic

Painful length-dependent sensorimotor peripheral neuropathy caused by axonal damage and selective loss of large-diameter myelinated fibers may be observed with thalidomide.211 This usually occurs after prolonged administration of thalidomide, and the risk increases with cumulative doses exceeding 50 g, thus affecting up to 70% of patients who are treated for >6 months.212 Similar to other toxic neuropathies, this may improve but not completely resolve with drug cessation. Serial electrophysiologic studies to monitor sensory nerve action potential amplitudes and/or somatosensory evoked potential latencies have been advocated as sensitive tests to gauge the severity of neuropathy; however, there is insufficient evidence to suggest that these are useful for early detection.213 A pharmacogenetic explanation for susceptibility to neurotoxicity from thalidomide may be provided by the acetylator status of an individual patient, with the ‘slow acetylators’ at increased risk for neuropathy because of slower excretion of the drug.214 The incidence of neuropathy with lenalidomide, in contrast, is reduced markedly (<10%).

Similar to thalidomide, bortezomib also causes a length-dependent, axonal, small- and large-fiber polyneuropathy that predominantly is sensory in nature in approximately 33% of treated patients.215 Other drugs that induce sensory neuropathy, although at a lower frequency, include sorafenib, the FTI tipifarnib, and vandetanib. A shared attribute among these drugs is their potential antiangiogenic property, which may provide a mechanistic explanation for the neuropathy induced. Indeed, currently, it is believed that the neuropathic effect of antitumor drugs occurs through damage to vasa nervorum endothelial cells, in part through a VEGF-mediated mechanism.216 This is plausible, because the role of VEGF-dependent neuroprotection has been implicated previously in various neurodegenerative disorders.217 Loss of VEGF-mediated neuroprotection also may explain in part the increased incidence of neuropathy reported with the combination of bevacizumab with either oxaliplatin79 or paclitaxel,218 although this commonly is ascribed to the longer duration of therapy that patients received because of delayed disease progression compared with the treatment regimens without bevacizumab.

Others

Certain dermatologic manifestations, such as hair and/or skin depigmentation (sunitinib, sorafenib, lapatinib, imatinib), are related to c-kit or PDGFRβ effects based on observations in murine models of pigmentation disorders.219–221 Hand-foot skin reactions in up to 33% of patients who received sorafenib (fewer patients with sunitinib) also have been described, although the underlying mechanism is poorly understood. Subungual splinter hemorrhages that occur with these drugs are asymptomatic and idiopathic in nature. Stevens-Johnson syndrome and toxic epidermal necrolysis have been reported in conjunction with thalidomide therapy, although these are rare, because the most common dermatologic toxicity associated with IMiDs is a pruritic truncal and appendicular maculopapular rash. Skin eruptions caused by EGFR-directed therapies are characterized by interfollicular and intrafollicular papulopustules and usually are evident during the first 2 weeks of treatment. Polymorphic variations in the target, such as the CA repeats in intron 1 of the HER1/EGFR gene, may underlie the susceptibility to develop rash and tumor response.222

Diarrhea may occur in tandem with skin rash as part of the mucocutaneous toxicity experienced by patients receiving sorafenib, FTI, or EGFR-related therapy and also can occur independent of skin toxicities, such as with bortezomib therapy. Although mucositis is observed rarely with treatment using signal-transduction agents, oral stomatitis (generally mild) is commonplace among patients receiving mTOR inhibitors. Whereas reversible elevation of liver enzymes of various degrees may be observed in most of the targeted agents, to date, veno-occlusive disease has been associated uniquely with gemtuzumab.223 Bowel perforation and fistulae formation are dreaded complications arising from anti-VEGF therapy, particularly with bevacizumab, predominantly in patients with intra-abdominal/pelvic tumors such as those encountered in clinical trials involving patients with CRC, ovarian cancer, and pancreatic cancer. This is likely a function of multiple interdependent conditions that impair mucosal healing, such as pre-existing abnormalities in intestinal vasculature from prior surgery, radiation and/or atherosclerotic disease, location of underlying disease, or typhlitis-like manifestation of myelosuppression with combination regimens. Although this complication arises infrequently (1.5% in patients with CRC), incidence has been reported as high as 11% in patients, particularly among women with ovarian cancer.82, 84, 85, 224

Conclusions

Over the last 60 years, the development of cancer therapeutics has evolved from DNA/mitosis-directed cytotoxic agents to targeting aberrant pathways, dysregulated signaling molecules, and tumor-specific antigens. Although some successes have been obtained, considerable refinement still is needed in our approach. It has become clear that, because of the redundancy in signaling pathways and survival mechanisms, narrowly targeted agents are not as effective as multitargeted agents, with the occasional exceptions (as exemplified by CML or GIST) of dependence on a dominant aberrant signaling node for which narrowly targeted agents can be successful. Further research to ascertain determinants of drug sensitivity conceptually would lead to a plethora of anticancer agents with clinical use based on pathway-driven selection, akin to antibiotic therapy. Moreover, the current research interest in combining various ‘targeted’ agents recapitulates the leitmotif with the use of traditional chemotherapy agents. Advancing the field of systemic cancer therapy requires sustained efforts in refining the molecular classification and diagnosis of cancer to complement traditional anatomic pathologic classification. We envision dynamic decision algorithms for therapy in the future based on molecular signatures unique to each individual patient and his/her cancer. Nonetheless, therapeutic manipulation of various signal-transduction pathways will be at a stalemate despite an extensive knowledge of the small-world network of intracellular signals and the dynamics of target activation and processing in the absence of concomitant improvement in drug design and delivery.

Ancillary