Antibody–drug conjugates and other nanomedicines: the frontier of gynaecological cancer treatment

Abstract

Gynaecological cancers: malignancies of the cervix, uterus, ovaries, vagina and vulva, are responsible for over 1.1 million new cancer cases and almost half a million deaths annually. Ovarian cancer in particular is difficult to treat due to often being diagnosed at a late stage, and the incidence of uterine and vulvar malignancies are both on the rise. The field of nanomedicine is beginning to introduce drugs into the clinic for oncological applications exemplified by the liposomal drugs, Doxil and Myocet, the nanoparticle, Abraxane and antibody–drug conjugates (ADCs), Kadcyla and Adcetris. With many more agents currently undergoing clinical trials, the field of nanomedicine promises to have a significant impact on cancer therapy. This review considers the state of the art for nanomedicines currently on the market and those being clinically evaluated for the treatment of gynaecological cancers. In particular, it focuses on ADCs and presents a methodology for their rational design and evaluation.

1. Introduction

Nanomedicine is the application of nano-sized agents for the treatment, diagnosis and prevention of disease. Beyond this basic tenet, however, nanomedicine champions the design-orientated approach to combatting disease, capitalizing on an understanding of disease pathophysiology to engineer agents with an anticipated effect. Progress in nanomedicine has been rapid. In the late 1960s, phospholipid-based liposomes and polymer-based nanoparticles were the first nano-sized agents to be investigated for clinical applications [1]. Since then, numerous other types of nanomedicine agents have been under evaluation including iron-oxide and gold nanoparticles, albumin–drug conjugates, solid lipid nanoparticles, carbon nanostructures and antibody–drug conjugates (ADCs; figure 1).

Figure 1.

Schematic of an ADC showing the main components of antibody, linker and payload with examples. SPDB: N-succinimidyl-4-(2-pyridyldithio) butanoate, SPP: N-succinimidyl 4-(2-pyridyldithio)-pentanoate, vc: valine–citrulline, va: valine–alanine, mcc: maleimidomethyl cyclohexane-1-carboxylate, mc: maleimidocaproyl, MMAE/MMAF: monomethyl auristatin E/F. (Online version in colour.)

To date, nanomedicine's biggest contribution to medicine has come from the formulation of drug delivery systems (DDS) such as liposomes, nanoparticles and ADCs. The packaging of drugs in DDS can lead to their prolonged circulation at therapeutic levels by reducing renal clearance. Their increased cellular permeability can significantly enhance cellular uptake of hydrophilic drugs, as well as facilitating their transport across the blood brain barrier [2]. Furthermore, DDS can serve to target their payloads to particular tissues and in the case of ADCs, to selectively target diseased cells, thus reducing the off-target toxicities associated with many free drugs [3,4]. Nanomedicine has also been applied to diagnostics, such as the use of nanoparticles and nanomaterials in the detection of disease biomarkers and disease-related genetic mutations, and as imaging agents for tumour detection [5–7]. A third set of nano-agents called theranostics, combine therapeutic and diagnostic capabilities in a single entity, delivering a mixture of drugs and imaging agents to a tumour site, or radiolabelled antitumour antibodies which provide radiotherapy as well as imaging [8,9]. Nanomedicine has also been employed successfully in vaccinology, where nanoparticles are used as antigen delivery systems and/or adjuvants [10].

It is anticipated that nanomedicine will have a major impact on the treatment of gynaecological cancers, including cancers of the uterus, ovaries, vagina, vulva and cervix which make up a huge proportion of new cancer incidence among women. Cervical cancer is directly attributed to infection by the human papilloma virus (HPV). While it remains the fourth most common cancer among women globally, early detection is possible through routine screening and the disease is easily preventable by immunization against HPV (http://www.cancerresearchuk.org/about-cancer/type/cervical-cancer). By contrast, vaginal and vulvar cancers are far more rare. They can be, but are not always, associated with HPV and incidence for both cancers is higher in those aged above 60 (http://www.nhs.uk/conditions/Cancer-of-the-vagina, http://www.nhs.uk/Conditions/Cancer-of-the-vulva). Ovarian cancer is known as the ‘silent killer’ as its symptoms are often non-specific and over 70% of cases are diagnosed at an advanced stage (Stages III and IV) [11]. The 5-year survival rate for late stage patients is only 25%, making ovarian cancer the biggest killer among the gynaecological cancers in the developed world [11,12]. Uterine, or endometrial cancer, is the most prevalent gynaecological cancer in the developed world and the fifth most commonly diagnosed cancer globally [13]. Though the disease generally develops post-menopausally, 14% of diagnoses are in pre-menopausal women [13]. Its incidence has shown an upward trajectory thought to be associated with increasing levels of obesity among pre-menopausal women [13]. While the 5-year survival rate for endometrial cancer is encouraging (approx. 85%), patients diagnosed with stages III and IV cancers have survival rates of approximately 60% and 30%, respectively [14]. Currently, treatment of gynaecological cancer will almost always involve some form of surgery, with hysterectomy typical for endometrial, ovarian and advanced cervical cancers. Depending on stage and cancer type, surgery will be followed by chemotherapy, radiotherapy or a combination of the two [13,15]. The development of nanomedicines that prevent the need for surgery, and thus retain reproductive function in pre-menopausal women is highly desirable.

The combined impact of gynaecological cancers is alarming, with recent cancer statistics showing that they account for over 16% of annual cancer diagnoses among women and nearly half a million mortalities [12]. It is clear that novel treatments will be needed to improve patient outcomes, especially for those with advanced stage disease. Nanomedicine is regarded as having the potential to swing the cancer statistics in favour of survival. In this review, we describe the types of nanomedicine agents in advanced development for the treatment of gynaecological cancers and examine individual nanotherapeutics currently in use and in clinical development for that purpose. These drugs, listed in table 1, can be classified into liposomal drugs, nanoparticle drugs and ADCs.

Table 1.

Nanomedicine agents on the market, or in clinical trials indicated for the treatment of gynaecological cancer. SCLC, small cell lung cancer; NSCLC, non-small-cell lung cancer.

name	targeting mechanism	active component	market/trial status	indications
liposome-based drugs
Doxil/Caelyx [16–21]	passive	doxorubicin	on the market	ovarian cancer, Karposi's sarcoma, multiple myeloma
LEP-ETU [22,23]	passive	paclitaxel	phase IV	ovarian and other solid tumours
Myocet [24–26]	passive	doxorubicin	marketed in Europe and Canada	ovarian and other solid tumours
S-CKD602 [27]	passive	CKD602	phase I/IIa	ovarian cancer, SCLC
OSI-211 [28]	passive	topotecan	phase II	ovarian, lung, head and neck cancers
siRNA-EphA2 [29,30]	passive	siRNA-EphA2	phase I	ovarian and other solid tumours
nanoparticle-based drugs
abraxane [31–34]	passive	paclitaxel	on the market	metastatic breast and pancreatic cancers, NSCLC, and in trials for ovarian cancer
CRLX101 [35–37]	passive	camptothecin	phases I and II	ovarian and other solid tumours
opaxio/PPX [38,39]	passive	paclitaxel	phase III	ovarian cancer, NSCLC
antibody–drug conjugates
BMS-986148 [40]	anti-mesothelin mAb	undisclosed	phases I and IIa	ovarian and other solid tumours
anetumab ravtansine/BAY 94-9343 [41]	anti-mesothelin mAb	DM4	phases I and II	ovarian and other solid tumours
lifastuzumab vedotin /RG-7599/DNIB0600A [42]	anti-NaPi2b mAb	MMAE	phases I and II	ovarian, fallopian tube, peritoneal cancers, NSCLC
IMGN853/Mirvetuximab soravtansine [43,44]	anti-FRα mAb	DM4	phases I and II	ovarian, endometrial. fallopian tube, peritoneal cancers
SAR566658 [45,46]	anti-CA6 mAb	DM4	phase I	ovarian, breast cancers
tisotumab vedotin/HuMAX-TF [47]	anti-TF mAb	MMAE	phases I and II	ovarian, endometrial, cervical, prostate, bladder cancers
PF-06647263 [48]	anti-EFNA4 mAb	calicheamicin	phase I	ovarian, breast cancers

2. Nanomedicine: the therapeutic payloads

The majority of nanomedicine drugs (nanomedicines) in development combine a therapeutic payload with a drug delivery system (DDS) for improved targeting, uptake and reduced off-target toxicities. A number of nanomedicines on the market and in development make use of drugs already being used in their free form such as pegylated cisplatin for cancer treatment, pegylated cytokines for treatment of hepatitis B and C, liposomal amphotericin b (an antifungal agent), liposomal morphine for pain relief and Fe³⁺ complexes for iron deficiency [49]. The use of DDS in the tumour-targeted delivery of kinase inhibitors is also beginning to emerge, with for example nanoparticles loaded with the BCR-Abl inhibitor Gleevec demonstrating increased efficacy in breast cancer cell lines and reduced off-target toxicities in animal studies compared with free Gleevec [50]. Many nanomedicine payloads, however, must be used exclusively in combination with a drug delivery vehicle, because either they are too toxic to be used as free agents, or because their uptake in free form is too inefficient to provide a therapeutic benefit. Examples of the former include the potent cytotoxins used in ADCs. An example of the latter is nucleic acid-based drugs such as micro RNA (miRNA) and short-interfering RNA (siRNA), where due to drug instability, low cellular uptake, immunogenicity and possible off-target effects, DDS have proved essential for the use of such drugs in treating systemic disease [51].

For some nanomedicines, there is no distinguishable division between drug payload and DDS, rather it is the nanostructure itself which provides the therapeutic effect. Perhaps the most promising example of this is nanoparticle-based hyperthermia. Tumour cells are more sensitive to hyperthermia than normal cells due to the hypoxic and acidic tumour microenvironment [52]. Hyperthermia treatment has traditionally involved either perfusion of a limb, or organ with a heated fluid (for local treatment), or irradiation with infrared, radiofrequency and microwave electromagnetic energy [53]. These methods, however, are nonspecific with normal tissue invariably affected and can cause serious side effects, including cardiac failure [54]. Nanoparticle-based hyperthermia therapy allows for a more targeted hyperthermia treatment with potentially higher efficacy and reduced collateral damage. The types of nanoparticles being investigated for hyperthermia therapy include various types of gold nanoparticles, superparamagnetic iron-oxide nanoparticles (SPIONs) and carbon nanotubes (CNTs). These nanoparticles all absorb energy in one form and release it as heat: near infrared energy or radio frequency energy is absorbed by gold nanoparticles and CNTs, and SPIONs release heat when excited under a magnetic field [54].

3. Nanomedicine targeting

Nanoscale DDS for cancer therapy rely, in part, on tumour targeting for their efficacy of which there are two broad mechanisms: passive and active.

3.1. Passive targeting

Unlike normal vasculature, the blood vessels of tumours are highly fenestrated and lacking in smooth muscle making them highly permeable [55]. The high levels of angiogenic and vascular permeability factors produced by cancer cells cause tumours to become hypervascularized and the vasculature to be in a perpetually dilated state [55]. These features allow for the efficient uptake of nutrients and oxygen required for tumour growth. In addition, the defective lymphatics characteristic of solid tumours prevent loss of nutrients and macromolecules through tissue drainage [55]. These pathophysiological differences between solid tumours and normal tissue is termed the enhanced permeability and retention (EPR) effect, and it is the main mechanism through which nanomedicines are passively targeted to tumour sites. The high permeability of tumour vasculature and the increased retentive properties of tumours allow for preferential accumulation of an antitumour drug, provided it is small enough to pass through the gaps in tumour microvasculature, but large enough to evade renal clearance and diffusion into normal tissue. Studies with sarcoma-implanted mice injected with radiolabelled proteins of different size determined the optimum molecular weight for a particle to avoid rapid blood clearance while efficiently accumulating in tumour tissue to be 15–70 kDa [56]. Other important parameters in determining a particle's pharmacokinetics and ability to extravasate into the intratumoural space include size and shape: the optimum diameter of spherical particles for tumour deposition has been determined to be 100 nm and oblate spheroid nanoparticles have been shown to be superior to spherical particles in this respect [57,58]. Oblate particles experience lateral drift, which causes them to spend more time at blood vessel walls [59] and results in lower uptake by circulating macrophages [60]. It is prudent to point out, however, that size and shape dependency studies for nanoparticle uptake should be considered in the context of the tumour models with which they were conducted as different tumour types at different stages of progression show variation in the size of endothelium gaps, interstitial pressures and vascular geometry [58].

3.2. Active targeting

In active targeting, ligands which interact with cell surface structures at the tumour site are conjugated to the surface of nanomedicines steering them to the tumour, or surrounding vasculature [61]. Several studies have shown that active targeting of nanoparticles does not affect biodistribution (which is primarily governed by size, shape and vascular permeability), but does enhance cellular uptake [62–64]. The main internalization mechanism of ligand-bound nanomedicines is receptor-mediated endocytosis following interaction of the drug with its target antigen [65]. The characteristics of the tumour antigen are therefore crucial to the efficacy of the targeted therapeutic. In general, to be considered a good target, a tumour antigen should be highly expressed on the tumour cell surface relative to normal cells and should internalize efficiently upon ligand binding [66]. Moreover, to maximize release of therapeutic within the cell, the antigen should be transferred into the lysosome, avoiding recycling to the cell surface [66]. An example of an antigen being evaluated in our laboratory for tumour targeting is the chemokine, fractalkine (CX3CL1). CX3CL1 exists as a membrane-bound protein involved in leucocyte–endothelial cell adhesion and in a soluble form, acting as a chemoattractant to leucocytes [67]. CX3CL1 has also been reported to play a role in epithelial ovarian cancer (EOC) growth and has been shown to be highly expressed in EOC biopsies with minimal expression on normal ovarian tissue [68]. Recent data from our laboratory (figure 2) have shown that CX3CL1 is also highly expressed in the Type II, ERα (oestrogen receptor α)-negative endometrial cancer cell line HEC1A, and only minimally expressed in biopsies of healthy endometrial tissue. While possible issues could arise from off-target expression of CX3CL1, our data and that of other groups [68] warrants further evaluation of this antigen for targeting in gynaecological cancers.

Figure 2.

Rational approach to ADC development. Selection of tumour antigens is conducted in silico to generate a list of ADC targets (a). Expression levels and internalization properties of target antigens are then tested in vitro with the most promising candidates selected for antibody generation. Panel (b) shows our experimental data for the expression of CX3CL1, identified in our in silico screen, in endometrial Type II cancer cell line (HEC1A) and healthy cells isolated from a non-malignant tissue biopsy. Immunoblotting shows cancer cell specific expression (Vinculin = positive control). Immunofluorescences confocal microscopy shows membrane localized CX3CL1 only in HEC1A cells. Internalization (c) of anti-CX3CL1 mAb in HEC1A cells is revealed by a punctuated pattern of cytoplasmic localized antibody consistent with endosomal accumulation. In vitro evaluation (d) of anti-HER2 MMAE ADC in HEC1A shows decreased cell viability at higher ADC exposure, compared with healthy cells.

4. Nanomedicines for gynaecological cancers

4.1. Liposomal nanocarriers

Liposomes are small, spherical vesicles of one (unilamellar) or more (multilamellar) phospholipid bilayers, which enclose an aqueous internal compartment [69]. With the ability to carry both hydrophilic macromolecules in their aqueous core and lipophilic macromolecules within their lipid membranes, liposomes can act as suitable carriers for a broad range of drugs [70]. Their ability to self-assemble, biocompatibility and large carrying capacity all contribute to their popularity as nanomedicine agents. Furthermore, the liposome tool box contains a multitude of lipids and lipid modifications allowing the formulation of liposomes, which vary widely in their biophysical and biological properties [71]. For instance, surface charge has been shown to influence the targeting of lysosomes, with cationic liposomes showing more efficient targeting to angiogenic tumour vessels and, due to their more efficient extravasation, neutral and anionic liposomes proving more suitable for targeting to extravascular tumour tissue [72]. Pegylation, or the addition of polyethylene glycol (PEG) polymer chains to liposomes, increases the stability of liposomes in vivo. The PEG molecules sterically interfere with liposome opsonization in the blood stream and clearance through the reticuloendothelial system (RES) significantly increasing circulation times and decreasing toxicity [71]. In addition, the conjugation of ligands, antibodies and imaging agents to liposomes can transform them into actively targeted DDS and imaging agents [73]. Liposome-based drugs are currently in use for the treatment of cancers, fungal infections, pain management, macular degeneration and as vaccines [69–71].

4.2. Liposomal-based drugs on the market and in clinical development for the treatment of gynaecological cancer

In 1995, the liposome-based drug, Doxil (Caelyx) became the first nanomedicine to enter the market, receiving FDA approval for the treatment of Kaposi sarcoma and later, in 1999, for the treatment of platinum resistant ovarian cancer [16]. Doxil is a pegylated liposome loaded with the anthracycline, doxorubicin, a DNA intercalating agent routinely used to treat a range of cancers, including breast, lung, ovarian and gastric cancers [17]. Doxorubicin is associated with several side effects the most significant of which is cardiomyopathy, which limits cumulative dosing of the free drug to 450 mg m⁻² [74]. While a phase III clinical trial of Doxil versus doxorubicin in metastatic breast cancer patients showed no significant difference in efficacy [18], it, along with several other trials indicated a significantly lower incidence of cardiotoxicities in patients receiving Doxil [19]. Studies have confirmed that there is preferential accumulation of Doxil in the tumour and compared with free doxorubicin, reduced uptake of the liposomal drug in organs such as the heart; an effect attributable to the EPR effect [20]. Despite improved targeting, however, Doxil is not without its toxicities, the most common of which are asthenia and palmar-plantar erythrodysesthesia [21].

Myocet is a second doxorubicin-loaded liposome in use and is approved in Europe and Canada for metastatic breast cancer and in clinical trials for the treatment of ovarian, endometrial and fallopian tube cancers. Although both drugs contain doxorubicin as an antitumour agent, the Myocet liposome is composed of an egg phosphatidylcholine and cholesterol and unlike Doxil, is not pegylated [24]. In terms of efficacy, the majority of trials have shown little difference between Myocet and Doxil; however, Myocet benefits from a reduced toxicity profile, a fact attributed to its altered pharmacokinetics. The circulation time of Myocet has been shown to be significantly lower than Doxil and release of doxorubicin from the Myocet liposome occurs at a higher rate; both of which contribute to a shortened time of exposure of healthy tissue to doxorubicin [25]. The pharmacokinetic differences between these drugs have been attributed to the lack of pegylation in Myocet [26] and illustrates how influential differences in DDS design can be on drug tolerability.

In 2015, liposome encapsulated paclitaxel, or LEP-ETU, received orphan drug status for the treatment of ovarian cancer. Paclitaxel binds tubulin, enhancing its polymerization and leading to a reorganization of the microtubule cytoskeleton that results in G2/M arrest [75]. Paclitaxel is not water soluble and is prepared in polyoxyethylated castor oil for intravenous administration. The castor oil-based solvent is associated with strong hypersensitivity reactions which can be fatal [76]. LEP-ETU and other nanomedicine-based paclitaxel delivery systems eliminate the need for this solvent and could potentially reduce the toxicity of paclitaxel therapy [22]. A bioequivalence trial involving 32 patients with advanced cancer showed similar efficacies for LEP-ETU compared with standard preparation paclitaxel [23].

The only siRNA drug currently in clinical trials for the treatment of gynaecological cancers is siRNA-EphA2-DOPC, a liposome encapsulated siRNA targeting the transmembrane tyrosine kinase, ephrin type-A receptor 2 (EphA2). EphA2 has been shown to be overexpressed in numerous cancers including ovarian cancer where it is associated with tumour progression and a poor outcome [77]. SiRNA-EphA2-DOPC is passively targeted to the tumour site via the EPR effect, but its siRNA payload provides an additional element of tumour selectivity. Preclinical studies have demonstrated significant reductions in tumour size in mice carrying ovarian cancer xenografts when given the siRNA-EphA2-DOPC both via intravenous and intraperitoneal administration [29,30]. The antitumour effect seen with the liposomal-based drug was 30 times greater than that seen following administration of free EphA2 siRNA [30].

4.3. Non-liposomal nanoparticles

Nanoparticles can be made from a broad range of biological and synthetic materials and form a diverse array of structures [3]. The majority of nanoparticles in clinical development for cancer treatment are polymeric nanoparticles (PNPs). PNPs are solid particles, or particulate dispersions of particles that can take the form of nanocapsules or nanospheres [78]. Nanospheres are spherical polymer matrices on which a therapeutic payload is equally distributed throughout, whereas in nanocapsules, the therapeutic cargo is encapsulated by a polymer membrane [78]. As with liposomes, the size of PNPs facilitates tumour targeting via the EPR and can be further enhanced by surface conjugation of tumour antigen ligands [79]. One of the most significant advantages of PNPs as DDS is the potential for controlled and sustained drug release. Drug release from PNPs occurs through passive mechanisms, such as diffusion through pores, or gaps in the polymer matrices, osmotic pressure pumping drugs out of a non-swelling system, the degradation of the polymer surface, or the homogeneous degradation of the entire particle [80]. The choice of polymer(s) will determine the mechanisms of drug release, and drug release kinetics are influenced by the density of drug–polymer packing, drug–polymer interactions, drug–drug interactions and polymer matrix pore size [80]. Recently, nanoparticles have been developed which release payloads in response to environmental changes such as pH, redox potential, the presence of reactive oxygen species, or in the presence of lysosomal proteases [80]. PNPs which release drug in response to external stimuli, such as through heat, UV and infrared radiation, or ultrasound, are also being investigated [80].

4.4. Nanoparticle-based drugs in use and clinical development for the treatment of gynaecological cancer

Nab-paclitaxel (Abraxane) is the brand name for albumin-bound paclitaxel, a nanoparticle-based drug which received FDA approval for use in the treatment of metastatic breast cancer in 2005, NSCLC (non-small-cell lung cancer) in 2012, pancreatic cancer in 2013 and is currently in clinical trials for use in ovarian cancer patients. Free paclitaxel is often used as a first line treatment for ovarian cancer [81]. It is postulated that the conjugation of paclitaxel to albumin will improve its intratumoural uptake via gp60-dependent transcytosis through the endothelial layer of blood vessels [31]. Albumin binds gp60 on the surface of endothelial cells resulting in its internalization and transport through the cell in caveolae before being released into the interstitial space. Several phase II trial reports have been published with Nab-paclitaxel used as either a single agent or in combination with carboplatin to treat recurrent, or persistent epithelial ovarian cancer [32,33]. These studies have indicated high tolerability for the drug and encouraging efficacy in patient groups with aggressive disease. Individual case studies using Abraxane in recurrent ovarian cancer have shown that a total remission of the cancer can be achieved [34].

Paclitaxel Poliglumex (PPX) is a second nanoparticle drug featuring a paclitaxel payload and is currently in phase III trials for numerous malignancies including ovarian and fallopian tube cancers. In PPX, paclitaxel is conjugated to the synthetic polyamino acid, poly(l-glutamic acid), improving its tumour targeting and prolonging its plasma circulation [38]. The glutamic acid residues block the tubulin binding site of paclitaxel rendering the drug inactive [39]. Passive targeting through the EPR effect leads to accumulation of PPX in tumours, where it is taken up by local phagocytic cells such as tumour macrophages [39]. Active paclitaxel is released upon cleavage of the glutamic acid residues by cathepsin B in the tumour cell lysosomes [39]. Phase II trial data show PPX yields similar response rates as free paclitaxel when used as a first line therapy for ovarian cancer treatment [38]. Thrombocytopenia is a common adverse effect of PPX, which is not seen with paclitaxel; however, their associated treatment toxicities are otherwise very similar [38]. PPX is currently being evaluated in a phase III trial to determine whether it can offer any benefit as a maintenance therapy in ovarian cancer patients [38].

The polymeric nanoparticle drug CRLX101 is currently being trialled for the treatment of various solid tumours including ovarian cancer. In CRLX101, the topoisomerase and hypoxia inducible factor 1 alpha (HIF-1α) inhibitor camptothecin (CPT) is conjugated to a cyclodextrin-PEG co-polymer, which self-assembles to form nanoparticles [35]. Tumour biopsies taken in clinical trials have shown that the drug accumulates preferentially in tumour tissue versus nearby non-neoplastic tissue [36]. A phase I/IIa trial in which CRLX101 was administered as a single agent in patients with advanced solid tumours has shown the drug to be well tolerated and several phase Ib and phase II trials involving CRLX202 have been initiated for various malignancies including ovarian cancer [37].

5. Antibody–drug conjugates

ADCs (represented in figure 1) are a quintessential example of actively targeted nanomedicines. Composed of a monoclonal antibody (mAb) conjugated to cytotoxic drugs via blood-stable linkers, ADCs interact with a target antigen locally delivering a cytotoxic payload to the tumour site. Upon binding antigens on the tumour cell surface, the ADC is internalized and releases its drug component intracellularly leading to cell death [82]. ADCs targeting tumour vasculature, or supporting stromal tissue release their payload externally at the tumour site [83].

5.1. Antibody–drug conjugate payloads

The cytotoxic ‘payloads’ used in ADCs are typically several orders of magnitude more potent than free circulating chemotherapeutics [83]. Such potencies are required, because the efficiency of drug uptake in the tumour is currently very low. The diffusion of ADCs into the intratumoural space is hindered by their large size (more than 150 kDa) and the high interstitial fluid pressure found in solid tumours; it has been estimated that only between 0.003 and 0.08% of injected antibody dose will accumulate per gram of tumour [66]. The vast majority of drugs being used in ADCs exert their antitumour effect either by intercalating with the DNA double helix, or interfering with microtubule formation. Examples of DNA damaging agents include calicheamicin, duocarmycin, daunomycin (daunorubicin) and pyrrolobenzodiazepines (PBD); all of which intercalate to the minor groove of the DNA double helix. Calicheamicin induces double-stranded DNA breaks and duocarmycin is a sequence-specific alkylating agent [83,84]. Both PBDs and daunomycin interfere with DNA replication; the former by cross-linking opposing DNA strands and the latter by inhibiting topoisomerase II progression [83]. Drugs which interfere with microtubule dynamics include the auristatins, such as monomethyl auristatin E and F (MMAE and MMAF) and the maytansinoids, DM1 and DM4. These drugs have proven the most popular ADC payload with a 2015 estimate placing them in 75% of all ADCs in clinical development [85]. Both classes of drug interfere with microtubule stability through the binding of tubulin, resulting in G2/M cell-cycle arrest [86,87].

The choice of ADC payload is an important strategic decision; the expression of the target antigen, tumour characteristics and the nature of cells being targeted must all be considered. For instance, when targeting a tumour antigen with high expression in normal tissues, selection of a payload that selectively targets proliferating cells, such as MMAE, may help to reduce off-target effects. Conversely, when directing ADCs against cancer cells which are less proliferative such as cancer stem cells, a DNA damaging agent, such as calicheamicin, might be preferred. An additional factor, which influences choice of payload is its capacity to exhibit the bystander effect, in other words their ability to cross the membranes of nearby cells following efflux from primary target cells, or release from lysed cells [88]. This ability typically depends on the drug's lipophilicity. MMAE, for instance, exhibits a strong bystander effect as seen in clinical studies of the CD30 targeting ADC, SGN-35 [89]. By contrast, the MMAE analogue, MMAF, in which the terminal ephedrine group is substituted for phenylalanine, does not show a bystander effect and is far less toxic against a broad panel of CD30 expressing haematological cell lines [83]. While the bystander effect can result in toxicity to healthy cells, it can also be a powerful mechanism for killing tumour cells not expressing target antigens and for treating tumour by targeting tumour stroma and vasculature.

Despite the high cytotoxicity of the payloads described above, the enormous diversity in pathophysiology between tumour types and the heterogeneity seen between and even within individual tumours, means there is still a need for the development of novel ADC payloads. For instance, although the auristatins, maytansinoids and daunomycin are highly effective in killing proliferating cells, these cells have been shown in various tumour types to make up only a small proportion of total tumour tissue [90]. Furthermore, drug resistance is a persistent issue in cancer therapy and has been shown to exist against the majority of ADC payloads in clinical trials including calicheamicin, DM1, MMAE, doxorubicin and daunomycin [91]. There is therefore a clear need for a wide spectrum of payloads operating by different mechanisms. One agent showing great promise as an ADC payload is α-amanitin, a toxin found in several species of Amanita mushrooms. α-amanitin is a potent inhibitor of RNA polymerase II causing a block of transcription and inducing apoptosis; the toxin is effective against both dividing and resting tumour cells [92]. The drug has been tested in ADC format in a preclinical study in which an α-amanitin-anti-EpCAM antibody achieved significant shrinking of pancreatic tumour xenografts in mice. The cryptophycins, microtubule-binding agents isolated from cyanobacteria, are another class of compounds being evaluated as ADC payloads. These agents were shown to be more potent than MMAE conjugated ADCs and unsusceptible to extrusion by the permeability glycoprotein (P-gp) pump; a common source of chemotherapy resistance [93]. In addition, kinase inhibitors are currently being evaluated in nanoparticles and the potential exists for them to be adapted as ADC payloads.

5.2. The antibody: considerations

The mAb serves as the vehicle which transports the ADC's highly cytotoxic payload to the tumour site. The earliest mAbs developed for therapeutic use in 1975 were murine antibodies, however, these proved highly immunogenic and rapidly cleared from circulation [94]. Since then, the evolution of antibody engineering has led to the development of chimeric (containing a human constant region and murine variable region) and humanized antibodies (human in all regions except at antigen-interacting sequences) in the 1980s and the ability to produce fully human mAbs in the 1990s [95]. All three of these antibody types have been used in ADCs, however, the latest generations of ADC predominantly use either humanized or fully human ADCs [85,88].

One of the characteristics of antibodies that makes them useful drug delivery vehicles is their prolonged blood-circulation time, which permits them to be administered less frequently than conventional chemotherapeutics. With a mass of approximately 150 kDa, the size of IgG antibodies prevents their renal clearance [88]. In addition, IgG antibodies bind to the neonatal Fc receptors (FcRn) expressed on the surface and within the endosomes of vascular endothelial cells, macrophages and monocytes. Whereas the majority of plasma proteins taken up by these cells end up in the lysosomes and are digested, binding of antibodies to FcRn leads to them being shuttled back to the cell surface where they can dissociate and re-enter circulation; a process which also serves to protect these cells from ADC-induced toxicity [96]. The binding of antibodies to FcRn effectively produces an antibody reservoir that prolongs the circulatory half-life of IgG1, IgG2 and IgG4 antibodies to approximately three weeks [96].

In addition to their function in tumour targeting, the ADC antibody could potentially play a direct therapeutic role when the antigen it targets has a pro-tumour activity. The anti-HER2 mAb, Trastuzumab is often cited to illustrate this effect. HER2 is overexpressed in 20–30% of all breast cancers as well as in endometrial, ovarian and other cancers, where it is associated with cell proliferation and survival [97]. The binding of Trastuzumab with HER2 causes its internalization and degradation as well as preventing dimerization. Together these effects suppress signalling through the PI3 K/Akt, Ras/Raf and MEK pathways inhibiting tumour cell proliferation and survival [97]. The success of Trastuzumab in improving time to progression and survival rates of breast cancer patients prompted the development of the ADC Trastuzumab emtansine (Kadcyla); conceived to combine the antitumour effect of Trastuzumab with that of the cytotoxic payload DM1. However, it has yet to be ascertained the extent to which the antibody component of Kadcyla, when incorporated in the ADC, directly contributes to its antitumour effect.

A further mechanism through which antibodies can potentiate an antitumour effect is by stimulating a localized, immune response against tumour cells through processes known as antibody-dependent cell cytotoxicity (ADCC), complement-dependent cytotoxicity (CDC) and complement-dependent cell cytotoxicity (CDCC). In ADCC, cells of the innate immune system such as macrophages and natural killer cells (NKs) become activated through the interaction of the Fcγ receptors on these cells with the Fc region of the antibodies bound to the tumour cells [98]. Once activated, the immune cells will kill the tumour cells by phagocytosis in the case of macrophages, or in the case of NKs, by the release of a cocktail of proteases and other proteins that lyse the tumour cells. CDC and CDCC involve the binding of complement complex, C1 to the Fc region of the tumour bound antibody, which initiates the classical complement pathway cascade [99]. This cascade can either directly lead to tumour cell death by the formation of membrane attack complexes (as in CDC), or target the tumour cell for engulfment by phagocytes (CDCC). The extent to which antibodies induce ADCC, CDC and CDCC is dependent on isotype with IgG1 antibodies proving more efficient than IgG2 and IgG4 antibodies in this respect [100]. This principle is also demonstrated in ADCs with Adcetris and Kadcyla, both of which feature IgG1 antibodies, exhibiting immuno-stimulatory effects, while Mylotarg, which features an IgG4 antibody, does not [88,101,102]. The capacity of antibodies in ADCs to interact with components of the immune system is also likely to be influenced by the structure and location of the conjugated payload.

5.3. Conjugation chemistry: linkers and spacers

The ADC linker serves as an adaptor between antibody and payload. It is the main determinant of drug release and therefore plays a major role in drug pharmacokinetics, physico-chemical properties and ultimately, clinical efficacy. The linker must be stable in plasma to prevent premature drug release, but allow release of the payload upon cellular uptake. In addition, to avoid ADC aggregation, linker hydrophobicity must be considered, particularly when in combination with a hydrophobic payload. ADC aggregation can prevent efficient drug loading during ADC manufacture as well as potentially causing hepatotoxicity and immunogenicity when administered [103]. Furthermore, the hydrophobicity of the drug-linker metabolites generated upon cleavage can impact their efflux from multi-drug resistance transporters, as well as their cell membrane permeability and ability to exhibit a bystander effect [103].

Linkers can be categorized into cleavable and non-cleavable. Cleavable linkers facilitate drug release upon exposure to particular conditions in the intracellular environment and include pH-, protease- and glutathione-sensitive linkers. The prototypical pH-sensitive linker contains a hydrazone group that is hydrolysed in the acidic environment of endosomes and lysosomes. The hydrazone linker features in the ADC, Inotuzumab ozogamicin (INO), which is currently undergoing phase III clinical trials for the treatment of acute lymphoblastic leukaemia (ALL). Protease-sensitive linkers contain a dipeptide, such as valine–citrulline (vc), which becomes enzymatically cleaved by lysosomal proteases. These linkers have been included in numerous ADCs for the conjugation of MMAE, MMAF, duocarmycin and PBD dimer payloads [103]. Glutathione-sensitive linkers contain disulfide bridges, which mainly through the action of the thiol containing glutathione, become reduced in the cytosol leading to payload release. Such disulfide-containing linkers (e.g. SPDB, SPP) feature in several ADCs with maytansinoid payloads [104]. Non-cleavable linkers rely on the total degradation of ADCs within lysosomes and result in drug metabolites released with a terminal amino acid. The most frequently used are maleimidocaproyl (mc) and maleimidomethyl cyclohexane-1-carboxyalte (mcc), the latter of which features in Kadcyla [103]. While the added stability of non-cleavable linkers serves as a benefit to some ADCs, the modification of payload caused by the terminal amino acids can reduce the activity of certain drugs as well as diminishing their ability to induce a bystander effect [105].

Additional components are often added to basic linkers to improve the drug release profile of ADCs. An example of this is the use of spacers in ADCs with MMAE payloads. Seattle Genetics has employed an mc-vc-PABC (para-amino benzyloxycarbonyl)-MMAE linker-drug platform in 12 of the ADCs it has developed [103]. The mc and PABC spacers either side of the vc linker provides room for efficient proteolysis by lysosomal protease, cathepsin B and because PABC undergoes self-immolation, MMAE is released in its free (and most potent) form [103]. PEG spacers have been added to linkers to increase their solubility, thus reducing ADC aggregation and plasma clearance and improving drug loading [106]. In addition, the inclusion of PEG spacers in ADCs containing maytansinoid payloads was shown to improve the potency of these ADCs in multi-drug resistant cancer cell lines possibly by decreasing the interaction of drug metabolites with drug efflux pumps [103].

The sites and levels of drug conjugation are also important factors in ADC design. One of the issues which ADC developers have faced is heterogeneity in the distribution of payload on the antibody and in the quantities of drug loaded per antibody. An excessive drug to antibody ratio can lead to ADC aggregation and plasma clearance, whereas ADCs containing too little drug could compete with optimally loaded species for antigen binding [107]. Furthermore, payloads that are sub-optimally distributed can interfere with the interaction of an ADC with its target antigen [107]. Originally, conjugation to the ɛ-amines of antibody lysine residues, or to the sulfhydryl groups of cysteine residues were the drug–antibody loading methods of choice. The large number of lysine residues in an antibody (a typical IgG1 antibody contains approx. 100 lysine residues of which approx. 40 are modifiable [108]) means batch-to-batch variability in antibody–drug conjugation reactions is inevitable [107]. ADC heterogeneity can be greatly reduced when conjugating payloads to cysteine residues where controlled reducing conditions lead to the selective exposure of just the eight sulfhydryl groups of the four inter-chain disulfide bridges. However, this is still far from the ideal scenario of producing batches of ADC containing only a single, optimized species. To address the issue of ADC heterogeneity, several technologies have been developed to facilitate site-specific drug conjugation. The production of antibodies in which two engineered cysteines (one per antibody heavy chains) were inserted, termed THIOMABs, have enabled selective payload conjugation through a carefully controlled reduction and re-oxidation protocol [109]. In a comparison study, the THIOMAB drug conjugate equivalent of an anti-MUC16-MMAE ADC proved as effective as the ADC, but caused fewer adverse effects in rats and cynomolgus monkeys with implanted ovarian cancer xenografts [109]. Other techniques for controlled conjugation involve engineering amino acid tags at defined locations on the antibody. The tags are post-translationally modified to carry reactive groups for site-specific drug conjugation [110]. Non-natural amino acids, such as para-azido-l-phenylalanine and selenocysteine have also been engineered into antibodies to provide sites with unique conjugation chemistries allowing site-specific drug conjugation that leaves native amino acids untouched [111].

5.4. Antibody–drug conjugates in clinical development for the treatment of gynaecological cancer

BMS-986148 (Bristol-Myers Squibb) and BAY 94-9343 (Bayer Healthcare AG) are two ADCs targeting mesothelin, a cell surface glycoprotein expressed in the mesothelial cells that line the pleura, peritoneum, pericardium and surface epithelial cells of the ovary [112]. It is highly expressed in numerous malignancies including ovarian cancer and uterine adenocarcinoma, where it has been associated with poor prognosis [112]. The function of mesothelin is not fully understood, but studies in ovarian cancer cell lines have shown it to bind MUC16 (CA125) and it has been suggested that this interaction facilitates peritoneal metastasis of ovarian tumours [113]. BAY 94-9343 has a DM4 payload conjugated to the mAb via a reducible disulfide linker: SPDB (N-succinimidyl-4-(2-pyridyldithio) butanoate) linker. This combination of drug and linker was chosen for its bystander effect as a means of targeting tumour cells not expressing mesothelin, while the selective toxicity of DM4 in proliferating cells is thought to limit its off-target effects [41]. BAY 94-9343 is undergoing phase I trials for treatment of various cancers including platinum resistant ovarian cancer, both as a solitary agent and in combination with other chemotherapeutics.

Lifastuzumab vedotin (RG-7599 or DNIB0600A) is an ADC targeting NaPi2b (type II sodium-phosphate cotransporter), a transmembrane sodium-dependent phosphate transporter involved in maintaining local phosphate concentrations [114]. NaPi2b expression has been identified in tumours of the thyroid, lung, breast and ovaries, but also in normal tissues of the lung and breast raising the concern that an anti-NaPi2b could produce dangerous off-target toxicities [115]. However, tests in rats and cynomolgus monkeys showed the main off-target toxicities (bone marrow, liver and testicles) were those typically associated with MMAE and did not correspond to damage in tissues expressing NaPi2b [42]. Lifastuzumab vedotin is currently in phase I and II trials for the treatment of NSCLC, ovarian, fallopian tube and peritoneal cancers.

The folate receptor α (FRα) mediates the endocytotic uptake of folate, a metabolite, which through its roles in amino acid, DNA and RNA metabolism as well as in methylation reactions, is essential for cell growth and proliferation [116]. FRα is highly expressed in numerous tumours including ovarian, NSCLC, kidney, endometrial, colorectal and breast cancers making it a promising target for ADC design [117]. IMGN853 (Mirvetuximab Soravtansine) is an anti-FRα ADC with a DM4 payload being developed by Immunogen and Merck for treatment of FRα-expressing tumours. Preclinical studies using ovarian cancer xenografts in mice have shown IMGN853 to have potent antitumour effects both as a single agent and in combination with bevacizumab (a VEGF antagonist) and carboplatin [43,44]. The ADC was shown to illicit a bystander effect in xenograft tumours suggesting it can be effective in tumours with heterogeneous FRα expression [44]. Phase I trials with IMGN853 are currently being conducted in ovarian cancer patients with phase II trials recruiting patients with ovarian, endometrial, fallopian tube and primary peritoneal cancers.

SAR566658 is an ADC being developed in a collaboration between Immunogen and Sanofi that targets a tumour-associated sialoglycotope on MUC1, known as CA6 and delivering a DM4 payload [45]. CA6 is expressed in ovarian, breast, cervical, lung and pancreatic cancers [118]. SAR566658 is still in early clinical development, but indications of an ongoing phase I trial show the ADC to be generally well tolerated with evidence for clinical benefit in breast and ovarian cancer patients [46].

PF-06647263 (Pfizer), an ADC targeting EFNA4, entered clinical trials in 2014 with recruitment for a phase I study with triple negative breast cancer and ovarian cancer. It is unique among the ADCs discussed in that it has been developed specifically to target tumour initiating cells (TIC), a tumour subpopulation hypothesized to be responsible for tumour development and metastasis. A fascinating study by Damelin and colleagues describes the approach they took to identify EFNA4, a cell surface ligand of the Ephrin receptors, as a candidate for the targeting of TICs in ovarian and triple negative breast cancers (TNBC) [47]. In brief, TICs were identified in ovarian and TNBC patient derived xenografts (PDX) by sorting cells based on prospective TIC markers and then evaluating these subpopulations for TIC characteristics; namely the ability to form heterogeneous tumours and be serially transplanted. The group identified E-Cadherin (CD324) as such a TIC marker and through transcriptome analysis singled out EFNA4 as a cell surface antigen preferentially expressed in TNBC tumours versus normal tissue. Enzyme linked immunosorbent assays (ELISAs) confirmed that differences in the levels of EFNA4 mRNA corresponded to those of the protein. The ensuing ADC, PF-06647263 comprises an anti-EFNA4 IgG1 mAb conjugated to an AcButDMH-N-Ac-calicheamicin-g1 linker-payload. As TICs are associated with low proliferative rate, the choice of calicheamicin reflects its ability to kill cells regardless of cell-cycle status. PF-06647263 has demonstrated potent antitumour effects both in vitro and in vivo using ovarian PDX-implanted mice with clinical trial results highly anticipated.

Tisotumab vedotin is an MMAE bearing ADC conjugated to an anti-tissue factor (TF) mAb via a protease cleavable linker. TF is a transmembrane glycoprotein expressed predominantly on cells of the sub-endothelial wall of blood vessels, where it initiates coagulation upon exposure to blood at times of injury. TF is reportedly expressed in all types of cancer where it can be involved in angiogenesis, haematogenous metastasis, tumour growth and systemically increase the risk of thrombosis [119]. It represents an attractive target for the treatment of numerous cancers including gynaecological malignancies. Phases I and II trials for tisotumab vedotin are currently underway for patients with numerous solid tumours, including ovarian, endometrial, cervical, prostate and bladder cancers. Preliminary trial data show the drug to be efficacious at tolerable doses [48].

6. Outlook

Gynaecological cancers are second only to breast cancer as a cause of cancer-related mortality in women. The advent of HPV vaccines is expected to have a major impact in decreasing new incidences of cervical, vaginal and vulvar cancers. However, survival rates among patients diagnosed with ovarian cancer and high-grade uterine cancer subtypes have improved very little over the last 40–50 years, highlighting the need for novel therapies. Nanomedicines offer a new strategy in cancer treatment, in which tumour targeting and drug pharmacokinetics are placed at the forefront of a design-based approach to drug development. Three nanomedicines, Abraxane, Doxil and Myocet, are currently in use for the treatment of gynaecological cancers and the large number of nanomedicines in clinical development suggests this number is set to grow.

ADCs are already showing real promise in cancer treatment and represent the largest class of nanomedicines currently undergoing clinical trials for the treatment of gynaecological cancers. Kadcyla appears set to overtake Herceptin for treatment of HER2-positive breast cancers, and ADCs targeting HER2 along with other novel targets including receptor for advanced glycation endproducts (RAGE) are now being evaluated for their use in the treatment of gynaecological cancers in our laboratory. To enhance the delivery of ADCs to the clinic, a rational approach to their design is required. To this end, it is now possible to select all ADCable targets from the virtual human proteome, screen in silico for differential overexpression in cancers, and rapidly test for internalization efficiency before synthesizing the conjugate. In vitro evaluation of antibody libraries, either generated classically via hybridomas, or through phage display approaches to determine binding kinetics to native targets is now possible using high-throughput surface plasmon resonance techniques. Once optimized antibodies are selected for conjugation, in vitro and in vivo assays can quickly be undertaken. Through this approach, we have identified thousands of potential ADC targets in human cancers, and using the pipeline set out in figure 2, are now expediting the developmental ADC pipeline for gynaecological cancer treatment.

Author contributions

D.H. and J.G.-P.: acquisition of data; D.H., J.G.-P., G.D.H. and C.A.; analysis and interpretation of data; D.H. J.G.-P., G.D.H., L.W.F., D.G., L.M. and R.S.C.: revising the manuscript critically for important intellectual content; D.H. and R.S.C.: drafting the article. All authors made substantial contributions to conception and design of the review, and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All authors gave final approval of the manuscript to be published.

Competing interests

We declare we have no competing interests.

Funding

This work was funded by the Life Science Research Network Wales, an initiative funded through the Welsh Government's Ser Cymru programme and Tenovus Cancer Care (grant no. PhD2015/L35).