Peptide–drug conjugates (PDCs) present a smaller, synthetically tractable intermediate between cytotoxics and ADCs in lung cancer. Pre-clinically, highly potent T/NIR tumour delivery of tubulin or topoisomerase payloads at sub-toxic systemic exposures has been demonstrated by targeting integrins, somatostatin receptors or mutant p53, but fast renal loss, linker instability and antigen heterogeneity have limited efficacy to date. Current efforts to optimise PDC pharmacokinetics and biodistribution are using non-natural amino acids, stapled backbones or lipid tails to extend circulation while maintaining receptor uptake, and companion imaging to enrich for receptor-positive patients. If the hurdles of scalable, GMP-compatible production and immune-safe excipients can be solved, PDCs could provide a precision therapy option in oncogene negative lung cancers.
Non-small cell lung cancer (NSCLC) is the leading cause of cancer death globally as most patients present with disease that has already disseminated. Chemotherapy, while being the only treatment that can be applied to most patients, has not improved over the past decades. Dose intensification of platinum-based chemotherapy is not possible due to associated side-effects and dose reduction leads to early resistance. Targeted therapies are often limited by development of resistance and may not have activity across different mutations. In addition, many of these agents may have limited penetration into tumor tissues due to hypoxia, acidic pH, physical barriers like dense stroma, and immunosuppressive microenvironment. For example, small antibodies like Antibody-drug conjugates (ADCs) also face the issue of heterogenous antigen distribution, low binding affinities and recycling of Fc-receptors in target cells. Further, the molecular size of ADCs makes them less able to extravasate through the tumor microenvironment. The result is that current therapies have limited efficacy and most patients with metastatic disease have a median progression-free survival of only months. As such, there is a critical need for an improved drug delivery vehicle with high tumor penetration, molecular specificity, minimal immunogenicity, and the flexibility to exchange payloads. PDCs have shown the potential to address each of these requirements, while also being synthetically accessible by solid-phase peptide synthesis, and are regulated as small molecules rather than biologics.
An effective therapeutic index for pulmonary neoplasia requires that a targeting peptide both possess sufficient selectivity for the diseased over the normal parenchyma and deliver its payload to traverse the plasma membrane prior to systemic clearance. Peptides inspired by natural protein interactions, such as cyclic RGD or NGR, leverage the increased expression of integrin αvβ3 and aminopeptidase CD13 on newly-formed blood vessels and metastatic pulmonary cells, respectively. In addition to natural or synthetic sequences, conformational restriction of targeting peptides using lactam or disulfide bonds also enhances receptor binding affinity by reducing conformational entropy while also increasing resistance to proteolysis from circulating elastases, which are more abundant in smokers. In addition to passive targeting, newer generations of peptide-drug conjugates also utilize pH-responsive histidine or membrane-penetrating tryptophan tags to escape endosomes following cellular uptake; this dual targeting helps to mitigate off-target cytotoxicity in non-expressing pneumocytes. Structure–activity studies have also recently shown that N-methylation on certain backbone residues can alter the peptide conformation within the integrin binding pocket to transform partial agonists to full antagonists, thereby preventing downstream survival signaling that could negate the cytotoxic payload. In the case of lesions obscured by fibrotic stroma, penetratin or tat-derived basic peptides have been appended to the peptide terminus, creating bifunctional chimeras that transiently open tight junctions without permanently damaging the epithelium. In fact, selectivity is not entirely dependent on the affinity of the ligand–receptor interaction; the biophysical state of the conjugate also plays a key role in influencing peptide selectivity. Self-assembling, supramolecular nanofibrils generated by hydrophobic drug–peptide conjugates exhibit avidity effects that localize the conjugate within the tumor interstitium while slowly releasing the active drug via ester hydrolysis, thereby converting a systemic bolus dose into an intratumoral depot. These examples highlight that selectivity and penetration are not mutually exclusive parameters and that both are tunable variables that require iterative optimization between computational docking studies, organoid penetration assays, and orthotopic lung xenograft imaging rather than single-parameter screening.
Fig. 1 Structure of tumor-targeting drug conjugates.1,5
High therapeutic indices in lung cancer can only be achieved if the targeting peptide can discriminate between tumor and non-tumor tissue and deliver its payload across cell membranes before elimination by the systemic circulation. Cell penetrating peptides like cyclic RGD or NGR motifs leverage overexpression of integrin αvβ3 or aminopeptidase CD13 in neo-angiogenic endothelium and tumor cells, respectively. Stapling by lactam or disulfide bonds further increases affinity by reducing the entropic penalty upon binding and confers resistance to elastase proteolysis in the systemic circulation, which is elevated in the lungs of smokers. Second generation systems also include histidine residues which undergo a charge reversal at endosomal pH, or tryptophan residues which directly insert into the membrane to trigger translocation out of endosomes following cellular uptake; this two-punch approach also confers selectivity against non-targeted pneumocytes which do not express the receptor. Structure–activity studies have also shown that N-methylation of select peptide backbone residues can switch the binding pose of peptides from inside to outside the binding groove, thus changing peptide partial agonists to antagonists and blocking receptor-mediated cell survival signaling which would otherwise negate the intended effect of the cytotoxic payload. To reach tumors with a fibrotic stroma, cationic peptides from penetratin or tat have been conjugated to the distal end of the targeting peptide to form dual-function chimeras that transiently disrupt tight junctions without causing irreversible epithelial damage. It should also be noted that selectivity is not determined by ligand–receptor affinity alone; rather, the biophysical state of the conjugate also plays a role. Self-assembling nanofibrils formed by hydrophobic drug–peptide conjugates avidly bind to the tumor stroma and release the active drug through ester hydrolysis, effectively turning a systemic dose into an intratumoral depot. In summary, selectivity and penetration are co-dependent parameters that are optimized not through one-dimensional screens but rather an iterative process that includes computational docking, organoid penetration, and in vivo imaging in orthotopic lung xenografts.
The peptide–payload linker is the critical determinant of either whether the cytotoxic compound will remain cloaked in a quiescent state during delivery or become prematurely exposed in the bloodstream, consequently diminishing the therapeutic index carefully imprinted by the peptide targeting sequence. In the tumor microenvironment of the lung, this linkage must also be stable against the combined activity of numerous cathepsins, matrix metalloproteinases, and oxidative stress that is considerably more robust than in most other organs. Previous generations of these linkers based on ester linkages were easy to synthesize but were prone to nonspecific hydrolysis by ubiquitous esterases. This was resolved by replacing esters with valine–alanine or valine–citrulline dipeptide sequences cleaved by lysosomal cathepsin B following clathrin-mediated endocytosis. To avoid the residual negative charge that might affect drug efficacy, a self-immolative para-aminobenzyloxycarbonyl spacer was designed to bridge the dipeptide and cytotoxic payload. Cleavage of the spacer by cathepsin B results in spontaneous 1,6-elimination, effectively ejecting the unmodified payload out of the complex. Linkers containing sulfur were also shown to be vulnerable to thiol-disulfide exchange reactions, so they were replaced by ether or carbamate linkages that were stable to reduction, yet could be hydrolyzed by acid in late endosomes (pH < 5.5). A fourth environmental stimulus is hypoxia-activated azoreductase substrates: azo-linked SMANCS is stable in the oxygen-rich environment of alveolar capillaries, yet becomes reductively cleaved by azoreductase when it reaches the oxygen-poor hypoxic center of the tumor, allowing spatially regulated drug release to follow metabolic tumor heterogeneity. Finally, the linker should also not be so long as to interfere sterically with the peptide binding site. Excessively short linkers will have less conformational entropy and lower receptor binding affinities, while excessively long ones are more susceptible to nonspecific cleavage. PK studies in cigarette-smoke-treated mice have shown that an alkyl linker three carbons in length, and capped on either end with a cyclobutane to limit rotational and conformational freedom, is long enough to preserve plasma integrity for at least 4 h but short enough to guarantee full payload release in 1 h of internalization.
A decade ago the experimental pipeline for lung-targeted PDCs consisted of a series of small single-center phase I trials largely based on proof-of-concept in preclinical models. Over the last five years, however, this has shifted to a growing number of prospective early-stage programs, now falling within a common structural framework, and benefitting from shared learnings. The most visible of these is the shift away from single-agent cytotoxic payloads and towards immune-modifying or epigenetic warheads whose efficacy is not unmasked until after tumor-selective cleavage. With the refined emphasis on metabolism- and tumor-type specific targeting, eligibility criteria have evolved: from an all-comers metastatic cancer profile to a defined surface aminopeptidase or integrin biomarker status established by companion positron-emitting peptides pre-screening patients and essentially turning exploratory cohorts into in vivo pharmacodynamics platforms. On the regulatory front, we have also seen a move away from the requirement for comprehensive antibody-scale safety packages. However, regulatory comfort remains predicated on the rigorous characterization of linker metabolites within bronchoalveolar lavage fluid, and this requirement has driven the development of point-of-care microdialysis labs in phase I settings. Other operational best practices are being shared through an international consortium, now harmonizing infusional schedules, cannula dwell-times and steroid pre-medication policies and, in effect, establishing a master protocol approach that will speed cross-trial comparisons. So far none of the lung-directed PDCs have been considered for registrational evaluation but recent increases in the size of patient cohorts being recruited, a shift from tertiary to community oncology hubs and, most importantly, the first wave of publications reporting phase I data beyond mere safety dose-expansion studies suggest we are nearing the inflection point at which randomized data is required.
Current phase I/II trials employ a standard adaptive Bayesian framework, allowing the probability of toxicity to be updated after each patient has completed cycles 1 and 2. The majority of open protocols will now investigate two separate axes: vascular-targeted constructs that deliver tubulin-disrupting payloads to areas of neovascularization, and epitope-directed chimeras that translocate transcriptional modulators into cells expressing lineage-survival antigens. Dose-escalation will be performed using a staggered weekly infusion, instead of a day-1 bolus, in an attempt to leverage pulmonary endothelial recycling and blunt the cytokine response. Cohorts are being separated by prior immune checkpoint inhibitor use, because pre-existing T-cell infiltrates have been found to enhance the inflammatory signature of specific peptide linkers, which can result in a self-limiting pneumonitis being mistaken for dose-limiting toxicity. Pharmacokinetic data collection is also more robust: arterial lines are placed for cycle 1 to obtain early pulmonary extraction, and after the initial cycle dried-blood-spot microsampling is offered to patients during subsequent cycles to minimize inconvenience. Response assessment has also evolved beyond simple RECIST: hyperpolarized xenon magnetic resonance is being utilized in a subset of patients to measure regional ventilation-perfusion mismatch shortly after infusion, in order to obtain a functional read-out that often precedes anatomical reduction by several weeks. Biomarker discovery is being performed prospectively: single-cell RNA sequencing of brush biopsies taken from tumor-containing airways is being correlated against peripheral T-cell clonotypes, to allow researchers to determine whether peptide-mediated delivery can reshape the local immune landscape. Preliminary data has shown that conjugates with histone-targeting payloads drive a transcriptomic change that is reminiscent of viral mimicry. This has resulted in the parallel opening of expansion cohorts where low-dose interferon is added as a transcriptional primer. Crossover is allowed at the point of radiological progression, but tissue acquisition is still required, providing a rolling biorepository that will link molecular evolution to clinical outcome.
The new dogma is that peptide-targeted toxins are not intended to supplant existing standards of care but rather to act as enzymatic boosters which prime and reduce the threshold for the action of concurrently administered therapies. The most common strategy interlaces PDCs with vascular normalization and immune checkpoint inhibition with the expectation that transient pruning of immature tumor vasculature increases the local E:T ratio without risk of systemic hypoxia. The infusion schedule is intentionally staggered: the conjugate is given one day before the antibody in order to synchronize peak endothelial apoptosis with the pharmacodynamic peak of checkpoint saturation. Stereotactic body radiotherapy is also being integrated in 3-therapy regimens and may be given either as a 3-fraction mid-course boost or as a single high dose session in the second week of peptide treatment. Preclinical imaging data suggest that local irradiation up-regulates aminopeptidase N in the peri-necrotic rim, thus expanding the targetable tumor burden for the peptide vector. Chemotherapy backbones have not been foregone but rather re-positioned as metronomic primers. Etoposide or gemcitabine are given at low-dose every other week, a schedule which seems to blunt myelosuppression and enhance peptide uptake as a result of cell cycle synchrony. To reduce the risk of overlapping pneumotoxicity, prophylactic inhaled steroids are initiated one week before the first concomitant cycle and only tapered after evidence of radiographic stability is documented. Early translational endpoints include spatial transcriptomics of serial biopsies to determine whether payload induced immunogenic cell death expands the clonal T-cell repertoire recognizable by the concomitant checkpoint inhibitor. Adaptive randomization algorithms are used to assign the next cohort to the arm with the most promise to date, so that the combinatorial space is explored with statistical efficiency rather than empirical exuberance. Although mature survival data are still awaited, the repeated observation of abscopal responses outside the irradiated lobe and the low rate of grade-three pneumonitis have encouraged the investigators to open a seamless phase II basket in which not only NSCLC and SCLC but even high-grade pulmonary neuro-endocrine carcinomas are being treated according to a unified master protocol.
Fig. 2 Theoretical summary of the steps that lead to the internalization of the specific (overexpressed receptor) within cancer cells.2,5
In spite of the conceptual attractiveness of peptide–drug conjugates, their clinical application in lung cancer is often redirected or frustrated by a triad of important challenges with biological, chemical and pharmacokinetic/regulatory dimensions. First, the highly heterogeneous nature of lung tumors in terms of antigen density, stromal context and immune landscape can outpace efforts of medicinal chemistry and their rearrangement. Second, the selective pressure against tumor cells also creates more robust transcriptional and epigenetic escape mechanisms, so that each individual peptide–payload combination is almost preprogrammed to fail eventually. Finally, the lung parenchyma is extremely vulnerable to off-target effects; pneumonitis, broncho-vasculitis and cytokine-release syndromes have all been observed at doses well below the anticipated therapeutic window, and clinicians are often forced to make a difficult choice between cytoreduction and preservation of gas exchange. These and other challenges related to safety, efficacy and progression have coalesced into a translational bottleneck, where promising signals seen pre-clinically have often failed to reproduce in early clinical development, not due to a lack of activity, but because the target (the tumor) is a moving target that re-negotiates its existence well before adaptive trial designs can keep up.
It is important to emphasise that most non-small cell lung cancers are not homogeneous entities but spatially organized populations of genetically distinct sub-clones. This can result in dynamic paracrine interactions and spatial competition between sub-clones. In addition to the spatial heterogeneity, intra-tumor temporal heterogeneity is also common: for example, a primary tumor targeted at diagnosis with a peptide may already harbor small clusters of cells in other parts of the tumor, which do not express the peptide receptor and therefore are not affected by the targeting peptide. Likewise, pre-existing metastases may have different receptor status on different parts of the same tumor. Ongoing receptor down-regulation can be mediated through hypoxia or inflammation. This is also a clinical observation; in current clinical practice, only one biopsy sample from a tumor is obtained. These are usually image guided biopsies and are limited to the most accessible and largest quadrant of the tumor, but may not capture the chemo-naïve resistant sub-clone(s) that are otherwise present, and therefore not identified for target selection. In fact, even if a peptide target is present in a specific patient tumor, epigenetic regulation of that target may switch it off without any change in the DNA sequence, and tumor heterogeneity is therefore not adequately predicted by the genome sequence. Recent single cell phospho-proteomics have shown that the phosphorylation status may be different in different regions of a single tumor, meaning that a peptide designed on the basis of a signalling snapshot in one area of a tumor, may have a different phosphorylation topology in a different region, a few millimeters away. Tumor stromal content is also a major confounding factor; for example, collagen- or fibrosis-rich desmoplastic areas may steric-block peptide penetration, whilst very vascular areas allow rapid systemic wash-out, leading to a spatial pharmacokinetic profile that alternates between sub-therapeutic and supra-toxic levels. The effect of clonal competition, whereby sensitive cells may protect resistant cells from cytotoxic killing via, for example, lactate shuttling or glutamine donation from the former to the latter, also limits cytotoxic release by the conjugate. Some of these issues have been partially addressed by using cocktails of multi-receptor peptides, or by sequential antigen switching, although as the number of ligands on a construct increases, so does the risk of immunogenicity and manufacturing batch variability.
Resistance to peptide-targeting cytotoxins is extremely rare from a single point mutation; resistance develops as an orchestrated event that considers genomic, epigenomic, and microenvironmental signals in multi-step fashion. The most common mechanism is lysosomal sequestration, the upregulation of Rab7 trafficking which redirects the conjugate into recycling endosomes instead of proteolytic vesicles, causing it to be excreted before it can reach its intracellular target. The upregulation of ABC-transporter family pumps through demethylation of the gene's promoter and subsequent reactivation then pushes the now excreted drug through the lysosomal membrane and out of the cell, where it is sequestered within extracellular vesicles and expelled into the microenvironment. Paracrine mechanisms can also help tumor cells: cancer-associated fibroblasts release a hyaluronan-rich extracellular matrix which binds to the positive charge of the peptide, keeping it near the stromal interface and out of the way of its receptor. Immune-editing mechanisms are also manipulated; continuous exposure to interferon-γ, either from host T-cells or via immune checkpoint inhibitors, can upregulate HLA-class-I, which incidentally internalizes the target peptide through the antigen-processing complex, removing it from the tumor surface. In metabolic reprogramming, a hyper-glycolytic phenotype may reduce the pH in endosomes too quickly, which causes the linker to be cleaved in the trafficking vesicle before it can go sufficiently deep into the cell for the drug to be effluxed out. Finally, stemness may allow adenocarcinoma cells to convert to a neuro-endocrine phenotype which silences the initial receptor target and becomes dependent on a separate signaling pathway. Each escape mechanism is likely to not be exclusive of the others, and resistance through one of these channels may simply shift tumor cells to an alternate path; it is therefore unlikely that effective treatment would be monotherapy, but would need to target lysosomal sequestration, epigenetic reprogramming, and stromal interactions all at once.
Normal lung is especially sensitive to off-target effects, as the air-exchange surface has no physiologic redundancy and the immune-surveillance role is hyper-reactive to perceived threats. On one hand, PD-1 complexes are relatively low-molecular-weight, and can still induce a dose-limiting interstitial pneumonitis, with a radiographic appearance consistent with organizing pneumonia, leading to chronic steroid use with secondary infections. This may be mediated by the alveolar macrophages, via Toll-like receptor stimulation by a small amount of residual dimethyl-sulfoxide, a carrier solvent that is often not removed from lyophilized preparations; the excipient, in other words, may be the toxic moiety and not the peptide itself. On the other hand, the peptide backbone can be attractive to the positively-charged regions of the lung endothelium, which are exposed in pulmonary capillaries. This non-specific binding concentrates the local dose, causing disruption of the endothelial lining and capillaritis that can cause haemoptysis. In many ways, rapid renal excretion of peptide cleavage products (a design feature thought to be advantageous) can paradoxically saturate the renal tubule re-absorption capacity and cause a Fanconi-like syndrome (e.g., aminoaciduria, hypophosphataemia) with downstream respiratory-muscle fatigue. Safety in the liver is good, but the same aminopeptidase that is targeted by these drugs is expressed in sinusoidal endothelial cells at very low levels, which may cause sinusoidal dilatation that is sub-clinical until a patient has post-chemotherapy infection and bleeding diathesis. Immunogenicity, though probably less of an issue with peptides versus antibodies, can still occur, for example if non-native D-amino acids or cyclisation bonds are recognized as non-self haptens. Anti-peptide IgE can result in an acute bronchospasm with re-exposure that is an absolute contraindication to further use, and would eliminate further use of this or closely related peptide-based therapies.
PDCs are on the verge of transitioning from the fringe of experimental oncology to the standard of care for thoracic cancers; however, their ultimate legacy is likely to be defined not by incremental improvements in potency but rather by the willingness of the scientific community to adopt an entirely new conceptual framework for their design. Over the coming decade, a paradigm shift from static, one-size-fits-all systems to dynamic frameworks capable of autonomously rewriting their targeting programs in real-time in response to tumor evolution is likely to occur. Simultaneously, the regulatory landscape will need to mature to establish corresponding benchmarks that value adaptive manufacturing, support micro-sampling pharmacokinetics, and accept temporary formulation changes during trials. Should these cultural and technical prerequisites align, lung PDCs may find a unique foothold as a precision cytotoxic platform that is nimble enough to pursue clonal heterogeneity without the financial or immunological costs associated with biological modification. Herein, we describe two such themes that are likely to feature prominently in the next research cycle.
Future lung-targeted constructs may contain molecular antennae that continuously probe local proteomic microenvironment and trigger or block their own cytotoxic function depending on the answer. One strategy includes the attachment of a chemically scrambled peptide linker to the framework core; when a certain level of an oncometabolite is detected, the linker is metabolically ligated to the metabolite through an engineered metabolic trap, which causes a rotamer change of the carrier to bring the release site into proximity with the endogenous proteases. Reversible in nature, the prodrug can be switched on and off many times within a single pass and can establish a kind of closed feedback system to further limit off-target exposure in antigen-positive, metabolically silent tissue. In a similar design, RNA aptamer switches in the form of stem–loop hairpins unfold in the slightly elevated temperature found in inflammatory microenvironment, thereby exposing a cryptic cysteine that partakes in disulfide exchange and releases the drug payload where leukocyte migration is highest. Crucially, these self-monitoring constructs do not need external energy input nor exogenous imaging contrast to work, they rather utilize the chemical energy inherent to the tumour itself to translate disease activity into the signal that triggers drug release. This might become particularly useful in the adjuvant space after surgery where occult disease is still biochemically active but physically undetectable. Regulators have indicated an openness to "sense-and-respond" therapeutics, if orthogonal safety measures are designed to avoid uncontrolled drug activation, which can be achieved by the incorporation of acid-sensitive quenchers that permanently inactivate the drug conjugate if the pH within the endosome falls below a physiological threshold. If these measures can be proven in large animals, self-monitoring PDCs may shift the risk–benefit paradigm of early stage lung cancer towards a single infusion that will stay pharmacologically silent until occult recurrence stirs its killing potential.
Intravenous delivery is operationally attractive, but it requires each peptide to circulate through the entire vascular system before reaching the lungs, at which point it has become diffusely diluted and has formed secondary interactions with other targets en route. One way around this is to revisit the long-dormant goal of direct airway delivery using smart nebulizers that produce monodisperse droplets which are small enough to penetrate the airways and large enough not to be immediately exhaled. The alveolar epithelium is only several hundred nanometers thick, so once deposited, conjugates can diffuse directly into the interstitial space, effectively creating their own systemic portal from the lung and side-stepping the first-pass kidneys. Challenges with this approach include that ultrasonic nebulization shear cyclic peptides, and propellant-driven atomisers can electrostatically cross-link particles, leading to plugging of the smaller airways. Excipients from the material sciences that are breathable have recently been identified that form a self-assembling protective film around the particles as they are aerosolized, shielding the peptide from shear degradation forces, and then dissolving in seconds after reaching the mucosal surface. After deposition, the conjugate also needs to cross the mucociliary escalator, which is accomplished by transiently loosening the tight junctions between ciliated cells using small amounts of thiolated sugars. This provides paracellular passage without causing long-term damage to the epithelial surface. The most exciting benefit of direct airway delivery is that it allows targeting of specific lobes of the lung, which is an option because you can impose a rotational magnetic field on the aerosol stream and direct inhaled particles to the anatomical quadrant with the most suspicious radiologic findings.
We specialize in developing PDCs for oncology, with a strong focus on lung cancer applications. By integrating expertise in tumor-targeting peptides, PEGylation, and linker stability, we support partners seeking to improve efficacy and safety in lung cancer treatment. Contact us to collaborate on advanced PDC oncology programs.
1. How can peptide drug conjugates help in lung cancer therapy?
PDCs target receptors overexpressed on lung cancer cells, delivering cytotoxic drugs directly to tumors. This improves precision and reduces damage to healthy lung tissue.
2. What is the mechanism of action of PDCs in lung tumors?
The peptide portion guides the conjugate to cancer cell receptors, after which the conjugate is absorbed. Inside the cell, the linker releases the drug payload to kill the cancer cell.
3. How do PDCs compare with existing lung cancer therapies?
Compared with chemotherapy and some targeted therapies, PDCs offer greater selectivity, potentially fewer systemic side effects, and improved efficacy against resistant tumors.
4. Are PDCs safe for lung cancer patients?
While still under investigation, early trials suggest that PDCs may have a better safety profile than conventional chemotherapy, though toxicity from drug payloads must be carefully managed.
5. Can PDCs be combined with immunotherapy or chemotherapy?
Yes, combination approaches are being studied. PDCs may enhance the effect of checkpoint inhibitors or serve as part of multi-drug regimens to overcome tumor resistance.
References
