Anti-Lung Cancer Agent Discovery and Development: A Review of Cross-Platform Correlation Across in silico, Network Pharmacology, in vitro, and in vivo Assessments

Abstract

Lung cancer remains one of the leading causes of cancer-related mortality worldwide, highlighting the urgent need for more effective and clinically translatable therapeutic agents. In recent years, drug discovery strategies integrating computational modeling with experimental validation have gained increasing attention; however, their predictive consistency across biological scales remains a critical challenge. This review systematically synthesises evidence from eleven primary studies that evaluated anti–lung cancer agents across computational, cellular, and organism-level platforms within a single developmental continuum. Eligible studies were required to report either in silico target prediction/molecular docking or network pharmacology analysis (or both), in direct conjunction with in vitro cytotoxic evaluation and in vivo efficacy assessment. Relevant studies were identified using a structured Boolean keyword strategy integrating computational, cellular, animal-based, and lung cancer–specific terms. Across the analyzed studies, computational and network approaches consistently converged on key oncogenic signaling hubs, particularly PI3K/AKT-centered pathways and receptor tyrosine kinases. These predictions were corroborated by cellular assays demonstrating selective antiproliferative activity in lung cancer cell lines and subsequently validated in animal models showing significant tumor growth suppression with acceptable safety profiles. Notably, agents optimized through targeted delivery systems, formulation strategies, or rational combination therapies exhibited the most consistent cross-platform performance. Collectively, the findings highlight that robust anti–lung cancer efficacy is most reliably achieved when molecular interaction data, systems-level network engagement, cellular responses, and in vivo outcomes are coherently integrated. This integrated evaluation framework provides a rational and predictive foundation for future anti–lung cancer drug discovery and development.

Graphical Abstract

Introduction

Lung cancer remains the leading cause of cancer-related mortality worldwide, accounting for a substantial proportion of global cancer deaths each year. Despite advances in screening, molecular diagnostics, and targeted therapeutics, the overall prognosis for lung cancer patients particularly those with advanced-stage disease remains poor. Non-small cell lung cancer (NSCLC), which constitutes most lung cancer cases, is characterized by pronounced molecular heterogeneity, complex oncogenic signaling networks, and dynamic tumor–microenvironment interactions that collectively complicate therapeutic intervention and long-term disease control.1–3 These features underscore the urgent need for more effective and rationally designed and developed anti–lung cancer agents that can achieve durable efficacy with acceptable safety profiles. Current therapeutic strategies for lung cancer, including chemotherapy, targeted therapy, and immunotherapy, have improved clinical outcomes in selected patient populations; however, their benefits are frequently limited by intrinsic or acquired resistance, dose-limiting toxicities, and interpatient variability in treatment response.4–6 Although molecularly targeted drugs and immune checkpoint inhibitors have transformed the treatment landscape, resistance mechanisms such as pathway reactivation, compensatory signaling, and phenotypic plasticity continue to drive disease progression and relapse. Consequently, the identification of novel agents or optimized therapeutic strategies that can effectively modulate key oncogenic pathways while maintaining translational relevance remains a central challenge in lung cancer drug development.

In this context, in silico methodologies encompassing molecular docking, molecular dynamics simulations, and related computational approaches have become indispensable tools in early-stage drug discovery. These techniques enable the rapid screening of compound libraries, the prediction of target–ligand interactions, and the preliminary assessment of binding stability and specificity.7–9 Complementing these approaches, network pharmacology has emerged as a powerful systems-level strategy to elucidate multitarget and pathway-oriented mechanisms, particularly for phytochemicals and complex formulations that exert pleiotropic biological activities.10–12 Nevertheless, computational predictions alone are inherently limited by model assumptions and cannot fully capture biological complexity without rigorous experimental validations. Experimental evaluation through in vitro and in vivo models remains essential to confirm biological activity, therapeutic selectivity, pharmacokinetic behavior, and safety. Cell-based assays provide critical insights into cytotoxicity, antiproliferative effects, and oncogenic pathway modulation in relevant lung cancer cell lines, while animal models allow assessment of tumor growth inhibition, biodistribution, toxicity, and overall therapeutic performance under physiological conditions.13–15 However, many studies report these experimental outcomes independently from their computational predictions, often without explicitly assessing whether molecular-level hypotheses translate coherently and consistently into cellular and organismal effects.

A critical gap in the current literature lies in the lack of integrative evaluation frameworks that systematically examine linear correlations and cross-platform coherence across in silico, network pharmacology, in vitro, and in vivo assessments. Although numerous reports claim promising anticancer activity based on isolated datasets, relatively few studies critically interrogate whether predicted target engagement and pathway modulation are consistently reflected in cellular phenotypes and ultimately in measurable antitumor efficacy in vivo.16–18 This fragmentation contributes to high attrition rates during drug development, as candidates with compelling single-platform outcomes often fail to demonstrate robust translational potential when subjected to more comprehensive evaluation. Addressing this gap represents a key opportunity to enhance the predictive reliability of anti–lung cancer drug discovery pipelines. Accordingly, this review aims to critically synthesize evidence from eleven primary studies that explicitly integrate in silico modeling, network pharmacology, in vitro cytotoxic evaluation, and in vivo efficacy assessment in the discovery and development of anti–lung cancer agents. By focusing on studies that span the full continuum from molecular-level prediction to organism-level validation, this review seeks to elucidate patterns of cross-platform and to highlight how coherent, multilevel alignment enhances translational relevance. Through this integrative perspective, the present work proposes a rational and predictive framework for future anti–lung cancer drug discovery grounded in cross-platform consistency, rather than isolated experimental success.

Methodology

This review was conducted using a structured and integrative literature analysis designed to identify anti–lung cancer agents that have been evaluated across in silico, network pharmacology, in vitro, and in vivo platforms within a single developmental continuum. Relevant studies were retrieved through a systematic database search employing Boolean logic that combined computational modeling terms, cell-based cytotoxicity and mechanistic assay indicators, animal efficacy and pharmacokinetic descriptors, and lung cancer–specific disease keywords, thereby ensuring comprehensive coverage of relevant studies that may not explicitly label their methodologies using standardized terminology. Articles were included only if they reported computational target prediction or network-based pathway analysis together with experimental validation in lung cancer cell lines and corresponding animal models, while studies limited to a single level of evaluation were excluded. From each eligible publication, data was systematically extracted and harmonized across four analytical domains: molecular targets and binding profiles derived from in silico or network pharmacology analyses, antiproliferative and mechanistic outcomes from in vitro assays, and antitumor efficacy, biodistribution, and toxicity findings from in vivo models. An integrative analytical framework was then applied to assess linear coherence and cross-platform concordance across these domains, allowing comparison of predicted molecular interactions and pathway modulation with observed cellular responses and organism-level therapeutic outcomes. This approach enabled systematic identification of convergent mechanistic patterns and high-potential anti–lung cancer candidates while acknowledging inherent heterogeneity in experimental models, dosing strategies, and biological endpoints across the included studies. Prisma flow diagram of the systematic review process for identifying multi-platform anti-lung cancer agents as shown in Figure 1.

Figure 1.

Prisma Flow Diagram of the Systematic Review Process.

Molecular Pathogenesis of Lung Cancer

Lung cancer is a highly heterogeneous malignancy characterized by complex genetic, epigenetic, and signaling dysregulations that collectively drive uncontrolled proliferation, resistance to apoptosis, angiogenesis, immune evasion, and metastatic dissemination. At the molecular level, dysregulation of receptor tyrosine kinases (RTKs) and downstream signaling cascades plays a central role in disease initiation and progression, with aberrant activation of pathways such as EGFR-driven PI3K/AKT/mTOR, MAPK, NF-κB, and STAT signaling consistently reported across multiple lung cancer subtypes.19–21 These pathways orchestrate fundamental oncogenic processes, including cell cycle progression, metabolic reprogramming, suppression of programmed cell death, and dynamic modulation of the tumor microenvironment.

In the context of non-small cell lung cancer (NSCLC), which accounts for most lung cancer cases, genetic alterations involving EGFR mutations, KRAS mutations, ALK rearrangements, and downstream effectors such as PIK3CA and AKT are particularly prominent and have become key therapeutic targets.22–24 In parallel, immune checkpoint regulation, notably through PD-1/PD-L1 signaling, has emerged as a critical determinant of tumor immune escape and therapeutic response.25 Angiogenic signaling mediated by VEGF and VEGFR further contributes to tumor growth and metastasis progression, reinforcing the multifactorial and interconnected nature of lung cancer pathogenesis.26

From a drug development perspective, these interconnected molecular events highlight why effective anti–lung cancer strategies increasingly require multitarget or pathway-oriented approaches rather than single-node inhibition. Contemporary research therefore frequently focuses on agents capable of simultaneously modulating proliferative signaling, apoptotic resistance, angiogenesis, and immune regulation, reflecting the biological complexity inherent to lung cancer progression and the limitations of mono-target therapies.27,28

Ultimately, the intricate interplay of these redundant and overlapping signaling cascades underscores a fundamental therapeutic challenge: suppressing a single molecular node often triggers compensatory bypass mechanisms that sustain tumor survival. This biological reality directly dictates the need to move beyond single-target paradigms toward comprehensive intervention strategies capable of dismantling interconnected oncogenic networks. Convergence of key oncogenic driven on downstream PIK3CA/AKT signaling in NSCLC as shown in Figure 2.

Figure 2.

Convergence of Key Oncogenic Driven on Downstream PIK3CA/AKT Signaling in NSCLC.

Drug Design Strategies for Anti–Lung Cancer Agents

Building upon this mechanistic understanding of network redundancy and pathway crosstalk, modern drug discovery has pivoted toward design frameworks capable of targeting multiple vulnerabilities simultaneously. Addressing the multifaceted pathogenesis of lung cancer requires innovative therapeutic architectures that can disrupt interconnected signaling hubs while optimizing drug delivery to the tumor site.

Traditional lung cancer drug discovery has historically emphasized single-target inhibitors, particularly those directed against dominant oncogenic drivers such as EGFR or ALK. While this strategy yielded clinically effective agents, therapeutic resistance, pathway redundancy, and tumor heterogeneity have substantially limited long-term efficacy.29,30 Consequently, modern anti–lung cancer drug design has progressively shifted toward integrative and system-oriented strategies that address network-level signaling interactions and adaptive tumor responses.

Within this framework, the rational design of anti–lung cancer agents increasingly incorporate polypharmacological profiles, targeted drug delivery systems, and pathway-oriented modulation to enhance therapeutic selectivity and durability.31,32 Phytochemicals, semi-synthetic derivatives, and nanocarrier-based formulations have gained considerable attention due to their capacity to simultaneously engage multiple molecular targets, improve bioavailability, and reduce systemic toxicity.33 These strategies are particularly relevant in lung cancer, where concurrent regulation of oncogenic signaling, apoptosis, inflammation, and angiogenesis is often required to achieve meaningful and sustained therapeutic outcomes.

Representative drug development efforts in lung cancer increasingly integrate computational prediction, cellular validation, and animal modeling to iteratively refine candidate agents. This paradigm supports a translational drug discovery pipeline in which molecular hypotheses are systematically tested and validated across progressively increasing levels of biological complexity.34

In silico Studies and Network Pharmacology for Anti–Lung Cancer Drug Discovery

In silico methodologies have become indispensable tools in contemporary anticancer drug discovery, offering efficient and cost-effective approaches to predict molecular interactions, binding affinities, and target specificity prior to experimental validation. Molecular docking and molecular dynamics simulations enable the identification of potential ligand–target interactions at atomic-level resolution, facilitating early-stage screening of candidate compounds against oncogenic proteins relevant to lung cancer, such as EGFR, AKT, PI3K, mTOR, and immune checkpoint associated molecules.35–37 Complementing these structure-based techniques, network pharmacology provides a systems-level approach integrating compound–target, protein–protein interaction, and pathway-enrichment analyses to capture multitarget mechanisms, providing a holistic framework to decipher complex biological networks.

Beyond single-target prediction, network pharmacology has emerged as a powerful systems-level strategy capable of capturing the multitarget and multipath way nature of complex diseases such as lung cancer. By integrating compound–target associations, protein–protein interaction networks, and pathway enrichment analyses, network pharmacology enables systematic elucidation of how bioactive compounds collectively modulate interconnected signaling cascades, including PI3K/AKT, MAPK, NF-κB, and cytokine-mediated pathways.38,39 This approach is particularly relevant for multi-component formulations and natural products, where therapeutic effects arise from the coordinated modulation of multiple molecular nodes, rather than isolated target inhibition.

In lung cancer drug discovery, integrated in silico and network-based analyses are commonly employed to prioritize therapeutic targets, predict synergistic or complementary mechanisms, and rationalize downstream experimental design. These computational frameworks provide a mechanistic foundation upon which subsequent cellular and animal studies are constructed, facilitating a more coherent and predictive translational research pipeline.40

In vitro Cytotoxic Evaluation of Anti–Lung Cancer Agents

In vitro evaluation remains a cornerstone of anticancer drug development, providing direct and quantifiable evidence of cytotoxicity, antiproliferative activity, and mechanistic engagement at the cellular level. Lung cancer cell lines such as A549, H1299, H1975, and H292 are widely employed due to their well-characterized genetic backgrounds and relevance to distinct molecular subtypes of lung cancer.41,42 Quantitative endpoints, including IC₅₀ values derived from MTT, MTS, or SRB assays, offer standardized and comparable measures of antiproliferative potency.

Beyond viability assessment, in vitro studies increasingly incorporate mechanistic assays to interrogate apoptosis induction, cell cycle arrest, angiogenesis inhibition, oxidative stress modulation, and pathway-specific protein expression.43 These cellular endpoints serve as functional downstream readouts of upstream molecular interactions predicted through in silico and network pharmacology analyses. Importantly, the inclusion of normal (non-malignant) cell lines enables preliminary evaluation of therapeutic selectivity, an essential consideration in lung cancer drug development given the narrow therapeutic window of many cytotoxic agents.

While in vitro systems cannot fully replicate tumor complexity, they provide a critical translational bridge by validating molecular hypotheses and narrowing candidate selection before progression to in vivo evaluation.44

In vivo Model Evaluation for Anti–Lung Cancer Efficacy

In vivo models represent the most integrative preclinical platform for evaluating anti–lung cancer agents, capturing pharmacokinetic behavior, biodistribution, therapeutic efficacy, and systemic toxicity within a living organism. Xenograft and syngeneic lung cancer models are widely utilized to assess tumor growth inhibition, survival outcomes, and treatment-related adverse effects.45 These models allow investigation of drug–tumor interactions in the context of tumor vascularization, immune components, and organ-specific distribution.

In addition to efficacy assessment, in vivo studies play a crucial role in validating targeted drug delivery strategies, such as ligand-decorated nanoparticles or carrier-mediated drug transport, which are increasingly explored to enhance lung tumor selectivity and reduce off-target toxicity.46 Histopathological analysis and biomarker evaluation further provide insight into treatment-induced changes at the tissue and organs level, complementing cellular and molecular findings.

Within the anti–lung cancer drug development continuum, in vivo evaluation represents the decisive translational step that determines whether molecular predictions and cellular responses translate into meaningful and reproducible therapeutic benefit, thereby informing progression towards clinical investigation.47,48 Summary of drug design strategies for anti–lung cancer agents as shown in Table 1.

Table 1.

Comparison of Single-Target versus Multitarget/Polypharmacological Drug Design Strategies for Anti–Lung Cancer Agents

Aspect	Single-Target Design	Multitarget / Polypharmacological Design
Design principle	Selective, potent modulation of a single dominant oncogenic protein or pathway (eg, EGFR, ALK).	Simultaneous modulation of multiple interconnected signaling nodes (eg, PI3K/AKT, MAPK, NF-κB, cytokine-mediated pathways) to address network redundancy.
Representative agent classes	Selective small-molecule kinase inhibitors (EGFR-TKIs, ALK-TKIs); target-specific monoclonal antibodies.	Multi-component phytochemicals and semi-synthetic derivatives; nanocarrier-based formulations; multitarget natural products; rationally designed drug combinations.
Rational basis	Based on a single, genetically/molecularly identified dominant oncogenic driver.	Based on protein–protein interaction (PPI) networks, network pharmacology, and pathway crosstalk.
Associated in silico approach	Molecular docking and molecular dynamics simulation against a single target protein to predict binding affinity.	Network pharmacology integrating compound–target associations, PPI networks, and pathway enrichment analysis to map multitarget mechanisms.
Advantages	High target selectivity; well-defined and readily validated mechanism of action; pharmacological profile more predictable and easier to optimize at early stages.	Potential to reduce resistance arising from pathway redundancy; more durable therapeutic effect; relevant for multi-component formulations and natural products; may lower systemic toxicity through synergistic effects at reduced doses.
Limitations	Susceptible to therapeutic resistance due to pathway redundancy and tumor heterogeneity; limited long-term efficacy.	Higher mechanistic complexity, making experimental validation more challenging; more demanding in silico prediction; potential off-target interactions and pharmacokinetic optimization challenges for multi-component systems.
Relevance to tumor heterogeneity	Less optimal for tumors with high genetic heterogeneity or adaptive resistance.	Better suited to address tumor heterogeneity and adaptive responses by modulating multiple molecular nodes concurrently.
Translational validation stages	In silico docking/MD → in vitro cytotoxicity assays (eg, IC50 in A549, H1299, H1975, H292) → in vivo xenograft/syngeneic models.	In silico network pharmacology and multitarget docking → in vitro mechanistic assays (apoptosis, cell cycle arrest, oxidative stress) → in vivo validation, including targeted delivery systems (eg, nanoparticles).

Results

In silico and Network Pharmacology Evaluation

This subsection summarizes the computational and network-based evaluations of candidate anti–lung cancer agents, encompassing molecular docking analyses and systems-level network pharmacology approaches. The data collectively describe predicted binding interactions with key lung cancer–associated molecular targets, including receptor tyrosine kinases, immune checkpoints, apoptosis- and survival-related proteins, and carrier–receptor interfaces relevant to targeted drug delivery.

Molecular docking analyses revealed substantial variability in binding affinities among the evaluated compounds. Among all candidates, lonchocarpin, a natural chalcone-derived compound, exhibited the mist favorable docking performance, achieving an exceptionally low binding energy of −41.36 kcal/mol against the target structure 4LVT. This interaction involved extensive contacts with multiple amino acid residues, including PHE101, PHE109, MET112, VAL130, LEU134, ALA146, PHE150, VAL153, and ARG143.49 High-affinity binding was also observed for cyclomorusin, isolated from Morus alba cortex mori, which showed a docking score of −11.6 kcal/mol towards AKT1 (6CCY).50

Consistently strong binding within the PI3K/AKT/mTOR signaling axis was demonstrated by sakuranin, which exhibited its most favorable interaction with AKT (6HHG, −10.5 kcal/mol), followed by PI3K (1E8X, −9.2 kcal/mol), mTOR (3JBZ, −8.7 kcal/mol), and ERK (5NGU, −8.1 kcal/mol).51 These results indicate robust and multi-nodal engagement across a central oncogenic signaling cascade, supporting the potential of sakuranin as a pathway-oriented therapeutic candidate.

Ligand–receptor targeting studies further highlighted the performance of carrier-based delivery systems. Transferrin (Tf) ligand displayed strong affinity toward the transferrin receptor, particularly at 1E7U site 1 (−9.8 kcal/mol) and 1E8W site 1 (−9.0 kcal/mol), with numerous stabilizing interactions involving hydrophobic and charged amino acid residues.52 Similarly, the chitosan–folate conjugate (CHI-FA) demonstrated favorable binding to folate receptor-α (FR- α; PDB ID: 4KM6) with a docking score of −7.6 kcal/mol, engaging multiple lysine, asparagine, serine, and valine residues.53

Docking analyses of phytochemical constituents from Bauhinia acuminata L,54 demonstrated diverse and target-specific binding profiles across several lung cancer–related targets. The strongest interactions within this group were observed for deoxycytidine, which exhibited high affinity towards EML4 (4CGC, −8.6 kcal/mol) and RET (2IVT, −7.8 kcal/mol), as well as rhoeagenine, which bound strongly to EGFR (2GS2, −8.1 kcal/mol). Additional notable interactions included (−)-caryophyllene oxide with EML4 (−8.1 kcal/mol), 9,12-octadecadienoic acid with PD-L1 (−7.2 kcal/mol) and EML4 (−7.1 kcal/mol), and 4-((1E)-3-hydroxy-1-propenyl)-2-methoxyphenol with PD-L1 (−7.0 kcal/mol), the latter forming the highest number of hydrogen bonds within this subgroup. For Artemisia judaica L, flavonoid and phenolic constituents demonstrated defined residue-level interactions with CDK-2 (2A4L) and EGFR (1M17).55 In parallel, oleanolic acid exhibited stable binding to bovine serum albumin (4JK4) with a docking score of −8.2 kcal/mol, supported by interactions involving ILE297, ARG336, HIS337, and surrounding hydrophobic residues.56 These favorable hydrophobic interactions and hydrogen bonding suggest effective carrier-binding potential and implications for enhanced pharmacokinetic stability. Curcumin, evaluated against neutrophil elastase (PDB ID: 1B0F), demonstrated measurable binding affinity, characterised by interactions with ARG178 and ARG217.57

Table 2 summarizes the predicted binding affinities, key interacting amino acid residues, and hydrogen bond interactions of phytochemicals, synthetic compounds, and ligand–carrier systems against major lung cancer–related molecular targets, including EGFR, ALK, RET, PD-L1, VEGFR, PI3K/AKT/mTOR signaling components, transferrin receptor, folate receptor-α, and carrier proteins. The data highlight compounds exhibiting the strongest binding performance across multiple oncogenic targets.

Table 2.

In silico Molecular Docking Analysis of Candidate Anti–Lung Cancer Agents

No.	Plant	API	In silico
			Target	Binding Energy (kcal/mol)	Amino Acid Residues	Number of Hydrogen Bond	Ref
1.	Bauhinia acuminata L	(-)-Caryophyllene oxide	2GS2, EGFR – Epidermal Growth Factor Receptor (Tyrosine Kinase)	−6.6	LYS 721	1	[54]
			4CGC, EML4 Echinoderm Microtubule-Associated Protein-Like 4	−8.1	N/A	2
			4TT7, ALK – Anaplastic Lymphoma Kinase	−6.8	N/A	1
			5NIU, PD-L1 Programmed Death Ligand-1	−5.9	N/A	2
		Rhoeagenine	2GS2, EGFR Epidermal Growth Factor Receptor (Tyrosine Kinase)	−8.1	THR 766, MET 769, THR 830	2
			2IVT, RET Receptor Tyrosine Kinase		LYS 758, SER 891
			2MWY, p53 tumor suppressor protein	−7.2	GLN 1, ASN 3	1
			4CGC, EML4 Echinoderm Microtubule-Associated Protein-Like 4	N/A	GLN 32, GLN 34, GLU 35, GLU 37
			1VPF, VEGFR Vascular Endothelial Growth Factor Receptor	−5.7	N/A	1
			4TT7, ALK Anaplastic Lymphoma Kinase	−5.9	MET 1199	2
			5NIU, PD-L1 Programmed Death Ligand-1	−6.3	VAL 130
		Deoxycytidine	2IVT, RET Receptor Tyrosine Kinase	−7.8	SER 891	1
			4CGC, EML4 Echinoderm Microtubule-Associated Protein-Like 4	−8.6	GLN 34
		R3,7,11,15-Tetramethyl-2- hexadecen-1-ol	2GS2, EGFR – Epidermal Growth Factor Receptor (Tyrosine Kinase)	−6.5	THR 766, GLN 767	2
			2IVT, RET – Receptor Tyrosine Kinase	−5.3	VAL 915, LYS 916	2
			2MWY, p53 tumor suppressor protein	−6.7	ARG 81	1
			1VPF, VEGFR – Vascular Endothelial Growth Factor Receptor	−5.6	GLY 59	1
			4TT7, ALK – Anaplastic Lymphoma Kinase	−5.8	ILE 1361, GLN 1364	3
		2-hexadecen-1-ol, 7,11,15- tetramethyl-[R-[R*,R*-(E)]]	2GS2, EGFR – Epidermal Growth Factor Receptor (Tyrosine Kinase)	−6.4	THR 766, GLN 767	2
			2MWY, p53 tumor suppressor protein	−6.0	GLN 1, TYR 77	2
			1VPF, VEGFR – Vascular Endothelial Growth Factor Receptor	5.3	TYR 25, CYS 104	2
			4TT7, ALK – Anaplastic Lymphoma Kinase	−6	ASN 1254, GLY 1269	2
			5NIU, PD-L1 Programmed Death Ligand-1	6.6	LYS 129, VAL 130, ASN 131	4
		−9,12-Octadecadienoic acid	2GS2, EGFR – Epidermal Growth Factor Receptor (Tyrosine Kinase)	−6.4	ASN 818, ASP 831	2
			2IVT, RET – Receptor Tyrosine Kinase	−5.9	SER 891	1
			4CGC, EML4 – Echinoderm Microtubule-Associated Protein-Like 4	−7.1	GLN 34, GLU 37, GLU 35	3
			4TT7, ALK – Anaplastic Lymphoma Kinase	−6.8	ARG1253	1
			5NIU, PD-L1 Programmed Death Ligand-1	−7.2	HIS 140, HIS 141	3
		−4-((1E)- 3-Hydroxy-1- propenyl)- 2-methoxyphenol	1VPF, VEGFR Vascular Endothelial Growth Factor Receptor	−5.2	GLU 30, GLY 59	4
			5NIU, PD-L1 Programmed Death Ligand-1	−7	ILE 116, ALA 121, ASP122	5
		Nonadecen-1-ol	2MWY (p53 tumor suppressor protein)	−6.5	ASN 3, LEU 87, ALA 88	3
			4CGC (EML4 Echinoderm Microtubule-Associated Protein-Like 4)	−4.4	ASP 19, VAL 20, GLN 21	3
			1VPF, VEGFR Vascular Endothelial Growth Factor Receptor	−5.8	GLU 64, LEU 66, LYS 107	3
			5NIU, PD-L1 Programmed Death Ligand-1	−6.6	LYS 129, VAL 130, ASN 131	4
		Z)-N-[2-(1-Fluoropropenyl) phenyl]propionamide	2GS2, EGFR Epidermal Growth Factor Receptor (Tyrosine Kinase)	−6.1	LEU 679,753, 754	3
			2MWY, p53 tumor suppressor protein	−5.7	ILE 2, GLU 128	2
			4CGC, EML4 Echinoderm Microtubule-Associated Protein-Like 4	−6.9	GLN 34, GLU 35, THR 39	4
			1VPF, VEGFR Vascular Endothelial Growth Factor Receptor	−5.7	SER 50	1
			4TT7, ALK Anaplastic Lymphoma Kinase	−5.9	MET 1199	2
			5NIU, PD-L1 Programmed Death Ligand-1	−6.3	VAL 130	1
		9,12–15 Octadecadienoic acid (Z,Z)	2IVT, RET – Receptor Tyrosine Kinase	7.2	GLY 733, ASN 789	2
			4TT7, ALK – Anaplastic Lymphoma Kinase	−6.7	ARG 1253, ASN 1254	12
		2-Benzenedicarboxylic acid, diethyl ester	2GS2, EGFR Epidermal Growth Factor Receptor (Tyrosine Kinase)	−5.6	THR 830	2
			2IVT, RET Receptor Tyrosine Kinase	−6.7	SER 891	2
			2MWY, p53 tumor suppressor protein	5.6	GLN 1, ASN 3, THR 86	4
			4CGC, EML4 Echinoderm Microtubule-Associated Protein-Like 4	6.5	GLN 34	1
			1VPF, VEGFR – Vascular Endothelial Growth Factor Receptor	−5.1	THR 31, LEU 32	2
			4TT7, ALK – Anaplastic Lymphoma Kinase	−5.6	MET 1199	2
		4-((1E)- 3-Hydroxy-1- propenyl)- 2-methoxyphenol	1VPF, VEGFR – Vascular Endothelial Growth Factor Receptor	5.2	GLU30, GLY 59	4
			5NIU, PD-L1 – Programmed Death Ligand-1	−7	ILE 116, ALA 121, ASP 122	5
2.	N/A	Curcumin	Neutrophil Elastase (1B0F)	27.4	ARG178, ARG217	N/A	[57]
3.	N/A	Transferrin (Tf-ligand)	1E7U (site 1) transferrin receptor	−9.8	ALA 885, LYS 808, LYS 833, LYS 807 MET 804, ILE 879, ILE 963, ILE 831, PRO 810, TRP 812, LEU 963, HIS 948, VAL 1091	3	[52]
			1E7U (site 2) transferrin receptor	−7.8
			1E8W (site 1) transferrin receptor	−9.0	ALA 885, LYS 883, MET 953, ILE 881, ILE 963, ILE 879, ILE 831, PRO 810, TRP 812, ILE 963, TYR 867	3
			1E8W (site 2) transferrin receptor	−7.2
4.	N/A	Sakuranin	1E8X (PI3K)	−9.2	N/A	N/A	[51]
			6HHG (AKT)	−10.5	N/A	N/A
			3JBZ (mTOR)	−8.7	N/A	N/A
			5NGU (ERK)	−8.1	N/A	N/A
5.	Morus alba L (cortex mori)	Cyclomorusin	6CCY (AKT1)	−11.6	N/A	N/A	[50]
6.	N/A	Chitosan-Folate (CHI-FA)	Folate receptor-α (FRα) 4KM6	−7.6	LYS 46, ASN 69, LYS 76, ASP 77 SER 67, LYS 41, VAL 78	N/A	[53]
7.	N/A	Lonchocarpin (natural chalcone)	4LVT (also validated on 1YSW, 2O2F, 2O21, 4AQ3, 4IEH)	−41.36	PHE 101, PHE 109, MET 112, VAL 130, LEU 134, ALA 146, PHE 150, VAL 153, ARG 143	N/A	[49]
8.	Artemesia judaica L	Flavonoid & phenolic compounds	CDK-2 (2A4L)	N/A	LEU 83, LYS 89	2	[55]
			EGFR (1M17)	N/A	MET 769	1
9.	N/A	Oleanolic acid	Bovine serum albumin (BSA) 4JK4	−8.2	ILE297, ARG336, HIS337, LEU304, PRO302, LEU301	2	[56]

Table 3 presents the network pharmacology outcomes of Qingjie Yifei Miao Fang and Jinfu’an Decoction, including compound identification strategies, compound–target network construction, predicted and intersecting disease targets, core hub genes, enriched KEGG pathways, and experimentally validated proteins. The results emphasize multi-target and pathway-level modulation relevant to lung cancer progression and metastasis.

Table 3.

Network Pharmacology Analysis of Multi-Component Anti–Lung Cancer Formulations

Aspect	Qingjie Yifei Miao Fang (QJYFMF).58	Jinfu’an Decoction (JFAD).59
Network pharmacology strategy	Compound-centric, database-driven network pharmacology	Serum compound-guided, target-oriented network pharmacology
Source of compounds	TCMSP database	UHPLC-QTOF-MS (drug-containing serum)
Compound screening criteria	Oral bioavailability (OB) ≥ 30%; drug-likeness (DL) ≥ 0.18	Serum-detected compounds only (no OB/DL filtering)
Number of identified compounds	98 active compounds	32 serum-detected compounds
Compound–target network construction	Conducted	Conducted
Compound ranking by network degree	Explicitly ranked and emphasized	Calculated but not a primary focus
Key compounds/major compounds	Individual compounds were not highlighted as primary bioactive agents	Propyl-2-(trimethylammonio)-ethyl phosphate; Ailanthoidol; Linoleic acid; C16-dihydrosphingosine
Degree values of major compounds	Not explicitly reported	57, 55, 53, and 43, respectively
Total predicted drug targets	901 targets	500 predicted targets (from 29 compounds)
Disease-related targets	Non-small cell lung cancer (GeneCards, OMIM)	Lung cancer metastasis-related targets (GeneCards)
Filtered disease targets	Not explicitly reported	1,077 targets
Drug–disease intersecting targets (Venn analysis)	Not explicitly reported	229 intersecting targets
Protein–protein interaction (PPI) analysis	Conducted	Conducted
Core targets (hub genes)	PTEN, AKT1, PIK3CA, MAPK1, TNF, IL6, CASP9, BAD	AKT1, PIK3CA, TNF, STAT3, IL1B, RHOA, RAC1, RHOC
Compound–target association pattern	Compounds mainly linked to apoptosis and survival signaling	Compounds mainly linked to PI3K/AKT and MAPK signaling
Main KEGG pathways	PI3K–Akt signaling pathway; MAPK signaling pathway; apoptosis	PI3K–Akt signaling pathway; NF-κB signaling pathway; cytokine–cytokine receptor interaction
Biological focus	Induction of apoptosis and enhancement of therapeutic sensitivity	Suppression of inflammation and metastasis
Experimental validation	Western blot	Western blot
Western blot–validated proteins	PTEN, PI3K, p-PI3K, AKT, p-AKT, BAD, cleaved caspase-9	PI3K, p-AKT, Lumican, p120-catenin, RhoA, Rac1, RhoC
Consistency between NP prediction and validation	High	High

Table 4 compiles in vitro evaluation data obtained from lung cancer cell models (A549, H1299, H292) and normal cell lines, including assay methods and IC₅₀ values. The table highlights agents and formulations demonstrating enhanced cytotoxic potency, improved selectivity, and superior performance compared with corresponding controls.

Table 4.

In vitro Cytotoxic and Antiproliferative Activity of Candidate Anti–Lung Cancer Agents

No.	Extract/Molecules/API	Method	In-vitro		Remarks	Ref
			Cell Line	IC50
1.	Bauhinia					[54]
1.1	Chloroform Extract	MTT Assay, 72 hours incubation 37°C, 5% CO2	Vero Cell A549 Cell	150.77 ± 1.1 µg/mL 74.23 ± 3.4 µg/mL	Extract and fraction concentrations were fixed at 100 µg/mL; among all tested samples, the ethyl acetate fraction exhibited the strongest inhibitory activity.
1.2	Ethyl Acetate Extract		Vero Cell A549 Cell	134.21 ± 4.4 µg/mL 89.35 ± 3.4 µg/mL
1.3	Methanol Extract		Vero Cell A549 Cell	139.22 ± 0.9 µg/mL 98.39 ± 4.5 µg/mL
1.4	Ethyl Acetate Fraction		Vero Cell A549 Cell	114.96 ± 3.2 µg/mL 54.23 ± 3.5 µg/mL
1.5	Methanol Fraction		Vero Cell A549 Cell	379.18 ± 3.0 µg/mL 450.96 ± 4.6 µg/mL
1.6	Acetone Fraction		Vero Cell A549 Cell	94.37 ± 0.7 µg/mL 595.51 ± 3.0 µg/mL
1.7	Chloroform Fraction		Vero Cell A549 Cell	896 ± 4.9 µg/mL 956.36 ± 2.9 µg/mL
1.8	Hexane Fraction		Vero Cell A549 Cell	> 1000 µg/mL > 1000 µg/mL
1.9	Doxorubicin (Control)		Vero Cell A549 Cell	6.70 ± 0.9 µg/mL 0.148 ± 0.04 µg/mL
2.	Curcumin					[57]
2.1	Curcumin 0.3 (µM) Curcumin 10 (µM)	Tube formation: matrigel assay, 4hours incubation CAM Assay: 72 hours incubation	HUVEC	P value < 0.01 P value < 0.001	Increasing concentrations resulted in progressively stronger inhibition of angiogenesis.
2.2	Curcumin 1 (µM) Curcumin 3 (µM) Curcumin 10 (µM) Curcumin 30 (µM)		Chicken CAM Neovascular	P value < 0.05 P value < 0.001 P value < 0.001 P value < 0.001
3.	Transferin					[52]
3.1	Ciszet-50 (Control)	MTT Assay: 24hours incubation post treatment	A549 Cell	5.45 ± 0.241 µg/mL	Transferrin-targeted micellar nanoparticles demonstrated a higher inhibitory potency.
3.2	CPT		A549 Cell	1.53 ± 0.042 µg/mL
3.3	CGPT		A549 Cell	1.25 ± 0.021 µg/mL
3.4	CGPT-Tf		A549 Cell	0.44 ± 0.012 µg/mL
4.	Sakuranin					[51]
4.1	Sakuranin	MTT Assay	A549 Cell	74.22 µg/mL	Sakuranin has inhibitory activities demonstrated in A549 cell.
5.	Mori Cortex
5.1	Cyclomorusin 24H	MTS Test (Proliferation)	NCI-H1299 A549	148.27 ± 3.04 µM 66.57 ± 2.56 µM	The cyclomorusin isolate exhibited 2.14-fold greater potency than cisplatin after 24 and 48 hours of incubation.	[50]
5.2	Cisplatin 24H (Control)		NCI-H1299 A549	>200 µM >200 µM
5.3	Cyclomorusin 48H		NCI-H1299 A549	97.63 ± 2.88 µM 38.89 ± 5.00 µM
5.4	Cisplatin 48H (Control)		NCI-H1299 A549	76.43 ± 1.52 µM 83.42 ± 4.72 µM
6.	Cisplatin					[53]
6.1	Ciszest-50 (Control)	MTT Assay	A549	6.86 ± 0.2 µg/mL	CPT: CDDP loaded Pluronic F127 TPGS Micelles) CPTT: CDDP + UCNP loaded loaded Pluronic F127 TPGS Micelles) CPTT-FA (Chitosan-Folic Acid decorated CDDP and UCNP loaded Pluronic F127 TPGS Micelles).
6.2	CPT		A549	1.38 ± 0.04 µg/mL
6.3	CPPTT		A549	1.23 ± 0.02 µg/mL
6.4	CPTT-FA		A549	0.54 ± 0.2 µg/mL
7.	Chalcone					[49]
7.1	Lonchocarpin (34)	MTT Assay	H292	10.0 µM	Compound 34 (lonchocarpin) exhibited the highest potency against H292 lung cancer cells, with no detectable toxicity toward Vero cells up to 100 µM; compounds 35–37 showed moderate activity, compounds 12, 15, 23, 26, and 33 displayed weak activity, while the remaining compounds were inactive.
7.2	Compound 35		H292	61.5 µM
7.3	Compound 36		H292	50.0 µM
7.4	Compound 37		H292	61.5 µM
7.5	Compound 12		H292	100.0 µM
7.6	Compound 15		H292	158.5 µM
7.7	Compound 23		H292	311.9 µM
7.8	Compound 26		H292	293.8 µM
7.9	Compound 33		H292	215.8 µM
7.10	Compound (1–11, 13–14, 16–22,24-32)		H292	>300 µM
8.	Artemesia judaica					[55]
8.1	Doxorubicin (Control)	MTT Assay	A549	9.98 ± 0.97 µg/mL	The Artemisia judaica extract exhibited strong anticancer activity against A549 lung cancer cells with selectivity toward normal WI-38 cells, while showing minimal effects on PC-3 and MDA-MB-231 cells.
8.2	A.Judaica Extract		A549 PC-3 MDA-MB-231 WI-38	14.2 ± 0.84 µg/mL 59.8 ± 3.25 µg/mL 98.6 ± 4.65 µg/mL 69.4 µg/mL
9.	Oleic Acid					[56]
9.1	CTX-OLA-ALB-NPs	MTT Assay	A549	4.34 ± 1.90 µg/mL	Cetuximab-functionalized albumin nanoparticles loaded with oleanolic acid demonstrated superior inhibitory activity.
9.2	OLA-ALB-NPs		A549	13.87 ± 1.28 µg/mL
9.3	Free OLA		A549	98.21 ± 1.45 µg/mL
9.4	Docotaxel (Control)		A549	25.3 ± 2.7 µg/mL
10.	Jinfuan Decoction					[58]
10.1	Low Dose JFAD-containing Serum	Transwell Assay	A549	Migration: 9.3% Invasion: 19.7% (P value < 0.001)	JFAD induces S-phase cell cycle arrest and reduced the migratory and invasive capabilities of A549 cells, indicating its potential antimetastatic properties.
10.2	Medium Dose JFAD-containing Serum		A549	Migration: 19.7% Invasion: 20.9% (P value < 0.001)
10.3	High Dose JFAD-containing Serum		A549	Migration: 28.5% Invasion: 44.5% (P value < 0.001)
11.	Yifei Miao Fang (QJYFMF) + Gefitinib					[59]
11.1	QJYFMF + GE	Transwell Assay	A549	Migration: P- value < 0.001 Invasion: P-value < 0.001	Transwell assay results showed that the number of invasive cells in the GE and QJYFMF groups was significantly reduced, and the invasive ability was decreased compared to control.

Table 5 summarizes in vivo evaluation data from xenograft and syngeneic lung cancer models, detailing experimental approaches, animal species, administered doses, and observed antitumor outcomes. The results highlight formulations and compounds exhibiting the strongest tumor growth inhibition, improved targeting efficiency, and favorable safety profiles.

Table 5.

In vivo Antitumor Efficacy of Candidate Anti–Lung Cancer Agents in Animal Models

No.	Extract/Molecules/API	In-vivo			Ref
		Method	Animal	Effectivity
1.	Bauhinia	Acute toxicity analysis, Tumor growth, antioxidant status, hematological and biochemical profiles, histopathological alterations, and p53 expression.	C57BL/6 mice	The Ethyl Acetate fraction of CEBA (250 mg/kg body weight/day) showed an optimal effective dose and potent anticancer efficacy against in vitro A549 lung cancer cell lines.	[54]
2.	Curcumin	A Lewis lung carcinoma mouse model, curcumin administration at different doses, tumor growth measurement, histological analysis, Ki-67 expression, microvessel density, and angiogenesis-related markers such as VEGF and VEGFR.	C57BL/6 mice	Curcumin at a dose of 300 mg/kg/day was the most effective in inhibiting the growth of Lewis lung carcinoma (LLC) cells.	[57]
3.	Transferin	Tumor xenograft modeling, intravenous administration of cisplatin-loaded micelles, tumor growth measurement, MRI-based biodistribution analysis, histopathological examination, and toxicity assessment.	Albino wistar rats	Transferrin-targeted micelles (CGPT-Tf) at a cisplatin dose of 5 mg/kg were the most effective in inhibiting lung cancer in vivo, significantly outperforming conventional cisplatin and non-targeted micelles.	[52]
4.	Sakuranin	A549 xenograft mouse modeling, intraperitoneal administration of sakuranin, tumor volume and weight measurement, body weight monitoring, and histopathological analysis.	BALB/c nude mice	In vivo investigations of sakuranin treatment in a dose at 200 mg/kg confirmed its anticancer efficacy by notably suppressing tumor growth.	[51]
5.	Mori Cortex	A lung cancer (NCI-H1299) xenograft mouse modeling, intraperitoneal administration of cyclomorusin, tumor volume monitoring, body weight assessment, and histopathological examination.	Nude mice	Cyclomorusin derived from cortex mori exhibited dose-dependent antitumor activity, with the 30 mg/kg dose producing the most pronounced inhibition of tumor growth in the in vivo lung cancer model.	[50]
6.	Cisplatin	Pharmacokinetic profiling, biodistribution and imaging studies, toxicity assessment, and histopathological analysis to assess the safety and targeting performance of folate-receptor targeted cisplatin micelles.	Wistar rats	The folate-receptor targeted micelles (CPTT-FA) at 5 mg/kg is the most effective formulation for enhancing cisplatin delivery and therapeutic potential in lung cancer.	[53]
7.	Chalcone	An S180 tumor-bearing mouse model, oral administration of the chalcone at different doses, tumor weight measurement, body weight monitoring, and histopathological analysis.	Male Kunming mice	The antitumor effect increased with higher chalcone lonchocarpin doses, showing the 100 mg/kg group had the strongest inhibition.	[49]
8.	Artemesia judaica	An A549 lung cancer xenograft model, administration of Artemesia judaica crude extract, tumor growth measurement, biochemical toxicity assessment, and histopathological analysis.	Swiss albino mice	The crude extract of Artemesia judaica L. administered at 100 mg/kg showed significant antitumor activity.	[55]
9.	Oleic Acid	Pharmacokinetic profiling, biodistribution on major organs (lung, liver, kidney, spleen, and heart) and imaging studies, toxicity assessment, and histopathological analysis	Wistar Rats	EGFR-targeted albumin nanoparticles of oleanolic acid (CTX-OLA-ALB-NPs), administered at an oleanolic acid dose of 5 mg/kg, demonstrated effective and targeted therapy against lung carcinoma.	[56]
10.	Jinfu’an Decoction	Tumor xenograft model, tumor volume measurement, bioluminescence imaging, hispatologycal and immunohistochemistry analysis.	BALB/c-nu mice	The medium dose exhibited the strongest and statistically significant antitumor effect (tumor inhibition rates: 35.68%), indicating that Jinfu’an decoction at 4× the clinical dose was the most effective dose for inhibiting tumor growth in vivo analysis.	[58]
11.	Yifei Miao Fang (QJYFMF) + Gefitinib	An A549 lung cancer xenograft model, tumor growth measurement and Western blot and qRT-PCR examination.	BALB/c nude mice	Combination of Qingjie Yifei Miao Fang (QJYFMF) with gefitinib demonstrated clear antitumor activity against non-small cell lung cancer. The treatment regimen consisted of a fixed dose of QJYFMF (25.09 g/kg/day) and gefitinib (32.5 mg/kg/day) administered orally for 3 weeks. The combination therapy produced the greatest reduction in tumor volume and weight compared with either monotherapy and increases of apoptosis and inhibition of the PTEN/PI3K/AKT signaling pathway, without body weight loss observation.	[59]

Network pharmacology analyses were derived from two independent systems-level investigations focusing on Qingjie Yifei Miao Fang (QJYFMF)58 and Jinfu’an Decoction (JFAD).59 In the QJYFMF study, a compound-centric strategy identified 98 active compounds based on oral bioavailability and drug-likeness criteria, resulting in the prediction of 901 protein targets associated with non-small cell lung cancer. Protein–protein interaction analysis identified PTEN, AKT1, PIK3CA, MAPK1, TNF, IL6, CASP9, and BAD as the most prominent hub targets. Pathway enrichment analysis highlighted the PI3K–Akt signaling pathway, MAPK signaling pathway, and apoptosis-related pathways as dominant regulatory networks. Western blot validation further confirmed modulation of PTEN, PI3K, p-PI3K, AKT, p-AKT, BAD, and cleaved caspase-9.

In contrast, the JFAD study employed a target-centric approach based on serum-detected compounds, identifying 32 circulating compounds using UHPLC-QTOF-MS. This analysis led to the prediction of approximately 500 drug targets, with 229 intersecting lung cancer–related targets identified following disease association filtering. Network degree analysis identified propyl-2-(trimethylammonio)-ethyl phosphate, ailanthoidol, linoleic acid, and C16-dihydrosphingosine as the most dominant compounds, with degree values of 57, 55, 53, and 43, respectively. Core hub targets identified from the PPI network included AKT1, PIK3CA, TNF, STAT3, IL1B, RHOA, RAC1, and RHOC, while KEGG pathway enrichment emphasized the PI3K–Akt, NF-κB, and cytokine–cytokine receptor interaction pathways. Western blot analysis validated alterations in PI3K, p-AKT, lumican, p120-catenin, RhoA, Rac1, and RhoC.

Collectively, these computational and network-based findings consistently identified high-affinity molecular interactions and densely connected signaling networks, with PI3K–Akt–centered pathways repeatedly emerging as dominant regulatory features across both molecular docking and network pharmacology datasets.

In vitro Evaluation

In vitro cytotoxic, antiproliferative, and antiangiogenic activities of candidate anti–lung cancer agents were evaluated across multiple lung cancer cell models, primarily A549, NCI-H1299, and H292, using standardized viability and proliferation assays. Normal (non-malignant) cell lines were included in selected studies to assess cellular selectivity.

In extract-based evaluations, fractionated preparations demonstrated distinct cytotoxic profiles against A549 lung cancer cells. Among all tested samples, the ethyl acetate fraction exhibited the strongest antiproliferative activity, achieving an IC₅₀ value of 54.23 ± 3.5 µg/mL, which was substantially lower than those of the corresponding crude extracts and other fractions. In contrast, hexane and chloroform fractions showed minimal cytotoxicity, with IC₅₀ values exceeding 1000 µg/mL. The reference control doxorubicin displayed substantially higher potency, with an IC₅₀ of 0.148 ± 0.04 µg/mL in A549 cells.54

Antiangiogenic activity was most prominently observed in curcumin-treated models. In both HUVEC-based tube formation assays and chicken chorioallantoic membrane (CAM) models, curcumin significantly suppressed neovascularization. The highest inhibitory effects were consistently recorded at concentrations of 10–30 µM, with strong statistical significance (p < 0.001) across both experimental systems.57

Nanocarrier-based formulations demonstrated a progressive enhancement of cytotoxic efficacy. In transferrin-targeted delivery systems, cisplatin-loaded micelles (CGPT-Tf) achieved a markedly reduced IC₅₀ of 0.44 ± 0.012 µg/mL in A549 cells, representing the most potent formulation within the series. This activity was substantially stronger than that of free cisplatin (5.45 ± 0.241 µg/mL) and non-targeted micellar formulations.52

Evaluation of sakuranin revealed measurable antiproliferative activity against A549 cells, with an IC₅₀ of 74.22 µg/mL, confirming its cytotoxic potential in lung cancer cells.51 In studies involving Cortex Mori–derived compounds, cyclomorusin exhibited strong and time-dependent antiproliferative effects. At 48 hours, cyclomorusin achieved IC₅₀ values of 97.63 ± 2.88 µM in NCI-H1299 cells and 38.89 ± 5.00 µM in A549 cells, outperforming the cisplatin control under equivalent experimental conditions, particularly in A549 cells.50

Further enhancement of cisplatin efficacy was observed in advanced micellar delivery systems incorporating targeting ligands. The chitosan–folic acid–decorated formulation (CPTT-FA) exhibited the lowest IC₅₀ value of 0.54 ± 0.2 µg/mL in A549 cells, surpassing both non-targeted micelles and free cisplatin, thereby highlighting the benefit of folate receptor-mediated targeting.53

Among a library of natural chalcone derivatives, lonchocarpin (compound 34) emerged as the most potent agent against H292 lung cancer cells, achieving an IC₅₀ of 10.0 µM. Importantly, lonchocarpin displayed no detectable cytotoxicity toward Vero cells at concentrations up to 100 µM, indicating high tumor selectivity. Other chalcone analogues showed moderate to weak antiproliferative activity, while the majority of compounds exhibited IC₅₀ values exceeding 100 µM.49

Crude extract evaluation of Artemisia judaica demonstrated selective cytotoxicity activity towards lung cancer cells. The extract achieved an IC₅₀ of 14.2 ± 0.84 µg/mL in A549 cells, while exhibiting reduced antiproliferative activity against PC-3 and MDA-MB-231 cells and only moderate effects on normal WI-38 fibroblast cells. The doxorubicin control showed an IC₅₀ of 9.98 ± 0.97 µg/mL in A549 cells, indicating comparable but higher potency relative to the plant extract.55

Nanoparticle-mediated delivery significantly enhanced the cytotoxic performance of triterpenoid compounds. Cetuximab-functionalized albumin nanoparticles loaded with oleanolic acid (CTX-OLA-ALB-NPs) demonstrated the strongest antiproliferative activity, with an IC₅₀ of 4.34 ± 1.90 µg/mL in A549 cells. This formulation markedly outperformed non-targeted nanoparticles (13.87 ± 1.28 µg/mL) and free oleanolic acid (98.21 ± 1.45 µg/mL), indicating a substantial enhancement of antiproliferative efficacy through EGFR-targeted nano formulation strategies.56 A summary chart illustrating representative IC₅₀ ± SEM values for the most potent formulations as shown in Figure 3.

Figure 3.

Summary of In Vitro IC₅₀ Potency of Selected NCSLC Treatments. Statistical significance between groups is indicated as follows: *Significant, ***Extremely significant. Bars represent mean IC₅₀ ± SEM and are derived from references.50,52–56

In vivo Evaluation

The in vivo antitumor efficacy of candidate anti–lung cancer agents assessed across syngeneic and xenograft models, incorporating tumor growth inhibition, pharmacokinetic and biodistribution analyses, toxicity profiling, and histopathological evaluation.

Across extract- and compound-based interventions, clear dose-dependent antitumor activity was consistently observed. Among plant-derived preparations, fractionated extracts of Bauhinia acuminata demonstrated robust tumor suppression, with the ethyl acetate fraction at 250 mg/kg/day producing the most pronounced inhibition of tumor growth in C57BL/6 mice, while maintaining favorable hematological, biochemical, and histopathological profiles.54 Similarly, the crude extract of Artemisia judaica administered at 100 mg/kg significantly suppressed tumor growth in A549 xenograft models, with acceptable systemic toxicity indicators.55

Polyphenolic agents exhibited strong antitumor efficacy in murine lung cancer models. In a Lewis lung carcinoma system, curcumin achieved substantial tumor growth inhibition across tested doses, with the 300 mg/kg/day regimen yielding the strongest antitumor effect in C57BL/6 mice, supported by consistent reductions in tumor progression metrics.57 In xenograft models, sakuranin administered intraperitoneally at 200 mg/kg significantly reduced tumor volume and tumor weight in BALB/c nude mice.51 Moreover, lonchocarpin showed a dose-dependent antitumor effect, with the 100 mg/kg dose producing the strongest inhibition of tumor growth, likely through suppression of the PI3K/AKT signaling pathway.49

Isolated bioactive compounds exhibited pronounced in vivo potency. Cyclomorusin, derived from Cortex Mori, displayed dose-dependent antitumor activity in NCI-H1299 xenograft models, with the 30 mg/kg dose producing the most substantial inhibition of tumor growth, without adverse effects on body weight, indicating a favorable safety profile.50

Targeted and ligand-decorated drug delivery systems consistently outperformed non-targeted counterparts. Transferrin-targeted cisplatin micelles (CGPT-Tf) administered at a cisplatin dose of 5 mg/kg demonstrated superior tumor suppression compared with free cisplatin and non-targeted micelles in albino Wistar rat models.52 Parallel findings were observed for folate receptor–targeted cisplatin micelles (CPTT-FA), which emerged as the most effective formulation at 5 mg/kg, enhancing therapeutic efficacy, while maintaining favorable biodistribution and safety profiles.53

Nano formulated triterpenoids further reinforced the advantages of receptor-targeted strategies. EGFR-targeted albumin nanoparticles loaded with oleanolic acid (CTX-OLA-ALB-NPs) administered at 5 mg/kg achieved efficient tumor targeting and significant antitumor activity in Wistar rat models, supported by organ-specific biodistribution patterns and histopathological evidence of therapeutic findings.56

Herbal formulation–based therapies also demonstrated significant in vivo efficacy. In BALB/c-nu xenograft models, Jinfu’an Decoction produced the strongest and statistically significant inhibition of tumor growth at the medium-dose level, corresponding to four times the clinical dose, with a tumor inhibition rate of 35.68%.58 Moreover, combination therapy with Qingjie Yifei Miao Fang (QJYFMF) and gefitinib yielded the greatest reduction in tumor volume and weight in A549 xenograft models, surpassing either monotherapy and proceeding without observable body weight loss over the treatment period.59

Discussion

Integration of in silico, in vitro, and in vivo Findings

The integrated analysis of in silico, network pharmacology, in vitro, and in vivo datasets reveals a coherent and biologically rational continuum that underpinning the discovery, design, and development of high-potential anti–lung cancer agents. Rather than representing isolated or modular layers of evidence, these four evaluation platforms collectively form a linear translational pipeline in which molecular-level interactions drive cellular responses, system-level network perturbations amplify biological vulnerability, and organism-level outcomes ultimately reflect the convergence of these effects within the tumor microenvironment.

At the molecular level, the strongest in silico signals were defined not merely by favorable binding energies, but by extensive residue engagement within functionally critical domains of oncogenic proteins. Compounds such as lonchocarpin, cyclomorusin, and sakuranin exhibited either exceptionally low docking scores or consistent high-affinity binding across multiple nodes of the PI3K/AKT/mTOR signaling axis. These interactions predominantly occurred at regions associated with kinase activation, substrate recognition, or regulatory stability, thereby predisposing these compounds to exert downstream biological effects. Importantly, ligand–receptor docking analyses of carrier-based systems, including transferrin–transferrin receptor and folate–folate receptor-α interactions, demonstrated structurally stable and energetically favorable binding modes that rationalize enhanced cellular internalization and tumor-selective targeting. Thus, the in-silico layer established not only target affinity but also structural plausibility for functional interference and targeted drug delivery.

These molecular predictions translated directly into observable in vitro phenotypes. Compounds and formulations with the strongest docking performance consistently demonstrated lower IC₅₀ values, time-dependent antiproliferative activity, and improved selectivity profiles in lung cancer cell lines. The extreme docking score of lonchocarpin was mirrored by its superior cytotoxic potency against H292 cells with negligible toxicity towards normal cells, indicating that extensive hydrophobic and aromatic residue engagement at the molecular level corresponds to efficient disruption of cellular survival machinery. Similarly, compounds exhibiting multi-target engagement within the PI3K/AKT/mTOR cascade, such as cyclomorusin and sakuranin, demonstrated sustained antiproliferative effects, particularly upon prolonged exposure, reflecting cumulative pathway suppression rather than transient receptor blockade. In contrast, compounds or fractions with weaker or fragmented docking profiles consistently displayed reduced cellular potency, underscoring the predictive value of molecular interaction density and target centrality in anticancer drug screening.

Network pharmacology further amplified this molecular–cellular relationship by contextualizing individual target interactions within interconnected signaling architectures. Both Qingjie Yifei Miao Fang and Jinfu’an Decoction revealed densely connected protein–protein interaction networks dominated by PI3K/AKT-centered hubs, inflammatory mediators, apoptosis regulators, and cytoskeletal signaling proteins. The recurrent emergence of AKT1, PIK3CA, PTEN, STAT3, and RHO-family proteins as central network hubs explain why multi-component formulations exhibited robust in vitro efficacy despite intrinsic chemical heterogeneity. Network-level convergence generates biological redundancy, ensuring that simultaneous inhibition at multiple nodes produces amplified phenotypic effects even when individual target affinities vary. The high concordance between predicted hub targets and experimentally validated protein modulation confirms that systems-level vulnerability, rather than single-target inhibition, underlies the observed cellular outcomes.

The transition from in vitro to in vivo efficacy reflects the successful escalation of pharmacodynamic effects into complex biological systems. Agents that demonstrated low IC₅₀ values, sustained cytotoxicity, and favorable selectivity profiles in cell-based assays consistently produced significant tumor growth suppression in animal models. Importantly, nano formulated and receptor-targeted systems exhibited superior in vivo performance compared with free drugs, despite similar or even lower administered doses. This phenomenon is directly attributable to carrier–receptor interactions predicted in silico, which facilitate preferential tumor accumulation, prolonged systemic circulation, and reduced off-target exposure. As a result, enhanced intracellular drug delivery observed in vitro translated into improved tumor targeting, reduced systemic toxicity, and more pronounced tumor regression in vivo.

The consistent superiority of targeted nano formulations across all evaluation platforms highlights the critical role of rational drug delivery design in lung cancer therapy. Transferrin- and folate-decorated delivery systems exploited overexpressed receptors on lung cancer cells, enabling receptor-mediated endocytosis that augmented intracellular drug concentrations. EGFR-targeted albumin nanoparticles further exemplified this principle by coupling molecular recognition with pharmacokinetic optimization. Collectively, these delivery strategies bridge molecular docking predictions with organism-level outcomes, demonstrating that structural compatibility at the receptor interface governs not only binding affinity but also therapeutic index, biodistribution and overall antitumor efficacy.

Across all analytical layers, the PI3K–AKT signaling axis consistently emerged as the dominant biological framework governing therapeutic responsiveness. This convergence is mechanistically rational, as PI3K–AKT integrates proliferative signaling, resistance to apoptosis, metabolic adaptation, angiogenesis, and inflammatory crosstalk processes that collectively sustain lung cancer progression. Compounds and formulations capable of disrupting this axis at multiple regulatory levels consistently demonstrated superior cellular inhibition and tumor suppression. The recurrent identification of this pathway across molecular docking studies, network hubs, protein validation, cytotoxic assays, and animal models underscores its central role as a convergence point for effective anti–lung cancer intervention.

Taken together, the linear correlation observed across in silico, network pharmacology, in vitro, and in vivo platforms confirms the robustness of an integrated drug discovery paradigm. Candidate agents that maintained internal consistency across molecular affinity, network centrality, cellular efficacy, and organism-level response exhibited the highest translational potential. This coherence validates the use of multi-layered computational and experimental integration as a predictive framework for anti–lung cancer drug development, reducing empirical attrition and enhancing confidence in candidate prioritization. The present findings collectively demonstrate that rational integration of molecular modeling, systems biology, and experimental pharmacology provides a powerful strategy for advancing effective and biologically grounded lung cancer therapeutics.

Several methodological and experimental constraints across the reviewed studies limit the direct translation of these multi-platform findings into clinical applications. Most notably, none of the 11 included studies validated their computational docking or network-predicted target interactions using gold-standard biophysical affinity assays such as Surface Plasmon Resonance (SPR) or Isothermal Titration Calorimetry (ITC), leaving predicted molecular interactions without physical thermodynamic verification. Furthermore, systemic safety evaluations in animal models were highly inconsistent, as only 6 of the 11 studies conducted thorough organ histopathology and full clinical biochemistry panels. These experimental gaps are further compounded by limitations in computational modeling, where protocols frequently relied on rigid receptor backbones and static energy minimization without accounting for induced-fit conformational changes or protein flexibility. Finally, predictive models largely omitted critical pharmacokinetic failure modes—such as metabolic stability, plasma protein binding, off-target transporter interactions, and systemic clearance—which represent primary drivers of translational failure in downstream clinical drug development.

Challenges and Future Perspectives

When evaluating the computational docking metrics reported across these studies, it is critical to emphasize that predicted binding energies (eg, values ranging from −41.36 kcal/mol to −7.0 kcal/mol) function strictly as relative, intra-assay scoring metrics rather than absolute thermodynamic measurements suitable for cross-study comparison. Free energy scoring functions in molecular docking rely on target-specific force fields, varying solvent models, diverse grid box dimensions, and different receptor backbone constraints. Consequently, a docking score calculated for one specific protein–ligand system cannot be directly compared to a score derived from a different target, binding pocket, or computational protocol. These values serve primarily to prioritize candidate compounds within a single defined target context or screen, providing structural hypotheses for residue engagement rather than a unified quantitative scale of biological potency.

Despite the compelling and coherent evidence generated across in silico, network pharmacology, in vitro, and in vivo platforms in the eleven core studies analyzed in this review, several conceptual and methodological challenges remain that must be addressed to strengthen translational relevance and guide future anti–lung cancer drug development.

The integration of in silico, network pharmacology, in vitro, and in vivo platforms demonstrates a coherent translational continuum where targeted molecular interactions drive cellular cytotoxicity, network perturbation, and tumor regression. High-affinity candidates and multi-component formulations consistently disrupted central oncogenic hubs—particularly the PI3K/AKT axis—while receptor-targeted nano-formulations successfully translated docking predictions into enhanced in vivo tumor accumulation. However, these findings are tempered by notable methodological limitations: none of the 11 studies validated binding predictions using direct biophysical assays (SPR or ITC), only 6 reported complete in vivo safety panels, computational models omitted protein flexibility and pharmacokinetic failure modes, and network pharmacology analyses remain subject to database annotation bias toward heavily studied pathways. Overall, despite these constraints, this multi-layered paradigm offers a robust framework for prioritizing candidate anti–lung cancer therapeutics while minimizing empirical attrition.

One of the primary challenges lies in the heterogeneity of computational evaluation frameworks. Although molecular docking and related in silico approaches consistently identified strong interactions between candidate agents and key lung cancer–associated targets particularly within the EGFR and PI3K/AKT/mTOR signaling axes, the quantitative docking scores cannot be interpreted as a unified thermodynamic scale across different receptors, ligands, and docking protocols. Variations in receptor preparation, scoring functions, binding site definition, and ligand physicochemical diversity inevitably influence predicted affinities. Consequently, docking results are best regarded as relative ranking tools within a defined target context, rather than absolute predictors of biological potency. Future studies would benefit from harmonized docking workflows, inclusion of molecular dynamics simulations to assess binding stability, and integration of free-energy–based calculations for top-ranked complexes, thereby enhancing predictive confidence and translational reliability.

A second major challenge concerns the interpretation of multi-component systems, particularly evident in studies employing crude extracts, fractions, and traditional herbal formulations. Network pharmacology has proven highly valuable for mapping multi-target interactions and identifying pathway convergence, most notably involving the PI3K–Akt, MAPK, apoptosis, and inflammatory signaling pathways. However, network-based predictions are inherently constrained by database completeness, annotation bias toward well-characterized proteins, and assumptions regarding compound bioavailability. Importantly, the comparison between compound-centric and serum-compound–centric strategies highlight a critical translational limitation: the biologically active entities in vivo may differ substantially from those predicted solely on in silico drug-likeness criteria. This discrepancy underscores the need for exposure-aware network pharmacology, in which computational predictions are systematically anchored to experimentally verified circulating compounds and metabolites, thereby enhancing biological and translational relevance.

At the cellular level, although in vitro assays provided robust evidence of cytotoxic, antiproliferative, and antiangiogenic effects particularly for targeted formulations and selected phytochemicals cell viability metrics alone cannot capture the full complexity of tumor biology. IC₅₀ values derived from monolayer cultures compress multiple biological variables into a single quantitative endpoint and do not adequately reflect tumor architecture, stromal interactions, immune modulation, or pharmacokinetic constraints. This limitation is especially pertinent for nano-delivery systems, where enhanced cellular uptake in vitro may overestimate therapeutic benefit under physiological conditions. The adoption of intermediate-complexity models, such as 3D tumor spheroids, patient-derived organoids, and co-culture systems incorporating stromal or immune components, therefore represents a critical future direction for bridging the gap between cell-based assays and in vivo efficacy.

From an in vivo perspective, while xenograft and syngeneic models consistently confirmed antitumor efficacy and therapeutic enhancement achieved through targeting strategies, model dependency remains a significant translational bottleneck. Subcutaneous xenograft models, although indispensable for early proof-of-concept evaluation studies, incompletely represent the lung tumor microenvironment and immune landscape, both of which strongly influence therapeutic response in patients. Moreover, the efficacy of receptor-targeted nanocarriers is subject to variability in receptor expression levels, tumor vascular permeability, and systemic clearance dynamics. Future investigations should increasingly prioritize the use of orthotopic lung cancer models, syngeneic or humanized immune systems, and integrated pharmacokinetic–pharmacodynamic (PK/PD) analyses to establish clinically relevant exposure–response relationships.

Looking forward, the most promising trajectory for advancing anti–lung cancer drug discovery lies in methodological integration and mechanistic triangulation. Computational predictions should be refined through standardized workflows and validated experimentally via pathway-specific perturbation studies. Network pharmacology outputs must evolve from descriptive pathway enrichment analyses towards causality-driven validation using genetic or pharmacological modulation of predicted hub targets. Experimentally, aligning in vitro assays, in vivo efficacy models, and molecular readouts with shared mechanistic endpoints will be essential to reinforce cross-platform coherence and predictability. Finally, for herbal formulations and complex natural systems that already demonstrate strong concordance between computational prediction, protein-level validation, andin vivo efficacy, the key future challenge lies in standardization and reproducibility. Rigourous chemical fingerprinting, identification of sentinel bioactive markers, batch-to-batch consistency, and exposure-driven quality control are indispensable prerequisites for translational development. Without such measures, even highly effective multi-component therapies risk remaining scientifically intriguing yet pharmaceutically undeployable.

Alongside traditional computational frameworks, the rapid maturation of artificial intelligence (AI) and machine learning (ML) is reshaping the early-stage drug discovery landscape. Emerging approaches—including deep-learning-based binding-affinity prediction, generative AI for de novo molecular design, and multi-omics data integration—offer powerful complements to classical docking and network-pharmacology pipelines. By leveraging neural networks trained on extensive structural, biophysical, and transcriptomic datasets, these advanced paradigms account for complex non-linear target interactions, subtle protein flexibility, and cell-type-specific network alterations that conventional static scoring functions routinely miss. Integrating AI/ML-driven virtual screening with multi-omics integration promises to significantly accelerate candidate prioritization, refine multi-target predictions, and improve clinical translatability for novel anti–lung cancer therapeutics.

Translating nano formulated and herbal anti–lung cancer candidates to the clinic requires navigating significant pharmacokinetic, regulatory, and production hurdles. Nanocarriers often face rapid immune clearance and variable tumor penetration, while multi-component herbal mixtures suffer from unpredictable metabolism and low bioavailability. Regulatory approval demands rigorous nanotoxicity profiling and biodistribution tracking for nanomedicines, alongside strict chemical fingerprinting and marker quantification for botanical extracts. Crucially, scaling up under Good Manufacturing Practice (GMP) requires stringent controls over particle size, encapsulation efficiency, and raw material consistency to ensure batch-to-batch reproducibility.

Overall, addressing these challenges will not only enhance the translational robustness of anti–lung cancer drug candidates but also reinforce the conceptual framework of linear correlation across computational, cellular, and animal-based evaluation platforms, as proposed in this review.

Conclusion

This review demonstrates that high-potential anti–lung cancer agents are most reliably prioritized not by isolated potency metrics, but by the linear convergence of molecular-level predictions, systems-level network engagement, cellular responses, and in vivo therapeutic performance. Across the eleven integrated studies, candidate agents that simultaneously exhibited strong target affinity within PI3K/AKT-centered oncogenic networks, multi-target pathway modulation, selective cytotoxicity in lung cancer cell lines, and consistent tumor suppression in animal models emerged as the most promising therapeutic leads. Notably, agents formulated through rational delivery strategies or evaluated within combination regimens showed the highest degree of cross-platform consistency, underscoring the importance of target accessibility, pharmacokinetics, and biological context in determining organism-level outcomes. However, cross-platform coherence should be interpreted as a predictive indicator of translational potential rather than a confirmatory guarantee of clinical success. Standard preclinical models inherently fail to fully recapitulate human disease, such as profound interspecies pharmacokinetic differences, the dynamic complexity of the human tumor microenvironment, and extensive patient inter-individual heterogeneity frequently introduce clinical translational barriers. Consequently, while multi-layered preclinical alignment significantly mitigates empirical attrition and strengthens candidate selection, downstream clinical validation remains indispensable. Collectively, these findings reinforce integrated, multi-platform assessment as a powerful and predictive framework for prioritizing candidates and guiding rational anti–lung cancer drug discovery.

Funding Statement

This study was funded by Universitas Padjadjaran through the Indonesian Endowment Fund for Education (LPDP) on behalf of the Indonesian Ministry of Higher Education, Science and Technology and managed under the EQUITY Program Based on the Decree Number 2024/UN6.RKT/Kep/HK/2025 and the Article Review Grant Agreement/Contract Number 3977/UN6.3.1/PT.00/2025.

Abbreviations

AKT1, AKT Serine/Threonine Kinase 1; BAD, BCL2 Associated Agonist of Cell Death; CASP9, Caspase-9; Cytoscape, Cytoscape Network Visualization Software; DL, Drug-Likeness; GeneCards, GeneCards Human Gene Database; IL1B, Interleukin-1 Beta; IL6, Interleukin-6; JFAD, Jinfu’an Decoction; KEGG, Kyoto Encyclopedia of Genes and Genomes; MAPK, Mitogen-Activated Protein Kinase; MAPK1, Mitogen-Activated Protein Kinase 1; MAPK3, Mitogen-Activated Protein Kinase 3; NF-κB, Nuclear Factor Kappa-Light-Chain-Enhancer of Activated B Cells; NP, Network Pharmacology; NSCLC, Non-Small Cell Lung Cancer; OB, Oral Bioavailability; OMIM, Online Mendelian Inheritance in Man; PI3K, Phosphatidylinositol 3-Kinase; PIK3CA, Phosphatidylinositol-4,5-Bisphosphate 3-Kinase Catalytic Subunit Alpha; PPI, Protein–Protein Interaction; PTEN, Phosphatase and Tensin Homolog; QJYFMF, Qingjie Yifei Miao Fang; RAC1, Rac Family Small GTPase 1; RHOA, Ras Homolog Family Member A; RHOC, Ras Homolog Family Member C; STAT3, Signal Transducer and Activator of Transcription 3; STRING, Search Tool for the Retrieval of Interacting Genes/Proteins; SwissTargetPrediction, SwissTargetPrediction Target Identification Tool; TCMSP, Traditional Chinese Medicine Systems Pharmacology Database; TNF, Tumor Necrosis Factor; UHPLC-QTOF-MS, Ultra-High Performance Liquid Chromatography–Quadrupole Time-of-Flight Mass Spectrometry; WB, Western Blot.

Disclosure

The authors report no conflicts of interest in this work.