Evaluating the efficacy and safety of antibody–drug conjugates in non-small cell lung cancer: a systematic review and meta-analysis

Abstract

Background

Numerous Antibody–Drug Conjugates (ADCs) have been investigated for non-small cell lung cancer (NSCLC), yielding mixed results. This study comprehensively evaluated the efficacy and safety of ADC therapies in NSCLC patients, particularly focusing on specific populations.

Methods

A literatures search was conducted to identify prospective trials published between January 2000 and June 2025. Only randomized and non-randomized phase II-IV clinical trials involving adult NSCLC patients treated with ADCs were selected. Efficacy endpoints were categorized based on the primary outcomes of the included studies.

Results

The analysis included 16 studies with 1872 participants. The pooled analysis showed a Objective Response Rate (ORR) was 34% (95% CI: 26%-42%) with high heterogeneity (I² = 91.7%). Subgroup analyses revealed significant variations in ORR between different ADC agents (P < 0.0001). In specific NSCLC subgroups, the ORR was 35% for Epidermal Growth Factor Receptor (EGFR)-mutant patients and 36% for those with actionable genomic alterations (AGAs). Notably, HER2-mutant patients achieved a significantly higher ORR of 55%, compared to 21% in populations lacking these mutations (P < 0.0001). ADC therapy may have limited efficacy against squamous cell carcinoma. All-grade and grade ≥ 3 treatment-related adverse events (TRAEs) occurred in 95% and 43% of patients, respectively, both showing high heterogeneity. The incidence of all-grade interstitial lung disease (ILD) was 10%, with the grade ≥ 3 incidence being 2%. The gastrointestinal system was the most frequently involved, but these were predominantly low-grade. In contrast, hematologic and respiratory system involvement were more common among grade ≥ 3 AEs. Pneumonitis and ILD were the leading causes of both treatment-related mortality and discontinuation.

Conclusion

ADC monotherapy has demonstrated considerable efficacy in previously treated NSCLC. Patients with non-squamous histology, EGFR mutations, or HER2 mutations may derive greater benefit from ADC therapy. However, it should be noted that the partially pooled results, derived from highly heterogeneous data, require cautious interpretation. Close monitoring and proactive management of hematologic and respiratory system-related toxicities are essential.

PROSPERO registration

CRD420251101467

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-025-15461-6.

Introduction

Lung cancer is the second most commonly diagnosed cancer and the leading cause of cancer-related death globally, with non-small cell lung cancer (NSCLC) being its predominant histological subtype, accounting for approximately 85% of cases [1]. Current first-line treatments for advanced NSCLC primarily include chemotherapy, immunotherapy, and targeted therapy for specific genetic mutations. However, upon the development of acquired resistance, subsequent treatment options become severely limited. Docetaxel-based chemotherapy currently serves as the standard later-line treatment, yet it offers limited efficacy and is associated with significant toxicity [2]. Novel second-line agents, such as amivantamab and ivonescimab have emerged, but their use is restricted to patients with specific biomarkers like Epidermal Growth Factor Receptor (EGFR) mutations or PD-L1-positive patients [3, 4]. This leaves a substantial unmet clinical need, underscoring the urgent demand for additional therapeutic options with superior antitumor activity.

Antibody–drug conjugates (ADCs) represent a class of emerging anticancer agents, composed of a monoclonal antibody (mAb) linked to a cytotoxic payload via a chemical linker. The antibody component of ADCs is predominantly based on humanized antibodies [5]. However, its large molecular size (150 kDa) results in relatively inefficient tumor penetration. Consequently, researchers are exploring the use of antibody fragments or peptides alone as novel drug carriers [6]. The linker, a biochemical component that conjugates the antibody to the cytotoxic payload, plays a dual role in modulating both the toxicity and efficacy profiles of ADCs by maintaining their stability in the systemic circulation [7]. The latest design of hydrophilic linker technology enables the incorporation of multiple payloads while maintaining excellent physicochemical properties and pharmacokinetic profiles [8].Upon internalization and linker-based release, the payload exerts potent cytotoxic effects on target cells. Common payloads are broadly categorized into two classes: Microtubule inhibitor and topoisomerase I inhibitor [9]. ADCs can also kill adjacent tumor cells through the "bystander effect," a phenomenon particularly associated with drugs utilizing cleavable linkers and hydrophobic payloads [10]. Additionally, the drug-to-antibody ratio (DAR) is a critical determinant of the pharmacological characteristics and clinical activity of ADCs. The DAR for FDA-approved ADCs typically ranges from 2 to 8 [11]. Moreover, to enhance therapeutic efficacy and mitigate adverse effects, ADC design is evolving through novel strategies and payload innovations. Key approaches include: employing bispecific antibodies to boost cellular internalization and/or improve tumor targeting; utilizing small-molecule drug conjugates (as alternatives to antibody scaffolds) to enhance tissue penetration, including into the central nervous system; and exploring non-cytotoxic payloads such as radioisotopes, enzymes, small molecules, and peptides [12].

Current targets for ADCs in NSCLC include human epidermal growth factor receptor 2 (HER2), HER3, trophoblast cell surface antigen 2 (TROP2), mesenchymal-epithelial transition factor (c-MET), carcinoembryonic antigen-related cell adhesion molecule 5 (CEACAM5), and B7 homolog 3 (B7-H3), among others. For each target, one or more specific ADCs have been developed, demonstrating encouraging efficacy with objective response rates (ORRs) ranging from 20 to 56% in phase I-II clinical trials [8, 13]. Notably, two phase III randomized controlled trials (RCTs), EVOKE-01 [14] and TROPION-Lung01 [15], reported that sacituzumab govitecan (SG) and datopotamab deruxtecan (Dato-DXd) did not demonstrate a statistically significant improvement in median overall survival (mOS) or median progression-free survival (mPFS) compared to docetaxel monotherapy in previously treated advanced NSCLC patients. Conversely, a subgroup analysis of TROPION-Lung01 revealed significant improvements with ADC therapy in specific tumor histologies and among patients with actionable genomic alterations (AGAs) (including EGFR, ALK, ROS1, NTRK, BRAF, MET exon 14 skipping, or RET alterations, excluding KRAS). Furthermore, recent results from the phase II RCT OptiTROP-Lung03 [16] indicated that sacituzumab tibotansutecan (Sac-TMT) showed superior benefits in ORR, PFS, and OS compared to docetaxel in an EGFR-mutant population. These divergent outcomes across studies underscore the critical importance of patient selection. By combining precise antibody targeting with potent cytotoxic payloads, ADCs can act on multiple tumor antigens, offering a key strategy to overcome treatment resistance [17, 18]. Several ADCs are now approved for NSCLC patients with specific biomarkers, such as high c-Met expression and resistant HER2/EGFR mutations [19–21]. Current research also includes combination approaches like Dato-DXd with immunotherapy for immune-related resistance [22] and ADC plus prior EGFR-TKI for non-target-mediated EGFR-TKI resistance [12]. Compared to traditional chemotherapy, the high specificity of ADC antibodies enables precise targeting, potentially reducing systemic toxicity, while their high affinity may allow for higher maximum tolerated doses [23]. Nonetheless, ADCs are still associated with adverse reactions such as hematological, respiratory, and gastrointestinal toxicities, which may result from the premature release of cytotoxic payloads into the systemic circulation [17]. Additionally, immune responses triggered by either the antibody or the cytotoxic component can lead to secondary tissue damage [8].

In summary, the overall potential and safety profile of ADCs require further elucidation. A systematic review and meta-analysis was conducted, employing a quantitative synthesis of published data from phase II-III prospective clinical trials to provide a comprehensive assessment of the efficacy and adverse events (AEs) profile of ADC therapies in advanced NSCLC. Furthermore, we conducted subgroup analyses to explore differences based on study design, cancer pathology, tumor mutation status, and different ADC targets/drugs. This work will provide consolidated evidence on the effectiveness and safety of ADCs, thereby informing their more effective clinical application in NSCLC.

Methods

Protocol and guidelines

This study was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [24]. The study protocol was prospectively registered on PROSPERO (Registration Number: CRD420251101467).

Search strategy

A systematic literature search was performed using PubMed, Embase, Web of Science, the Cochrane Library, and ClinicalTrials.gov for relevant clinical studies published from January 1, 2000, to June 30, 2025. Search strategies incorporated Medical Subject Headings (MeSH terms), supplementary concepts, and keywords. Key search terms included various specific ADCs (such as trastuzumab emtansine, sacituzumab govitecan, etc.), "antibody–drug conjugates", "non-small cell lung carcinoma", "NSCLC", "trial" and "study". The search was restricted to studies investigating the efficacy and safety of ADC-based therapies in NSCLC, focusing on prospective phase II-IV clinical trials. Additionally, the reference lists of retrieved articles were manually screened to identify any potentially eligible records. The detailed search strategy is provided in the Supplementary Material (Excel file).

Inclusion and exclusion criteria

Studies meeting the following criteria were included: (1) patients receiving ADC monotherapy or combination therapy, regardless of dose, treatment duration, or line of therapy; (2) adult patients with NSCLC; (3) reported data on efficacy outcomes and/or AEs; (4) prospective clinical trials (phase II-IV) with or without a control group. Studies were excluded based on the following criteria: (1) phase I trials, case reports, reviews, meta-analyses, non-peer-reviewed manuscripts, pooled analyses, editorials, expert opinions, conference abstracts, observational studies, retrospective studies, or preclinical studies; (2) studies lacking a full-text article or reported only in an abstract, including ongoing or withdrawn trials; (3) studies involving solid tumors without separate outcome data for the NSCLC cohort; (4) duplicate patient cohorts; (5) studies involving patients with NSCLC concurrent with other malignancies (6) non-English publications.

Study selection and data extraction

Two investigators independently identified eligible studies and extracted relevant information after searching the databases and removing duplicates. Data were extracted from the included studies by two independent reviewers. Any discrepancies between the investigators during the study selection or data extraction process were resolved through further discussion. The following metrics were collected: (1) basic information, including clinical trial number, first author's name, and publication year; (2) study characteristics, including trial phase, country, randomization, blinding, single-arm/two-arm design, ADC target, ADC drug name, NSCLC pathological type, number of patients, baseline patient characteristics, interventions, and follow-up time; (3) efficacy outcomes: the primary outcome was the ORR. Subgroup analyses were performed on the primary outcome ORR, including by tumor pathological type, ADC target, ADC drug, mutation type, and randomized/non-randomized study, to examine the influence of various factors on ORR. Secondary outcomes included: the disease control rate (DCR); the 1-year OS rate; the duration of response (DOR); PFS; and OS; (4) safety outcomes, including the number of patients who with treatment-related adverse events (TRAEs), drug discontinuation, and death. TRAEs included all-grade and grade ≥ 3 events. Treatment-related interstitial lung disease (ILD) was specifically analyzed, including all-grade and grade ≥ 3 events. AEs were graded according to the Common Terminology Criteria for Adverse Events (CTCAE), version 5.0. The terminology for AEs followed the Medical Dictionary for Regulatory Activities (MedDRA).

Statistical analysis

This analysis included only the treatment arms with ADC drugs. Event rates were calculated as the number of patients experiencing the event divided by the total number of patients receiving ADC-containing therapy. Rates were transformed using the logit function, and corresponding 95% confidence intervals (CIs) were calculated using the binomial method. Pooled analyses of DOR, PFS, and OS were performed using a random-effects model with log-transformation and inverse-variance weighting for data synthesis. The incidence of specific all-grade and grade ≥ 3 TRAEs with an overall incidence of at least 10% was reported. The risk of bias for randomized trials included in the systematic review was assessed using the Cochrane Risk of Bias 2 tool (ROB2) as recommended by the Cochrane Handbook for Systematic Reviews of Interventions [25]. For non-randomized intervention studies, the Risk Of Bias In Non-randomized Studies—of Interventions (ROBINS-I) tool was used [26]. The degree of heterogeneity among studies was assessed using the I² statistic. An I² of 0%–25% indicated low heterogeneity, 25%–50% indicated moderate heterogeneity, and an I² > 50% indicated high heterogeneity. When the I² value exceeded 50%, a random-effects model was used to summarize the effect size; otherwise, a common-effect model was applied. Sensitivity analyses were performed for all outcomes to identify sources of heterogeneity. The potential for publication bias was assessed using Egger's test and funnel plots. If publication bias was detected, the trim-and-fill method was used to adjust the effect size. A p-value of less than 0.05 was considered statistically significant. All statistical analyses were performed using R software (version 4.5.1, R Foundation for Statistical Computing).

Results

Literature search and study selection

A systematic literature search identified 557 records from the databases. After removing 145 duplicates, 412 records were screened based on their titles and abstracts, of which 161 were deemed eligible for full-text review. Ultimately, 16 studies met the inclusion criteria and were selected for further analysis, encompassing a total of 1872 subjects who received ADC-containing regimens. A flowchart summarizing the study selection process is presented in Fig. 1.

Fig. 1.

Flow diagram of study selection

Characteristics of included studies

The characteristics of the 16 included studies, comprising 19 cohorts, are presented in Table 1. All included studies were conducted in patients with locally advanced, metastatic, or recurrent NSCLC, including 2 studies in adenocarcinoma, 4 in non-squamous carcinoma, 1 in squamous cell carcinoma, and 9 in unspecified NSCLC. The ADCs used in the studies included: Trastuzumab emtansine (T-DM1) (n = 4), Datopotamab deruxtecan (Dato-DXd) (n = 2), Sacituzumab tirumotecan (Sac-TMT) (n = 3), Telisotuzumab vedotin (Teliso-V) (n = 2), Trastuzumab deruxtecan (T-DXd) (n = 2), Patritumab deruxtecan (Patritu-DXd) (n = 1), Sacituzumab govitecan (SG) (n = 1), and Trastuzumab rezetecan (SHR-A1811) (n = 1). The characteristics and approved indications of these ADCs are summarized in Table 2. These ADCs targeted c-MET (n = 2), HER2 (n = 7), HER3 (n = 1), and TROP-2 (n = 6). Regarding study design, there were 14 phase II trials and 2 phase III trials, consisting of 5 randomized trials and 11 non-randomized trials. Among the studies, two were phase III RCTs comparing the experimental arms of Dato-DXd (n = 1) and SG (n = 1) against docetaxel control groups. One study was a phase II RCT comparing the experimental arm of Sac-TMT (n = 1) against a docetaxel control group. All studies evaluated ADC monotherapy, with no combination therapies included. In terms of blinding, one study was double-blind (participants and investigators), while the remainder were open-label studies. Ten studies were multinational, 3 were multi-center trials conducted in China, 2 were multi-center trials in Japan, and 1 was a single-center trial in the United States.

Table 1.

Characteristics of included studies

Study Identifier(NCT)	First author	Year	Country	Randomization	Phase	Blinding	One/two arms	Pathology	Target	adc	Dosages	N	Median follow-up time,month	Median age, years	Male,n(%)	Smoke, never,n(%)	ECOG PS	Brain metastasis,n(%)	lines of previous cancer therapy Median (range)	Primary endpoints	Secondary endpoints
LUMINOSITY [19] (NCT03539536)	Camidge	2024	Multinational	No	II	Open label	One	nonsquamous EGFR-wildtype NSCLC	c-Met Protein–Overexpressing	Telisotuzumab Vedotin	1.9 mg/kg/Q2w	161	4.9(0.1–26)	64.0 (33–83)	111 (68.9)	35 (21.7)	0–2	33 (20.5)	1 (1–3)	ORR(ICR)	DOR, DCR, PFS, OS, AE
DESTINY-Lung01 [27](NCT03505710) -Cohort 1	Smit	2024	Multinational	No	II	Open label	ONE	nonsquamous NSCLC	HER2-overexpressing	Trastuzumab deruxtecan	6.4 mg/kg/Q3W	49	4.1(1.4—7.1)	63.0 (58.0–68.0)	30 (61)	16 (33)	0–1	17 (35)	≥ 1	ORR(ICR)	DOR, DCR, PFS, OS, AE
DESTINY-Lung01 Cohort 1A											5.4 mg/kg/Q3W	41	5.5(1.4—8.7)	62.0 (56.0–66·0)	22 (54)	9 (22)	0–1	12 (29)
DESTINY-Lung01 [28] Cohort 2	Li	2022							HER2-Mutant		6.4 mg/kg/Q3W	91	13.1(0.7–29.1)	60 (29–88)	31(34)	52 (57)	0–1	33 (36)	2 (0–7)	ORR(ICR)	DOR, PFS, OS, AE, biomarkers of response
HERTHENA-Lung01 [29] (NCT04619004)	Yu	2023	Multinational	YES	II	Open label	TWO	EGFR-Mutant NSCLC	EGFR-Mutant	Patritumab Deruxtecan	5.6 mg/kg/Q3W	225	18.9(14.9–27.5)	64.0 (37–82)	93 (41)	144 (64)	0–2	115 (51.1)	3.0(1–11)	ORR(ICR)	DOR(BICR), ORR and DOR (investigator), PFS, OS, Safety, etc
HORIZON-Lung [20](NCT04818333)	LI	2025	Multicenter (CHINA)	No	II	Open label	ONE	NSCLC	HER2-Mutant	Trastuzumab rezetecan	4.8 mg/kg/Q3W	94	8.7(7.0 ~ 10.4)	57.5 (50–64)	42 (45)	60 (64)	0–1	24 (26)	2 (1–2)	ORR(ICR)	ORR(investigator), DCR, DoR, PFS, OS, Safety
TROPION-Lung05 [21](NCT04484142)	Sands	2025	Multinational	No	II	Open label	ONE	NSCLC	Actionable Genomic Alterations	Datopotamab Deruxtecan	6.0 mg/kg/Q3W	137	15.2(9.1–20.5)	61 (29–79)	54 (39)	76 (55.5)	0–1	70 (51.1)	3 (1–9)	ORR(ICR)	DOR, DCR, clinical benefit rate, PFS, time to response, OS, safety
TROPION-Lung01 [15] (NCT04656652)	Ahn	2025	Multinational	YES	III	Open label	TWO	NSCLC	Unselected	Datopotamab Deruxtecan	6.0 mg/kg/Q3W	299	PFS:10.9(9.8—12.5) OS:23.1(22.0–24.8)	63.0(26.0 −84.0)	183 (61.2)	61 (20.4)	0–1	79 (26.4)	≥ 1	PFS and OS(ICR)	ORR, DOR, Safety
EVOKE-01 [14](NCT05089734)	Paz-Ares	2024	Multinational	YES	III	Open label	TWO	NSCLC	Unselected	Sacituzumab Govitecan	10 mg/kg/d1,d8/Q3W	299	12.7(6.0–24.0)	66 (31–84)	194 (64.9)		0–2	35 (11.7)	≥ 1	OS(ICR)	PFS(investigator), ORR, DCR, DOR, Safety
UMIN000017709 [30]	Hotta	2018	Multicenter (JAPAN)	NO	II	Open label	ONE	NSCLC	HER2-Positive	Trastuzumab Emtansine	3.6 mg/kg/Q3W	15	9.2	67 (45–77)	7 (47%)	10 (67%)	0–2		4 (1–7)	ORR(ICR)	Safety, OS, PFS
DESTINY-Lung02 [31](NCT04644237)	Goto	2023	Multinational	YES	II	Double-blind	TWO	NSCLC	HER2-Mutant	Trastuzumab deruxtecan	5.4 mg/kg/Q3W	102	11.5(1.1–20.6)	59.4 (31–84)	37 (36.3)	55 (53.9)	0–1	35 (34.3)	2 (1–12)	ORR(ICR)	ORR(investigator), DCR, PFS, OS and exploratory end point
											6.4 mg/kg/Q3W	50	11.8(0.6–21.0)	61.3 (28–86)	16 (32.0)	29 (58.0)	0–1	22 (44.0)	2 (1–7)
KL264-01 [32] cohort 3A(NCT04152499)	ZHAO	2025	Multinational	No	II	Open label	ONE	NSCLC	Unselected	Sacituzumab tirumotecan	5.0 mg/kg/Q2W	43	26.0(24.2–26.6)	58 (44–74)	27 (63)	NR	0–1	6 (14)	2 (1–10)	ORR(investigator)	safety, tolerability, pharmacokinetics, immunogenicity, DCR, DOR, PFS, OS and correlation between TROP2 expression, etc
SKB264-II-08 [32](NCT05631262) Cohort 1	ZHAO	2025	Multicenter (CHINA)	No	II	Open label	ONE	NSCLC	EGFR-mutant	Sacituzumab tirumotecan	5.0 mg/kg/Q2W	32	12.7(12.3–13.0)	58 (38–73)	14 (44)	NR	0–1	3 (9)	3 (1–5) EGFR TKIs and platinum-based chemotherapy	ORR and Safety(investigator)	pharmacokinetics, immunogenicity, DCR, DOR, PFS, OS and correlation between TROP2 expression and treatment outcomes
Cohort 2												32		57.5 (41–74)	14 (44)	NR	0–1	5 (16)	1 (1–2) TKIs but naive to chemotherapy
LUNG-MAP Sub-study [33](NCT03574753)	Waqar	2021	Multinational	No	II	Open label	ONE	Squamous Cell Lung Cancer	ceMET-positive	Telisotuzumab Vedotin	2.7 mg/kg/Q3W	23	NR	65.3(57.7–81.6)	13(57)	1(4)	0–1	2(9)	0–4	ORR(investigator)	PFS, OS, response within cohort, duration of response, and toxicities
NCT02289833 [34]	Peters	2019	Multinational	No	II	Open label	ONE	NSCLC	HER2-overexpressing	Trastuzumab Emtansine	3.6 mg/kg/Q3W	49	IHC2 + :23.1(0.9–26.7) IHC3 + :18.4(1.0–25.1)	61 (36–80)	29(59.1)	10(20.4)	0–1	4(8.2)	≥ 1	ORR(investigator)	PFS, DOR, OS and Safety(investigator)
JapicCTI-194620 [35]	Iwama	2022	Multicenter (JAPAN)	No	II	Open label	ONE	Lung adenocarcinoma	EGFR Exon 20 Insertion Mutations	Trastuzumab Emtansine	3.6 mg/kg/Q3W	22	8.0(0.3 ~ 24.7)	61.5 (35–80)	12 (54.5)	11(50.0)	0–2	9 (40.9)	1–2	ORR(investigator)	PFS, DOR, OS and Safety(investigator)
NCT02675829 [36]	Li	2018	Single-center(United States)	No	II	Open label	ONE	lung adenocarcinoma	HER2-Mutant	Trastuzumab Emtansine	3.6 mg/kg/Q3W	18	10	64 (47–74)	5(27.8)	7 (39)	KPS	2(11.1)	2(0–4)	ORR(investigator)	PFS, and toxicities(investigator)
OptiTROP-Lung03 [16] (NCT05631262)	Fang	2025	Multicenter (CHINA)	YES	II	Open label	TWO	nonsquamous NSCLC	EGFR-mutant	sacituzumab tirumotecan	5.0 mg/kg/Q2W	91	12.2 (9.5–15.6)	57 (37–75)	38 (42)	NR	0–1	18 (20)	≥ 1	ORR(ICR)	ORR(investigator), DCR, PFS, DOR, OS, Safety(ICR + investigator), correlation between TROP2 expression and treatment outcomes

Abbreviations: ADC Antibody–drug conjugate, ECOG PS Eastern Cooperative Oncology Group Performance Status, KPS Karnofsky Performance Status, N Total number of patients in arms containing ADC, EGFR Epidermal Growth Factor Receptor, TKIs Tyrosine Kinase Inhibitors, HER2 Human Epidermal Growth Factor Receptor 2, TROP2 Trophoblast Cell-Surface Antigen 2, NSCLC Non-small cell lung cancer, ORR Objective response rate, DCR Disease Control Rate, PFS Progression-free survival, OS Overall survival, DOR Duration of Response, AE Adverse Event, ICR Independent Committee Review, NR Not Reported

Table 2.

Characteristics of the ADCs used within the included clinical trials

ADC	Target	Linker	Payload	DAR	Research-based	FDA/EMA Authorization Details (Year of Approval, Therapeutic Indications and Dosage)
Telisotuzumab Vedotin(Teliso-V)	c-MET	VC(cleavable)	monomethyl auristatin E cytotoxic payload(MMAE) (Microtubule inhibitor)	3.1	LUMINOSITY [19]	U.S. FDA, May 2025(Accelerated Approval) First targeted therapy for adult patients with pretreated, locally advanced or metastatic non-squamous, EGFR wild-type NSCLC whose tumors have high c-Met protein expression (IHC 3 + in ≥ 50% of tumor cells)
Trastuzumab deruxtecan (T-DXd)	HER2	GGFG (cleavable)	topoisomerase I inhibitor(DXd)	8	DESTINY-Lung02 [31](Primary), DESTINY-Lung01 [27](Secondary)	U.S. FDA, August 2022(Accelerated Approval) For the treatment of adult patients with unresectable or metastatic HER2-mutant NSCLC who have received prior systemic therapy
Patritumab Deruxtecan (Patritu-DXd)	HER3	GGFG (cleavable)	topoisomerase I inhibitor(DXd)	8
Trastuzumab rezetecan(SHR-A1811)	HER2	GGFG (cleavable)	DNA topoisomerase I inhibitor	6	HORIZON-Lung [20]	China NMPA, May 2025(Conditional Approval) First therapy approved for adult patients with locally advanced or metastatic HER2-mutant NSCLC who have received at least one prior line of systemic therapy
Datopotamab Deruxtecan (Dato-DXd)	TROP-2	GGFG (cleavable)	topoisomerase I inhibitor(DXd)	4	TROPION-Lung05 [21] (Primary), TROPION-Lung01 [15](Secondary)	U.S. FDA, June 2025(Accelerated Approval) For the treatment of adult patients with locally advanced or metastatic EGFR-mutant NSCLC who have received prior EGFR-targeted therapy and platinum-based chemotherapy
Sacituzumab Govitecan (SG)	Trop-2	pH-sensitive carbonate (cleavable)	topoisomerase I inhibitor SN-38	7.6
Trastuzumab Emtansine(T-DM1)	HER2	MCC (noncleavable)	DM1- Emtansine (Microtubule inhibitor)	3.5
Sacituzumab tirumotecan(sac-TMT)	TROP2	unrevealed (cleavable)	topoisomerase I inhibitor (Tirumotecan)	8	OptiTROP-Lung03 [16]	China NMPA, March 2025 For the treatment of adult patients with EGFR mutation-positive locally advanced or metastatic non-squamous NSCLC that has progressed following therapy with EGFR-TKIs and platinum-containing chemotherapy

Abbreviations: ADC Antibody–drug conjugate, FDA Food and Drug Administration, EMA European Medicines Agency, NMPA National Medical Products Administration, NSCLC Non-Small Cell Lung Cancer, HER2 Human Epidermal Growth Factor Receptor 2, TROP2 Trophoblast Cell-Surface Antigen 2, EGFR Epidermal Growth Factor Receptor, TKIs Tyrosine Kinase Inhibitors, VC ValineCitrulline, GGFG Glycine-Glycine-Phenylalanine-Glycine, MCC Maleimidomethyl Cyclohexane1-Carboxylate, MMAE Monomethyl auristatin E

Quality assessment of included studies

A visual representation of the risk of bias assessment for the included randomized trials is shown in Fig. 2. The DESTINY-Lung02 [31] study demonstrated a low risk of bias across all domains, supported by robust randomization, effective blinding, objective outcome measurement criteria, and thorough data management. The EVOKE-01 [14], TROPION-Lung01 [15], and OptiTROP-Lung03 [16] studies, although randomized, were open-label without complete blinding. However, the use of an independent review committee (IRC) for blinded outcome assessment mitigated the potential performance bias inherent in the open-label design. Additionally, OptiTROP-Lung03 involved crossover therapy, which introduced deviations from the intended interventions. In contrast, the HERTHENA-Lung01 [29] study raised concerns regarding bias, primarily due to its open-label design and unclear description of the randomization method. A visual representation of the risk of bias assessment for the included non-randomized clinical trials is presented in Fig. 3. All studies demonstrated an overall moderate risk of bias. As single-arm studies, they all lacked randomization and utilized an open-label design. Additionally, the DESTINY-Lung01 [27] study exhibited selection bias, as patients were sequentially enrolled based on investigator selection, and there were deviations from the planned intervention protocol after enrollment. The UMIN000017709 [30] and JapicCTI-194620 [35] studies enrolled patients from specific hospitals, featured small sample sizes, and were conducted in only a few Japanese institutions. Furthermore, the JapicCTI-194620 [35] study employed unblinded outcome assessment, although the outcome measures themselves were objective. The LUMINOSITY [19] study showed bias due to missing data, with a rate of loss to follow-up or unmeasured data reaching 6.4%. The LUNGMAP Sub-study [33] and NCT02289833 [34] studies had deviations from the planned interventions, missing data, and unblinded outcome assessment. Notably, the LUNGMAP Sub-study [33] study also exhibited selective reporting bias, as ORR, PFS, and OS were not reported separately for Cohort 1 and Cohort 2. The KL264-01 [32] and SKB264-II-08 [32] studies were rated primarily due to their enrollment of specific Chinese patient populations, small sample sizes, and unblinded outcome assessment. Finally, the NCT02675829 [36] study was the only single-center trial, characterized by a small sample size and unblinded outcome assessment.

Fig. 2.

Risk of bias traffic light (above) and summary (below) plots of Revised Cochrane risk-ofbias tool for randomized trials (RoB 2.0) assessments for the randomized controlled trials (RCTs) included in the review. Each domain is judged as having low, some concerns or high risk of bias

Fig. 3.

Risk of bias trafffc light (above) and summary (below) plots of Risk of Bias in Nonrandomized Studies of Interventions (ROBINS-I) assessments for the non-randomized clinical trials included in the review. Each domain is judged as having low, moderate, serious, or critical risk of bias

Efffcacy

ORR

The pooled analysis showed that the ORR for ADC monotherapy as a later-line treatment for NSCLC ranged from 6.7% to 73.4%, with an overall pooled ORR of 34% (95% CI: 26%−42%) (Fig. 4). Due to significant heterogeneity among the study results (I² = 91.7%), meta-regression and subgroup analyses were performed. Meta-regression indicated that the study publication year was not a source of heterogeneity (p = 0.1064), while sample size was a significant source of heterogeneity (p < 0.0001). To further explore the sources of heterogeneity, subgroup analyses were conducted.

Fig. 4.

Objective response rate of antibody–drug conjugates monotherapy in pretreated patients with advanced or metastatic non-small cell lung cancer

ORR of randomized versus non-randomized studies

Subgroup analysis by study type (randomized vs. non-randomized) revealed no significant difference in ORR (Supplementary Figure S1). Due to high heterogeneity among the study results (I² = 91.7%), a random-effects model was used (p = 0.8362). The pooled effect sizes were similar between the two subgroups: the ORR for non-randomized studies was 33% (95% CI: 21%−45%), compared to 35% (95% CI: 19%−53%) for randomized studies. The high heterogeneity might also be influenced by study design, sample characteristics, or intervention details.

ORR of different pathological subtypes

The results showed that ADC therapy yielded a higher ORR in non-squamous NSCLC compared to squamous NSCLC (Figure S2). The ORR in the squamous cell carcinoma subgroup (9%, 95% CI: 2%—29%) was lower than that in the unspecified NSCLC subgroup (33%, 95% CI: 22%—45%), which in turn was lower than that in the non-squamous NSCLC subgroup (38%, 95% CI: 30%—47%). Due to high heterogeneity (I² = 91.7%), a random-effects model was applied, revealing a significant difference between the subgroups (p = 0.0082). However, the result for the squamous NSCLC subgroup is based on only one study and is thus unstable, requiring validation from more studies.

ORR for different mutation types

Subgroup analysis demonstrated that compared to the ORR in the Non-mutated subgroup (21%, 95% CI: 13%—31%), the ORR was higher in the EGFR-Mutant subgroup (35%, 95% CI: 17%—56%) and the AGAs subgroup (36%, 95% CI: 28%—44%), with the highest ORR observed in the HER2-Mutant subgroup (55%, 95% CI: 42%—68%) (Figure S3). Due to high heterogeneity (I² = 91.7%), a random-effects model was used, indicating a statistically significant difference in ORR among the different mutation types (p < 0.0001).

ORR for different ADC targets

There was no statistically significant difference in efficacy among different ADC targets (Figure S4). The analysis included four ADC targets: c-MET, HER2, HER3, and TROP-2. Although drugs targeting HER2 showed a higher ORR (39%, 95% CI: 26%—53%) compared to those targeting c-MET (22%, 95% CI: 11%—35%), HER3 (30%, 95% CI: 23%—36%), and TROP-2 (31%, 95% CI: 22%—40%), the high heterogeneity (I² = 91%) led to a non-significant difference between subgroups using the random-effects model (p = 0.3204).

ORR for different ADC drugs

Subgroup analysis revealed a statistically significant difference in efficacy among different ADC drugs (Figure S5). SHR-A1811 demonstrated the highest ORR at 73% (95% CI: 63%–82%), although this finding is supported by only one study. T-DXd also showed a relatively high ORR (45%, 95% CI: 35%−55%). In contrast, Teliso-V (22%, 95% CI: 11%−35%), SG (14%, 95% CI: 9%−18%), and T-DM1 (21%, 95% CI: 7%−40%) showed lower ORRs. Due to high heterogeneity (I² = 91.7%), a random-effects model was applied, indicating a significant difference in ORR among the different drugs (p < 0.0001).

DCR and 1-Year OS Rate

The pooled analysis showed DCR ranged from 33 to 98%, with an overall pooled DCR of 76% (95% CI: 68%−83%) (Fig. 5). Due to significant heterogeneity (I² = 90.6%), a meta-regression was performed, which identified both the study publication year (p = 0.0038) and sample size (p < 0.0001) as sources of heterogeneity. The 1-year OS rate ranged from 46 to 80%, with an overall pooled rate of 66% (95% CI: 57%−74%) (Fig. 6). Due to significant heterogeneity (I² = 87.7%), a meta-regression was conducted, which identified sample size as a source of heterogeneity (p < 0.0001).

Fig. 5.

Disease Control rate (DCR) of antibody–drug conjugates monotherapy in pretreated patients with advanced or metastatic non-small cell lung cancer

Fig. 6.

The 1-year overall survival rate of antibody–drug conjugates monotherapy in pretreated patients with advanced or metastatic non-small cell lung cancer

DOR, PFS and OS

A pooled analysis of all studies reporting DOR, PFS, and OS yielded the following results (Figs. 7, 8 and 9). The median DOR was 6.6 months (95% CI: 5.7—7.8 months), with moderate heterogeneity observed among the studies (I² = 40%). Regarding censored data, the censoring rates ranged from 15 to 35%. All censored data were processed uniformly using the Kaplan–Meier method and included in sensitivity analyses. The median DOR was 6.6 months for the full set and 7.0 months after excluding the three studies with undefined censoring, indicating minimal impact and no statistically significant difference (Cox model P = 0.901) (Figure S6-7). The mPFS was 4.9 months (95% CI: 3.9—6.0 months). Due to high heterogeneity (I² = 86.4%), a meta-regression was performed, which identified sample size as a source of heterogeneity (p = 0.0128). The censoring rates ranged from 10 to 45% among these studies. The overall median PFS was 4.9 months and it increased to 5.3 months after excluding the five studies with undefined censoring. The comparison showed a very limited impact, but there was a statistically significant difference (Cox model P = 0.043) (Figure S8-9). The mOS was 12.4 months (95% CI: 10.2—15.2 months), with significant heterogeneity (I² = 76.1%). Meta-regression analysis indicated that neither publication year (p = 0.1615) nor sample size (p = 0.3088) were significant sources of heterogeneity for OS. A subsequent meta-regression analysis specifically examining the seven drugs included in the OS analysis (Dato-DXd, Patritu-DXd, Sac-TMT, SG, T-DM1, T-DXd, Teliso-V) identified Sac-TMT as a source of heterogeneity (p = 0.0475). Among these studies, the censoring rates ranged from 35 to 60%. The median OS was 12.8 months for the full set and 12.8 months after excluding the four studies with undefined censoring, indicating minimal impact and no statistically significant difference (Cox model P = 0.15) (Figure S10-11).

Fig. 7.

Duration of Response of antibody–drug conjugates monotherapy in pretreated patients with advanced or metastatic non-small cell lung cancer

Fig. 8.

Progression-Free Survival of antibody–drug conjugates monotherapy in pretreated patients with advanced or metastatic non-small cell lung cancer

Fig. 9.

Overall Survival of antibody–drug conjugates monotherapy in pretreated patients with advanced or metastatic non-small cell lung cancer

Moreover, We performed a multivariate meta-regression analysis to examine the influence of follow-up duration on the aforementioned effect sizes. The results revealed no statistically significant effects of follow-up duration on DOR (P = 0.322), mPFS (P = 0.301), or mOS (P = 0.301).

Adverse events

TRAEs

A total of 11 studies, comprising 13 cohorts, reported all-grade TRAEs with an incidence ≥ 10%. The overall incidence was 95% (95% CI: 92% to 97%) (Fig. 10a), encompassing the following drugs: Dato-DXd, Patritu-DXd, Sac-TMT, SG, SHR-A1811, T-DXd, and Teliso-V. Given the significant heterogeneity (I² = 83.9%), the random-effects model revealed a statistically significant difference in incidence rates among the ADC drugs (p < 0.0001), with Sac-TMT having the highest incidence at 98% (Figure S12). A total of 13 studies, comprising 15 cohorts, reported grade ≥ 3 TRAEs with an incidence ≥ 10%. The overall incidence was 43% (95% CI: 36% to 51%) (Fig. 10b). All eight ADC drugs were included in this analysis. Due to significant heterogeneity (I² = 90.9%), the analysis revealed a statistically significant difference in incidence among the different drugs (p < 0.0001), with Sac-TMT again showing the highest incidence at 64% (Figure S13).

Fig. 10.

The incidence of treatment-related adverse events (TRAEs) of included studies (a) All-grade TRAEs; b Grade ≥ 3 TRAEs

We specifically analyzed the incidence of ILD. A total of 11 studies, comprising 13 cohorts, reported all-grade ILD. The overall incidence was 10% (95% CI: 5% to 15%) (Fig. 11a), involving the following drugs: Dato-DXd, Patritu-DXd, SHR-A1811, T-DXd, Teliso-V, Sac-TMT, and T-DM1. Due to significant heterogeneity (I² = 84.4%), the random-effects model showed a statistically significant difference in incidence among the different drugs (p < 0.0001), with T-DXd exhibiting the highest incidence at 18% (Figure S14). A total of 12 studies, comprising 14 cohorts, reported grade ≥ 3 ILD. The overall incidence was 2% (95% CI: 1% to 3%) (Fig. 11b). Moderate heterogeneity was observed among the study results (I² = 42.4%). Subgroup analysis indicated a statistically significant difference in incidence among the different drugs (p = 0.0198), with Teliso-V showing the highest incidence at 5% (Figure S15).

Fig. 11.

The incidence of treatment-related interstitial lung disease of included studies (a) All-grade; b Grade ≥ 3

Common adverse events

This analysis pooled all-grade TRAEs with an overall incidence ≥ 10%. However, since the studies by Li et al. [28] and Goto et al. [31] reported TRAEs with an incidence exceeding 20%, and the study by Ahn et al. [15] reported TRAEs exceeding 15%, the incidence of some AEs might be underestimated. All-grade TRAEs, ranked from highest to lowest incidence, are shown in Fig. 12a. The gastrointestinal system was most frequently involved, with nausea being the most common event, followed by neutropenia and anemia from the hematologic system. Grade ≥ 3 TRAEs, ranked from highest to lowest incidence, are shown in Fig. 12b. The hematologic system was most frequently affected, primarily neutropenia and leukopenia, followed by pneumonia from the respiratory system.

Fig. 12.

Incidence of the most common (≥ 10%) adverse events in included studies: a All-grade; b Grade ≥ 3

Eleven studies reported a total of 1,628 instances of all-grade TRAEs. Common all-grade TRAEs included gastrointestinal disorders (nausea, stomatitis, decreased appetite, diarrhea, vomiting, constipation), hematologic disorders (neutropenia, anemia, leukopenia, thrombocytopenia, lymphopenia), respiratory disorders (ILD, pneumonia, pneumonitis, dyspnea, cough, upper respiratory tract infection), metabolic disorders (hypertriglyceridemia, hyperglycemia, hypokalemia, hypocalcemia, hyponatremia, hypoalbuminemia), general disorders (decreased weight, asthenia, fatigue, rash, dizziness, peripheral edema), and other (alopecia, peripheral neuropathy, increased alanine aminotransferase, increased aspartate aminotransferase, increased blood creatinine). Thirteen studies reported a total of 765 instances of grade ≥ 3 TRAEs. Common grade ≥ 3 TRAEs included hematologic disorders (neutropenia, anemia, leukopenia, thrombocytopenia, lymphopenia), respiratory disorders (pneumonia, ILD, pneumonitis, dyspnea), gastrointestinal disorders (stomatitis, diarrhea, nausea, vomiting, decreased appetite), general disorders (fatigue, dizziness, asthenia, peripheral edema, decreased weight, rash), metabolic disorders (hypokalemia, hypertriglyceridemia, hypocalcemia, hyponatremia), and other (peripheral neuropathy, increased alanine aminotransferase, increased aspartate aminotransferase, alopecia).

Related to drug discontinuation and death

ADC-related treatment discontinuation and mortality were summarized in Table 3. The overall treatment discontinuation rate was 9% (95% CI: 5–13%) (Figure S16). The significant heterogeneity (I² = 86%) was attributed to sample size based on meta-regression findings (p < 0.0001). Pneumonitis and ILD were the most frequent respiratory AEs leading to discontinuation. No treatment-related discontinuations were reported for Patritu-DXd; discontinuations for SG were associated with gastrointestinal events (e.g., diarrhea, nausea, and abdominal pain), whereas those for Sac-TMT were linked to hematologic issues—none of which involved the respiratory system. In addition to respiratory events, T-DM1 discontinuations were also linked to infusion-related reactions, influenza, and brain hemorrhage, while Dato-DXd also reported discontinuations related to oral mucositis/stomatitis. The drug-related mortality rate, defined by grade 5 TRAEs, was 0.5% (95% CI: 0.1–1.0%). (Figure S17). Respiratory AEs—including pneumonitis, pneumonia, ILD, and respiratory failure—were the most common causes of death. Other fatal events included gastrointestinal perforation, sepsis, febrile neutropenia, neutropenic colitis, septic shock, hematologic toxicity, bronchopulmonary hemorrhage, and brain hemorrhage. No drug-related deaths were reported for T-DM1 or SHR-A1811.

Table 3.

Adverse events related to drug discontinuation and death

Author	Year	ADC	N	Drug-related discontinuation		Drug-related death
				N	AE	N	AE
Camidge	2024	Telisotuzumab Vedotin	172	37	most common: Pneumonitis (13), peripheral sensory neuropathy (12), peripheral sensorimotor neuropathy (4), ILD (2)	1	ILD(1)
Smit(6·4 mg/kg)	2024	Trastuzumab deruxtecan	49	12	Most common: pneumonitis (7)	1	Pneumonitis(1)
Smit(5·4 mg/kg)	2024	Trastuzumab deruxtecan	41	5	most common: pneumonitis (2)	1	ILD/pneumonitis(1)
LI	2022	Trastuzumab deruxtecan	91	23	pneumonitis(12),ILD(5)	2	ILD(2)
Yu	2023	Patritumab Deruxtecan	225	16	NR	4	Pneumonitis(1), pneumonia(1), GI perforation(1), and respiratory failure(1)
Li	2025	Trastuzumab rezetecan	94	1	ILD(1)	0	-
Sands	2025	Datopotamab Deruxtecan	137	7	Oral mucositis/stomatitis(1), ILD/pneumonitis(4)	0	-
Ahn	2025	Datopotamab Deruxtecan	297	24	most common: pneumonitis(13), ILD (3)	3	ILD/pneumonitis(2); sepsis(1)
Paz-Ares	2024	Sacituzumab Govitecan	296	20	Diarrhea(1), nausea(1), abdominal pain (1)	4	febrile neutropenia(1), neutropenic colitis(1), sepsis(1), septic shock(1)
Hotta	2018	Trastuzumab Emtansine	15	2	Interstitial pneumonia(1), aprolonged thrombocytopenia (1)	0	-
Goto5.4 (mg/kg)	2023	Trastuzumab Deruxtecan	101	14	ILD (6), pneumonitis (5)	1	ILD
Goto6.4(mg/kg)	2023	Trastuzumab Deruxtecan	50	10	ILD(4), pneumonitis (2)	1	ILD
Zhao	2025	Sacituzumab tirumotecan	107	1	Hematological toxicity(1)	1	Hematological toxicity(1)
Waqar	2021	Telisotuzumab Vedotin	23	4	Anemia (1), hyponatremia (1), hypophosphatemia (1), and peripheral sensory neuropathy (1)	3	bronchopulmonary hemorrhage(1),pneumonitis(2)
Peters	2019	Trastuzumab emtansine	49	2	infusion-related reaction(1), influenza(1)	0	-
Iwama	2022	Trastuzumab Emtansine	22	3	Interstitial pneumonia (1), appetite loss (1), brain haemorrhage (1)	1	brain haemorrhage (1)
Li	2018	Trastuzumab Emtansine	18	0	-	0	-
Fang	2025	Sacituzumab tirumotecan	91	0	-	0	-

Abbreviations: ADC Antibody–drug conjugate, N Number of patients, AE Adverse Event, ILD Interstitial Lung Disease, NR Not Reported

Quality assessment of research results

This study conducted sensitivity and publication bias analyses on outcomes related to ADC therapy, including efficacy, TRAEs, discontinuation rates, and mortality. The stability and reliability of all outcomes was demonstrated through sensitivity analyses, in which the sequential exclusion of individual studies had minimal impact on the pooled proportions, thereby confirming the robustness of the original estimates and the lack of undue influence from any single study (Figure S18-29).

Egger's test did not reveal significant publication bias for the following pooled outcomes: ORR (p = 0.3175), DCR (p = 0.9855), PFS (p = 0.3872), OS (p = 0.8777), DOR (p = 0.6988), all-grade TRAEs (p = 0.1670), grade ≥ 3 TRAEs (p = 0.6321), all-grade ILD (p = 0.35), grade ≥ 3 ILD (p = 0.8351), treatment discontinuation rate (p = 0.6171), and mortality rate (p = 0.4032). The corresponding funnel plots were provided in the supplementary materials (Figure S30-S40). Significant risk of publication bias was detected for the pooled 1-year OS rate (p = 0.0040), as shown in the corresponding funnel plot (Figure S41). The Trim-and-Fill method was used to adjust the effect size due to publication bias, estimating the number of missing studies to be 4. The adjusted 1-year OS rate was 53.76% (Figure S42).

Discussion

ADCs offer a novel therapeutic approach for NSCLC, demonstrating significant advantages and proven efficacy due to their high selectivity and potent antitumor activity [22]. However, AEs associated with ADCs are unavoidable, making it essential to monitor ADC-induced AEs and investigate their incidence. Currently, quantitative assessments of related efficacy and adverse reactions remain limited. To address this issue, we conducted a meta-analysis of 16 relevant studies. To our knowledge, this represents the most recent and comprehensive analysis evaluating the efficacy and TRAEs of ADCs in subsequent-line therapy for NSCLC.

Currently, docetaxel-based chemotherapy served as the standard subsequent-line treatment for NSCLC, with an overall ORR of 3%−14%, mPFS of 2.8–4.5 months, mOS of 10–12 months, and a 1-year OS rate of 32%, accompanied by significant toxicity [2, 37, 38]. Our study indicated that ADC monotherapy demonstrated an ORR of 33%, a mPFS of 4.9 months, a mOS of 12.4 months, and a 1-year OS rate of 54% in an unselected population of pretreated NSCLC patients, potentially representing a novel therapeutic option for later-line treatment. In addition, the phase 3 RATIONALE-303 trial demonstrated statistically significant and clinically meaningful improvement in OS with tislelizumab monotherapy compared to docetaxel in the second-line treatment of NSCLC (mOS: 17.2 months vs. 11.9 months; hazard ratio [HR] = 0.64, p < 0.0001) [39]. Notably, although the mOS observed with ADCs in our analysis is lower than that reported for tislelizumab, the RATIONALE-303 trial focused on second-line treatment, whereas the majority of the studies in our meta-analysis were conducted in third-line or later treatment. Furthermore, results from two phase III trials, EVOKE-01 [14] and TROPION-Lung01 [15], indicated that ADCs did not demonstrate a survival benefit over docetaxel. However, the phase II OptiTROP-Lung03 trial [16] reported multifaceted survival benefits of SAC-TMT compared to docetaxel in an EGFR-mutant population. These findings underscored the necessity of selecting specific populations for ADC therapy. Accordingly, we performed subgroup analyses of the primary outcome, ORR.

First, subgroup analysis revealed no significant difference in ORR among different ADC targets. However, notable disparities were evident among various ADC drugs, with T-DXd and SHR-A1811 exhibiting the highest ORRs at 45% and 73%, respectively. Although both T-DM1 and the two drugs mentioned above are HER2-targeted agents, T-DM1 carries an emtansine payload, a microtubule inhibitor that is effective but lacks membrane permeability. In contrast, T-DXd and SHR-A1811 utilize a topoisomerase I inhibitor payload, which exhibits greater efficacy and enhanced membrane permeability [12]. Furthermore, the two drugs feature cleavable linkers (enabling a bystander effect) and a higher DAR. Collectively, these properties result in enhanced tumor penetration, bystander killing, and overall cytotoxicity, translating into superior clinical outcomes, particularly in patients with HER2-low expression or heterogeneous HER2-positive tumors [18].

Second, subgroup analysis suggested that squamous histology might be associated with poorer prognosis with ADC therapy. The ORR for ADC-treated nonsquamous NSCLC was 38%, significantly higher than the 9% observed in squamous NSCLC. However, the data on squamous NSCLC were derived from only one clinical study, rendering the finding unstable and necessitating further validation in larger squamous cell cohorts. Remarkably, the therapeutic efficacy in this study also varied depending on prior exposure to immune checkpoint inhibitors (ICIs), with ICI-naïve patients exhibiting better responses than ICI-pretreated individuals [33]. Additionally, subgroup analyses from the TROPION-Lung01 [15] and ICARUS-Lung01 [40] trials consistently reported higher ORR in nonsquamous versus squamous NSCLC (31.2% vs. 9.2% and 30.5% vs. 5%, respectively), further supporting the potentially reduced efficacy of ADCs in squamous NSCLC.

Finally, the subgroup analysis demonstrated that ADC therapy yielded significantly higher ORRs in NSCLC patients with EGFR mutations, AGAs, or HER2 mutations compared to those without mutations (35%, 36%, 55%, and 21%, respectively; P < 0.0001). While targeted therapy remains the first-line choice for EGFR-mutant NSCLC, drug resistance is inevitable. Despite recent innovative advancements in second-line clinical trials such as ORIENT-31, AK112-301, and MARIPOSA-2, the PFS remains limited to 5.5–7.2 months [41], with no superior subsequent treatments available. Retrospective and real-world analyses further indicate that the PFS for salvage therapy beyond the third line is merely 2.8–3.3 months [42]. Currently, the primary ADC targets for EGFR mutation resistance are HER3 and TROP-2. Following treatment with third-generation EGFR tyrosine kinase inhibitors (EGFR-TKIs), blockade of the EGFR/AKT signaling pathway and suppression of AKT phosphorylation lead to upregulation of HER3 protein expression in cancer cells. HER3 expression is observed in 96.4%−100% of patients, enhancing the anticancer activity of HER3-DXd [41, 43]. Both the phase 2 HERTHENA-Lung01 [29] and phase 1 U1027 studies [44] have provided compelling evidence of the efficacy of HER3-DXd in EGFR-mutated NSCLC patients who have failed EGFR-TKIs and platinum-based chemotherapy (PBC). Both studies demonstrated clinical benefits across multiple subgroups, including those characterized by tumor HER3 expression, EGFR-TKIs resistance mechanisms, and the presence of brain metastases. Additionally, the EGFR-HER3 bispecific ADC zalontamab brengitecan has also exhibited significant efficacy in EGFR-mutated NSCLC, independent of HER3/EGFR expression levels [41]. TROP2 is overexpressed in over 50% of lung adenocarcinomas and squamous cell carcinomas [45], and is associated with poor patient prognosis [46]. Results from the global multicenter KL264-01 [32] and TROPION-Lung05 [21] studies reported Sac-TMT and Dato-DXd achieved ORRs of 55% and 43.6%, respectively, in the EGFR-mutated subgroup. Moreover, the OptiTROP-Lung03 [16] study, a randomized controlled trial comparing Sac-TMT with docetaxel, reported an ORR of 45% versus 16%, and a mPFS of 6.9 months versus 2.8 months (HR 0.30, P < 0.001). These findings collectively suggest that EGFR mutations may represent a promising target population for ADC-based therapy. Beyond EGFR mutations, The TROPION-Lung05 [21] study reported encouraging antitumor activity of Dato-DXd in heavily pretreated NSCLC with various AGAs, as evidenced by an ORR of 35.8%, a DoR of 7.0 months, a PFS of 5.4 months, and an OS of 13.6 months. Despite this promise, confirmation in larger phase III RCTs is required. Moreover, the HER2-mutant subgroup demonstrated the highest ORR, reaching 55%. HER2 activation in NSCLC occurs through three primary mechanisms: gene mutation (1%–4%), gene amplification (2%–5%), and protein overexpression (2%–30%), each associated with distinct prognostic and predictive implications [47]. Current targets primarily focus on HER2 mutations and protein overexpression. Studies have shown that SHR-A1811, T-DXd, and T-DM1 improve outcomes in HER2-positive NSCLC [20, 47, 48]. Specifically, in HER2-mutant NSCLC, Phase II trials reported encouraging ORRs of 73%, 50%, and 62%, respectively [47], results closely aligned with our findings. Moreover, both SHR-A1811 and T-DXd exhibited consistent and robust efficacy across all subgroups of HER2-mutant patients, regardless of prior HER2-TKI treatment or the presence of baseline brain metastases [20, 31]. In contrast, studies focusing on protein overexpression have yielded more modest outcomes. In DESTINY-Lung01 [27], T-DXd treatment in HER2-overexpressing patients led to a median DoR of only 6.2 months, notably lower than the 16.8 months observed in HER2-mutant patients in DESTINY-Lung02 [31]. Additionally, Peters et al. [34] reported an ORR of only 8.2% for T-DM1 in HER2-overexpressing NSCLC. In summary, these datas support the view that HER2 mutation may be more predictive of ADC response than HER2 protein overexpression alone. Preclinical experiments suggest that HER2-activating mutations may enhance receptor phosphorylation and ubiquitination by forming dimers with HER3, thereby promoting the internalization and intracellular uptake of receptor-ADC complexes, independent of protein expression levels [49]. This mechanism may underlie the notably high efficacy of these agents in patients with HER2-mutant NSCLC.

Distant metastasis, particularly brain metastasis, remains a predominant cause of mortality in NSCLC patients. ADCs including SHR-A1811, T-DXd, and Patritu-DXd (HER3-DXd) employ enzyme-cleavable GGFG linkers to release topoisomerase I inhibitors, producing potent bystander effects that enhance their antitumor activity in lung cancer brain metastases [50]. In the HORIZON-Lung study, SHR-A1811 demonstrated an ORR of 87.5% (21/24) in the brain metastasis subgroup [20]. Similarly, in the HERTHENA-Lung01 trial, Patritu-DXd achieved an ORR of 33.3% among 30 untreated EGFR-mutant NSCLC patients with brain metastases [29]. Furthermore, an exploratory pooled analysis of DESTINY-Lung01 and DESTINY-Lung02 presented at the 2023 ESMO congress reported an intracranial ORR of 50% for T-DXd at the 5.4 mg/kg every-3-weeks dosing regimen [22]. Collectively, these data highlight the substantial clinical potential of ADCs in treating lung cancer brain metastases.

ADCs demonstrated an overall favorable safety profile, with toxicities resembling those of conventional chemotherapy. These toxicities primarily arise from two mechanisms: payload-related toxicity, stemming from the premature release of the ADC payload in the bloodstream or tumor microenvironment, and target-related toxicity, resulting from the binding of ADCs to non-tumor cells expressing the target antigen [12, 51]. Our analysis revealed that among the ≥ grade 3 TRAEs of greatest clinical concern, neutropenia was the most frequently observed. In ADCs, payloads such as topoisomerase I inhibitors and microtubule inhibitors (e.g., MMAE) can suppress hematopoietic stem cell division, leading to impaired neutrophil production. Particularly, MMAE arrests cells in the M-phase of mitosis, preventing hematopoietic stem cells from proliferating into neutrophils. Beyond these direct cytotoxic effects, these payloads may also disrupt the bone marrow microenvironment (e.g., stromal cell function), further contributing to neutropenia [52]. Treatment discontinuation and fatal outcomes associated with ADCs are most commonly linked to pneumonitis and ILD. The higher incidence of pneumonitis and ILD in NSCLC underscores the particular vulnerability of this patient population to ADC-related pulmonary toxicity. This may be attributed to pre-existing impaired lung function, greater drug exposure in lung tissue, and prior therapies such as ICIs and taxane-based chemotherapy, which can predispose patients to pneumonitis or ILD [53]. Due to limited follow-up durations, clinical trials may not fully capture delayed AEs. Real-world evidence suggests that ADCs significantly increase the long-term risk of sepsis in cancer patients [54]. For instance, Paz-Ares et al. [14] reported two fatalities of sepsis and septic shock following treatment with SG. Additionally, ADCs may elevate the risk of neurotoxicity, which can lead to severe mortality, and have been associated with cardiovascular toxicity and peripheral neuropathy [55, 56]. Comparatively, TROP-2–targeted ADCs exhibit the lowest risk of cardiotoxicity [56]. In our included studies, no fatal cardiovascular or peripheral neuropathic events were reported. However, Teliso-V reported treatment discontinuations related to peripheral neuropathy, which was associated with its MMAE payload.

Based on our analysis of specific agents, we observed significant variations in the incidence of TRAEs across different agents. Sac-TMT exhibited the highest rates of both all-grade and grade ≥ 3 TRAEs. Regarding ILD, T-DXd showed the highest all-grade incidence, while Teliso-V was associated with the highest rate of grade ≥ 3 ILD. Notably, ADCs with higher DARs were associated with elevated TRAE rates, including ILD. This suggests that while increasing the drug payload per antibody may enhance therapeutic efficacy, it also elevates the risk of off-target effects. In particular, when topoisomerase I inhibitors were combined with high DAR values and cleavable linkers, the incidence of ILD increased, aligning with findings from prior studies [53]. The underlying mechanism may involve premature release of the highly membrane-permeable payload in the highly perfused lung tissue due to linker cleavage. The freely diffusing payload can then enter lung tissue and damage normal pulmonary cells—such as type II alveolar epithelial cells—that are undergoing DNA replication and are sensitive to topoisomerase I inhibition. This triggers widespread cellular injury, inflammation, and dysregulated repair processes, ultimately leading to ILD/pneumonitis. Conversely, ADCs with non-cleavable linkers and lower DAR values, such as T-DM1, exhibited lower TRAE rates. Non-cleavable linkers prevent premature payload release before reaching target tumor cells, reducing off-target organ damage. However, the lower DAR values also diminish the drug payload per antibody, potentially compromising therapeutic efficacy. Additionally, payloads with short half-lives represent an ideal strategy to mitigate potential off-target toxicity. Pharmacokinetic studies have shown that T-DM1 exhibits a short half-life—likely contributing to its lower incidence of AEs [57]. In comparison, Sac-TMT’s longer half-life may correlate with its higher TRAE rates. In summary, careful attention to the risk of ADC-related AEs and enhanced clinical monitoring are essential in real-world practice.

Current therapeutic approaches for NSCLC remain insufficient. Although novel agents like Proteolysis Targeting Chassimile Compounds (PROTACs) and molecular glues that modulate Cyclin-Dependent Kinase (CDK) function are emerging, ADCs constitute a fundamentally different approach, leveraging targeted delivery to concentrate potent cytotoxic payloads at the tumor site. Relative to these protein degradation strategies, ADCs often demonstrate faster and more robust efficacy and provide greater flexibility in target selection. Moreover, their targeted mechanism confers a substantially wider therapeutic window compared to traditional small-molecule drugs [58]. However, several unresolved challenges remain in the clinical translation of ADCs. Even with the most advanced current products, only about 2% of administered ADCs successfully reach the tumor surface [59]. Multiple biological processes can influence ADC activity, including target antigen turnover, internalization efficiency, lysosomal processing, and degradation rates. Additionally, addressing ADC resistance is crucial. When resistance mechanisms related to target antigens or payloads are identified, potential strategies include switching to ADCs targeting different antigens or employing alternative payloads, utilizing dual-payload ADCs or bispecific antibody ADCs [60], and exploring combination therapies—such as ADCs with EGFR-TKIs or ICIs [12, 41]. Furthermore, ADCs may play varying roles across different NSCLC subtypes, serving as either primary or auxiliary therapeutic agents depending on the context.

This study has several limitations. First, as the majority of included studies were single-arm trials lacking randomized controls, we were unable to calculate comparative effect sizes or fully control for confounding biases. Therefore, the interpretation of efficacy outcomes should be approached with caution, and establishing definitive causal relationships remains challenging. Second, substantial statistical heterogeneity was observed across several outcome measures. Although subgroup analyses and meta-regression were performed to investigate potential sources of this heterogeneity, they may not fully account for all the observed variations. Third, the majority of included studies, having enrolled unselected patient populations, lack the specificity required to address the practical challenges in studying specific, rare, and complex molecular subtypes of cancer. Finally, a key limitation is the heterogeneity in treatment lines: most enrolled patients received late-line therapy (≥ 3L), whereas historical comparators (e.g., docetaxel) often came from second-line settings. This difference not only complicates cross-trial survival comparisons and likely contributes to outcome variability, but also precluded a reliable subgroup analysis due to insufficient outcome data stratified by treatment line. Future research should address these limitations through multicenter collaborations to expand sample sizes and enhance generalizability. Randomized controlled trials will be essential to overcome the current shortcomings.

Conclusion

ADC monotherapy has demonstrated considerable efficacy in previously treated NSCLC. Patients with non-squamous histology, EGFR mutations, or HER2 mutations may derive greater benefit from ADC therapy. However, it should be noted that the partially pooled results, derived from highly heterogeneous data, require cautious interpretation. Future studies with more homogeneous populations are needed to confirm these findings. In addition, close monitoring and proactive management of hematologic and respiratory system-related toxicities are essential. It is noteworthy that the incidence of TRAEs varies among different ADCs, particularly with respect to ILD.

Abbreviations

ADC

Antibody-drug conjugate

AGAs

Actionable genomic alterations

B7-H3

B7 homolog 3

CDK

Cyclin-Dependent Kinase

CEACAM5

Carcinoembryonic antigen-related cell adhesion molecule 5

CIs

Confidence intervals

c-MET

Mesenchymal-epithelial transition factor

CTCAE

Common Terminology Criteria for Adverse Events

DAR

Drug-to-antibody ratio

Dato-DXd

Datopotamab deruxtecan

DCR

The disease control rate

DOR

Duration of response

ECOG PS

Eastern Cooperative Oncology Group Performance Status

EGFR

Epidermal Growth Factor Receptor

EMA

European Medicines Agency

FDA

Food and Drug Administration

GGFG

Glycine-Glycine-Phenylalanine-Glycine

HER2

Growth factor receptor 2

HER3

Growth factor receptor 3

ICR

Independent Committee Review

ILD

Interstitial lung disease

KPS

Karnofsky Performance Status

mAb

Monoclonal antibody

MCC

Maleimidomethyl Cyclohexane1-Carboxylate

MedDRA

Medical Dictionary for Regulatory Activities

MeSH

Medical Subject Headings

MMAE

Monomethyl auristatin E

NMPA

National Medical Products Administration

NR

Not Reported

NSCLC

Non-small cell lung cancer

ORR

Objective Response Rate

OS

Median overall survival

Patritu-DXd

Patritumab deruxtecan

PBC

Platinum-based chemotherapy

PFS

Progression-free survival

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

PROTACs

Proteolysis Targeting Chassimile Compounds

RCT

Randomized controlled trial

ROB2

Cochrane Risk of Bias 2

ROBINS-I

Risk Of Bias In Non-randomized Studies—of Interventions

Sac-TMT

Sacituzumab tibotansutecan

SG

Sacituzumab govitecan

SHR-A1811

Trastuzumab rezetecan

T-DXd

Trastuzumab deruxtecan

Teliso-V

Telisotuzumab vedotin

TKIs

Tyrosine Kinase Inhibitors

TRAE

Treatment-related adverse event

TROP2

Trophoblast cell surface antigen 2

VC

Valine Citrulline

Authors’ contributions

Linling Zhang: Conceptualization, Investigation, Methodology，Writing – original draft. Hongyuan Jia: Methodology, Validation, Software, Project administration, Writing – review & editing. Bin Niu: Resources, Supervision, Validation, Writing – original draft, Writing – review & editing.

Funding

The study is supported by the Wu Jieping Medical Foundation of China(Grant No.320.6750.2021–22-27).

Data availability

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable. This study does not contain any individual person’s data in any form.

Competing interests

The authors declare no competing interests.