Abstract
Introduction
Trastuzumab deruxtecan (T-DXd)-induced interstitial lung disease/pneumonitis (ILD) represents a clinically significant and potentially fatal toxicity. Discrepancies exist regarding its reported frequency and severity between clinical trials (CTs) and real-world data (RWD). This meta-analysis aims to evaluate the incidence of T-DXd-related ILD and investigate its differences between CTs and RWD.
Methods
A systematic review and meta-analysis was conducted in accordance with the PRISMA guidelines. Databases were searched from their inception through January 2026. CTs and real-world studies reporting T-DXd-related ILD were included in the analysis. Pooled incidences for all-grade, grade ≥3, and fatal ILD were calculated using random-effects models. Subgroup analyses comparing CTs and RWD, and meta-regression analyses for relevant outcomes were performed.
Results
Thirty-five studies (19 CTs, 16 RWD) including 6840 patients were analyzed. The pooled incidence was 8.8% for all-grade ILD, 1.6% for grade ≥3 ILD, and 0.26% for fatal ILD. RWD was independently associated with lower reported rates of all-grade and fatal ILD, while prior lines of therapy were the main predictor of grade ≥3 ILD.
Conclusion
ILD risk with T-DXd differs by severity and data source. Vigilant monitoring is essential, particularly in heavily pretreated patients.
Introduction
Trastuzumab deruxtecan (T-DXd) is an antibody-drug conjugate (ADC) that has demonstrated significant efficacy in human epidermal growth factor receptor 2 (HER2)-expressing solid tumors, establishing itself as a pivotal therapeutic option in contemporary oncology.1-5 Despite its potent anti-tumor activity, the risk of interstitial lung disease (ILD)/pneumonitis-hypothesized to be triggered by mechanisms such as alveolar macrophage involvement and the bystander effect-remains one of the most critical toxicities limiting its clinical utility.6,7 ILD encompasses a heterogeneous group of disorders affecting the lung parenchyma, characterized by a potentially irreversible and progressive clinical course. ADC-induced ILD most commonly presents with symptoms of dry cough, progressive dyspnea, and fatigue. Although it is detectable through radiological findings in its early stages, failure in prompt recognition or inadequate management may lead to the development of pulmonary fibrosis and, ultimately, fatal outcomes.8,9
In clinical trials (CTs), the incidence of ILD is typically reported with specific grades of severity, utilizing standardized adverse event reporting criteria and intensive monitoring protocols.10,11 However, uncertainty persists regarding the extent to which data derived from these controlled environments reflect real-world data (RWD), which encompasses more fragile, heterogeneous patient populations with various comorbidities and extensive prior treatments. It should also be acknowledged that methodological discrepancies in patient selection, monitoring frequency and imaging strategies complicate the direct comparison of ILD incidence and severity between CTs and RWD. 12 Specifically, elucidating how the determinant factors for mild versus severe ILD differ between clinical trial setting and real-world practice is a critical requirement for optimizing ADC-related toxicity management.8,12
In this context, comprehensive meta-analyses evaluating T-DXd-related ILD between CTs and RWD as well as investigating potential risk predictors systematically remain limited in the literature. Therefore, this study aims to evaluate the incidence patterns and potential predictors of T-DXd-related ILD through a holistic approach by integrating both CTs and RWD.
Methods
Literature Search and Study Selection
This systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines. 13 A comprehensive literature search was performed across PubMed/MEDLINE, Embase, and Cochrane Library databases from their inception until January 1, 2026. Searches were updated through January 10, 2026. Additionally, supplementary searches were conducted via Google Scholar, and reference lists of relevant studies were manually reviewed to identify additional publications. The search strategies utilized a combination of controlled vocabulary (MeSH/Emtree) and free-text keywords including “trastuzumab deruxtecan”, “T-DXd”, “interstitial lung disease”, “pneumonitis” and “pulmonary toxicity”. The complete list of MeSH terms and the detailed search string are provided in Supplementary Material 1. Only studies published in the English language were included in the analysis.
Discrepancies were noted in the terminology used for adverse events across the included studies (with some using only “ILD”, others only “pneumonitis”, and some using both). To ensure data integrity and prevent ambiguity, all findings were analyzed under a single clinical umbrella term (ILD/pneumonitis). In cases where both terms were reported separately for the same patient group, each case was counted only once to avoid double counting.
Prospective/retrospective CTs and observational real-world studies involving adult patients treated with T-DXd monotherapy (5.4 or 6.4 mg/kg) in metastatic setting were included. Selected studies were required to clearly report the incidence of ILD and the corresponding number of patients. To maintain statistical power, only studies with a sample size greater than 30 (n>30) were included. Phase 1 trials, case reports, reviews, editorials and conference abstracts without published full texts were excluded to ensure data homogeneity. For studies involving overlapping patient populations, the most recent or comprehensive dataset was prioritized and others were excluded. The study selection process is illustrated in the PRISMA flow diagram (Figure 1). PRISMA flow diagram illustrating the study identification, screening, eligibility, and inclusion process for the meta-analysis
Data Extraction and Quality Assessment
Study selection and data extraction were performed by two independent researchers (A.K.G. and N.C.D.). Title/abstract screening and full-text evaluations were conducted independently; any discrepancies were resolved through mutual consensus or consultation with a third senior researcher (O.K.).
Data extracted from each study included publication year, study design (CT or RWD), tumor type, sample size, median age, T-DXd dose (mg/kg), median number of prior treatment lines, follow-up duration and ILD outcomes. ILD data were categorized as all-grade, grade ≥3 and fatal events. Median values reported for the time to ILD onset were recorded as descriptive values; missing data were marked as not calculable. In studies where multiple T-DXd dose arms were reported, each dose arm was treated as an independent cohort.
The methodological quality of the included studies was evaluated using assessment tools appropriate for each study design. The Cochrane Risk of Bias Tool 2.0 was employed for randomized controlled trials, while the Newcastle–Ottawa Scale was utilized for observational real-world studies.
Outcomes Measures and Statistical Analysis
Primary endpoints were the pooled incidence rates for all-grade and grade ≥3 ILD associated with T-DXd Due to reporting heterogeneity, the time to ILD onset was not statistically pooled and reported median values were summarized descriptively instead. Statistical analyses were performed using a random-effects model to account for clinical and methodological heterogeneity among the studies. Proportions were analyzed using the Freeman–Tukey double arcsine transformation to ensure variance stabilization, particularly in rare events, and results were presented with 95% confidence intervals. 14 Heterogeneity was assessed using the I2 statistic and Cochran’s Q test.
Subgroup analyses using random-effects models evaluated ILD incidence across study design, cancer type, dose, median age, and prior treatment lines. Differences between groups were assessed using the Q-test, with the group variable included in the model as a moderator. Additionally, univariate and multivariate meta-regression analyses were performed to evaluate the impact of study type (CT vs. RWD), dose (5.4 vs. 6.4mg/kg), median age, cancer type (breast cancer vs. others) and number of prior treatment lines on all-grade and grade ≥3 ILD. Covariates for meta-regression were selected based on data availability across studies and their role as independent variables, with preference given to clinically meaningful parameters in cases of overlap. Formal assessment of publication bias was not conducted due to the limited number of studies and events. All analyses were performed using R software via the meta and metafor packages, and a two-sided p-value of <0.05 was considered the threshold for statistical significance.
Publication bias was evaluated for the primary outcome (all-grade ILD) using visual inspection of funnel plot asymmetry and Egger’s regression test. Given the predominance of single-arm studies and the low number of severe and fatal ILD events, formal assessments of publication bias were not performed for these outcomes, as such analyses may be unreliable in the presence of sparse data.
Results
Baseline Characteristics of Clinical Trials and Real-World Studies Included in the Meta-Analysis of Trastuzumab Deruxtecan–Associated Interstitial Lung Disease
CRC, colorectal cancer; GEJ, gastroesophageal junction; ILD, interstitial lung disease; NR, not reported; NSCLC, non-small cell lung cancer; RWD, real world data, T-DXd, trastuzumab deruxtecan.
*Gender data were reported for only 73 patients.
**Gender data were reported for only 30 patients.

Forest plot showing the pooled incidence of all-grade interstitial lung disease associated with trastuzumab deruxtecan

Forest plot of the pooled incidence of grade ≥3 interstitial lung disease associated with trastuzumab deruxtecan

Forest plot of the pooled incidence of fatal interstitial lung disease associated with trastuzumab deruxtecan
Data regarding the time to onset of ILD were reported in a limited number of studies, encompassing 3853 patients. The median of the reported median of medians for ILD onset was 127 days, with individual study medians ranging from 44 to 307 days. When considering the minimum and maximum values reported across these studies, the onset of ILD exhibited a broad spectrum, ranging from as early as 9 days to as late as 835 days. To address potential bias related to non-standardized imaging schedules in RWD, a sensitivity analysis restricted to CTs was performed, yielding a comparable median time to ILD onset of 124 days. In contrast, median onset times reported in real-world studies ranged from 87 to 307 days. Furthermore, among studies reporting ILD-related treatment discontinuation, 270 of 4174 patients (6.5%) discontinued therapy due to this adverse event.
Subgroup Analysis of All-Grade, Grade ≥3 and Fatal ILD Incidence in Clinical Trials and Real-World Settings
CI: Confidence Interval, CTs: clinical trials, ILD: Interstitial Lung Disease, RWD: real-world data.
In univariate meta-regression, real-world data (RWD) was significantly and inversely associated with the incidence of all-grade ILD (β = −0.102, p = 0.001). Cancer type was also significantly associated with ILD incidence, with lower rates observed in breast cancer compared to other tumor types (β = −0.072, p = 0.03), whereas no significant associations were found for median age, dose, or the number of prior treatment lines. In the multivariate analysis, RWD remained independently associated with a lower incidence of all-grade ILD (β = −0.105, p = 0.001), while cancer type did not retain statistical significance.
Regarding grade ≥3 ILD, univariate analysis demonstrated a significant association only with the median number of prior treatment lines (β = 0.017, p = 0.01), while RWD, age, dose, and cancer type were not associated with ILD incidence. In the multivariate model, the number of prior treatment lines remained independently associated with grade ≥3 ILD (β = 0.015, p = 0.04).
Univariable and Multivariable Meta-Regression Analyses Evaluating Study-Level Factors Associated With All-Grade, Grade ≥3 and Fatal Interstitial Lung Disease Incidence Following Trastuzumab Deruxtecan Treatment
CI: Confidence Interval, CT: clinical trials, PTL: Prior treatment lines, RWD: real-world data.
Visual inspection of the funnel plot did not reveal marked asymmetry for the primary outcome. Consistently, Egger’s regression test did not indicate significant small-study effects. The funnel plot is provided in Supplementary Figure S1.
Discussion
This meta-analysis reveals that T-DXd-associated ILD exhibits distinct patterns based on severity and highlights several discrepancies between CTs and RWD. The incidence of all-grade ILD was significantly higher in CTs compared to RWD, a finding that remained consistent across subgroup, univariate, and multivariate meta-regression analyses. Conversely, no significant difference was observed between CTs and RWD regarding grade ≥3 ILD; instead, the median number of prior treatment lines emerged as the primary determinant for the development of grade ≥3 ILD. Fatal ILD was reported at a higher rate in CTs than in RWD, maintaining its independence in multivariate analyses. These findings suggest that diagnostic and reporting discrepancies may play a role in milder forms of ILD, whereas biological factors related to patient and treatment burden appear to be more dominant in severe cases.
Previous studies in literature have highlighted significant discrepancies between CTs and RWD, particularly regarding safety outcomes.45-47 These variations can largely be attributed to differences in patient selection criteria, monitoring strategies, and the methodology used to define adverse events. 45 While CTs identify adverse events through prospective, intensive, and standardized monitoring, real-world practice may lead to underreporting of low-grade events or alternative classification of fatal outcomes.45,48 Despite the very low number of fatal events, consistent with various reports in oncology literature, the higher rate of fatal ILD observed in CTs compared to RWD in our study underscores the impact of these rigorous surveillance and reporting mechanisms.
In analyses evaluating clinical factors associated with the risk of T-DXd–related ILD in patients with metastatic breast cancer, a higher number of prior treatment lines, previous exposure to immune checkpoint inhibitors and the presence of baseline interstitial lung abnormalities have been identified as potential risk factors for ILD development. 40 These findings support the hypothesis that prolonged exposure to cytotoxic agents may induce cumulative vulnerability within the pulmonary parenchyma, thereby facilitating T-DXd–associated lung injury.49,50 In the present meta-analysis, the inclusion of multiple tumor types resulted in substantial heterogeneity with respect to prior treatment regimens. Nevertheless, our results-particularly those related to grade ≥3 ILD-are noteworthy in that they underscore an association between higher prior treatment burden and development of severe ILD. As the use of T-DXd continues to expand in routine clinical practice, future studies incorporating more granular subgroup analyses are warranted to better delineate the impact of prior treatment strategies, concomitant comorbidities, baseline lung disease, and both previous and concurrent medication use on ILD risk.
In T-DXd dose evaluation studies, it has been shown that the risk of ILD increases at the higher dose of 6.4 mg/kg.51,52 Although a pooled analysis by Powell et al, covering phase 1 and 2 clinical trials, reported that high therapeutic doses (>6.4mg/kg) increased the risk of ILD, differences within the approved dose range (5.4–6.4mg/kg) have not been consistently demonstrated in the literature. 53 In the DestinyLung-02 study, ILD was observed at a rate of 12.9% with a dose of 5.4 mg/kg, compared to 28% with a 6.4mg/kg dose. 23 Similarly, in the DestinyCRC-02 study, ILD occurred at a rate of 8.4% with the 5.4mg/kg dose and 12.8% with the 6.4 mg/kg dose. 25 In a meta-analysis by Soares et al, patients receiving a T-DXd dose of 6.4 mg/kg developed a significantly higher rate of ILD (22.7%) compared to those receiving a dose of 5.4 mg/kg (9.3%). 54 However, no significant difference was found between the 5.4mg/kg and 6.4mg/kg doses in our analyses despite the inclusion of the DestinyLung-02 and DestinyCRC-02 trials. Although clinical trials generally demonstrate a dose-dependent relationship between T-DXd and the risk of ILD, this association is particularly observed when the dose exceeds 6.4 mg/kg.23,25 In our analysis, no significant difference was identified between the recommended therapeutic doses of 5.4 and 6.4mg/kg in terms of ILD development.
Following the successful results of T-DXd, particularly in breast cancer, its use in early-stage patients was evaluated in the DESTINY-Breast11 trial.1,2,16-18,55 Considering the duration of treatment in early stage disease as well as the limited four-cycle administration of T-DXd in the DESTINY-Breast11 trial, and its potential for curative use, ILD-and especially fatal pulmonary toxicities-have become even more critical in this patient group. In the DESTINY-Breast11 trial, the incidence of all-grade ILD was reported as 4.9% in the T-DXd monotherapy arm and 4.4% in the T-DXd–THP arm. 55 In contrast, our meta-analysis, which includes a metastatic patient population where treatment is continued until progression, found the incidence of all-grade ILD to be 8.8%, a rate that reaches 11.4% in clinical trials. In our study, the reported median time to onset for ILD was 127 days; when evaluated together, these findings suggest that the use of T-DXd in early stage with a limited duration and cycle number may be associated with a lower risk of ILD development compared to long-term exposure in metastatic setting. Nonetheless, further studies are needed to more clearly establish the relationship between the cumulative dose of T-DXd and the risk of ILD development.
This meta-analysis has several limitations. Since the analyses are based on study-level reported data, a detailed assessment of patient-level risk factors was not possible. In the meta-regression analyses, only a limited number of clinical parameters could be evaluated; specifically, the type and duration of prior treatments, concomitant comorbidities, and baseline lung diseases could not be examined in detail. The fact that key parameters-such as treatment dose, time to ILD onset and ILD-related treatment discontinuation rates-are frequently underreported in real-world studies may have restricted the scope of the analyses. Furthermore, the inclusion of different tumor types led to clinical heterogeneity, and the limited number of events for rare outcomes may have complicated the interpretation of the results.
Despite these limitations, our study has several significant strengths. To our knowledge, this is one of the most comprehensive meta-analyses to date, integrating both clinical trial data and real-world evidence to provide a holistic overview of ILD risk associated with T-DXd. The inclusion of diverse tumor types allowed for a broader assessment of the safety profile across different patient populations. Furthermore, the use of rigorous statistical methods, including meta-regression and subgroup analyses, enabled a more nuanced exploration of potential risk factors. These findings provide valuable insights for clinicians in optimizing monitoring strategies and managing treatment-related toxicities in daily practice.
Conclusion
Our findings demonstrate that the incidence of T-DXd-related ILD exhibits distinct patterns based on its severity, with significant differences between CTs and RWD in terms of all-grade and fatal ILD. Although CTs include highly selected patient populations, the higher rate of all-grade and fatal ILD compared to RWD might be associated with the use of more standardized and intensive monitoring protocols in CTs. Additionally, particular attention should be paid to heavily pretreated patients for the development of severe T-DXd-related ILD. These results altogether emphasize the importance of vigilant monitoring and risk stratification for ILD occurrence in patients receiving T-DXd. Future patient-level and prospective studies are expected to identify clinical and biological markers that can more precisely predict T-DXd-related ILD risk.
Supplemental Material
Supplemental Material - Trastuzumab Deruxtecan–Induced Interstitial Lung Disease: A Systematic Review and Meta-Analysis of Clinical Trials and Real-World Evidence
Supplemental Material for Trastuzumab Deruxtecan–Induced Interstitial Lung Disease: A Systematic Review and Meta-Analysis of Clinical Trials and Real-World Evidence by Ali Kaan Güren, Nazım Can Demircan, Murat Sarı, Osman Köstek, İbrahim Vedat Bayoğlu in Cancer Control.
Supplemental Material
Supplemental Material - Trastuzumab Deruxtecan–Induced Interstitial Lung Disease: A Systematic Review and Meta-Analysis of Clinical Trials and Real-World Evidence
Supplemental Material for Trastuzumab Deruxtecan–Induced Interstitial Lung Disease: A Systematic Review and Meta-Analysis of Clinical Trials and Real-World Evidence by Ali Kaan Güren, Nazım Can Demircan, Murat Sarı, Osman Köstek, İbrahim Vedat Bayoğlu in Cancer Control.
Supplemental Material
Supplemental Material - Trastuzumab Deruxtecan–Induced Interstitial Lung Disease: A Systematic Review and Meta-Analysis of Clinical Trials and Real-World Evidence
Supplemental Material for Trastuzumab Deruxtecan–Induced Interstitial Lung Disease: A Systematic Review and Meta-Analysis of Clinical Trials and Real-World Evidence by Ali Kaan Güren, Nazım Can Demircan, Murat Sarı, Osman Köstek, İbrahim Vedat Bayoğlu in Cancer Control.
Supplemental Material
Supplemental Material - Trastuzumab Deruxtecan–Induced Interstitial Lung Disease: A Systematic Review and Meta-Analysis of Clinical Trials and Real-World Evidence
Supplemental Material for Trastuzumab Deruxtecan–Induced Interstitial Lung Disease: A Systematic Review and Meta-Analysis of Clinical Trials and Real-World Evidence by Ali Kaan Güren, Nazım Can Demircan, Murat Sarı, Osman Köstek, İbrahim Vedat Bayoğlu in Cancer Control.
Supplemental Material
Supplemental Material - Trastuzumab Deruxtecan–Induced Interstitial Lung Disease: A Systematic Review and Meta-Analysis of Clinical Trials and Real-World Evidence
Supplemental Material for Trastuzumab Deruxtecan–Induced Interstitial Lung Disease: A Systematic Review and Meta-Analysis of Clinical Trials and Real-World Evidence by Ali Kaan Güren, Nazım Can Demircan, Murat Sarı, Osman Köstek, İbrahim Vedat Bayoğlu in Cancer Control.
Footnotes
Ethical Considerations
Ethics approval was not required for this study, as it is a systematic review and meta-analysis based solely on data from previously published studies and does not involve any new studies with human participants or animals.
Author Contributions
All authors contributed to the study conception and design. Material preparation, data collection, and analysis were performed by the authors. All authors contributed to manuscript drafting and approved the final version.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
All data analyzed during this study are included in the published articles cited in the manuscript.
Use of Artificial Intelligence
No artificial intelligence (AI) tools were used for any aspect of this study.
Supplemental Material
Supplemental material for this article is available online.
Appendix
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
