The safety of trastuzumab deruxtecan (T-DXd) in breast cancer brain metastases with a focus on interstitial lung disease/pneumonitis: A systematic review and meta-analysis.

INTRODUCTION
Human epidermal growth factor receptor 2 (HER2)‐positive breast cancer (BC) constitutes 15%–20% of all BCs. This subtype is characterized by rapid tumor growth, aggressive behavior, and a predisposition to recurrence and metastasis. The advent of antibody–drug conjugates (ADCs) has significantly improved survival outcomes for patients with HER2‐positive, advanced BC (ABC). In 2013, the first HER2‐targeted ADC, trastuzumab emtansine (T‐DM1), received approval for second‐line treatment of HER2‐positive ABC based on the phase 3 EMILIA trial (ClinicalTrials.gov identifier NCT00829166).
1
Subsequently, the next‐generation ADC trastuzumab deruxtecan (T‐DXd) was approved for unresectable or metastatic, HER2‐positive BC in adults previously treated with one or more anti‐HER2 regimens based on the DESTINY‐Breast03 study (ClinicalTrials.gov identifier NCT03529110).
2

Brain metastases (BM) represent a major contributor to poor prognosis in patients with ABC.
3
The 24‐month cumulative incidence of BM in HER2‐positive patients reached 49.0%, significantly exceeding the rates observed in hormone receptor‐positive BC (43.6%) and triple‐negative BC (44.8%).
4
Conventional chemotherapy and numerous macromolecular‐targeted agents like trastuzumab exhibit limited efficacy against intracranial lesions because of impaired blood–brain barrier (BBB) penetration.
5
Retrospective exploratory analysis of EMILIA indicated no significant benefit for T‐DM1 over chemotherapy in patients with BM (median progression‐free survival, 5.9 months for T‐DM1 vs. chemotherapy 5.7 months for chemotherapy).
6
In contrast, the cleavable linker design of T‐DXd enabled payload delivery across the BBB, conferring intracranial antitumor activity.
7
The BM subgroup analysis of the DESTINY‐Breast03 trial demonstrated superior intracranial outcomes with T‐DXd: an intracranial objective response rate of 65.7% versus 34.3% with T‐DM1 and a median intracranial progression‐free survival of 15.0 months versus 3.0 months, respectively (hazard ratio, 0.25; 95% CI, 0.13–0.45).
8
These data established the definitive clinical advantage of T‐DXd over prior therapies for controlling BM.
However, treatment‐emergent adverse events (TEAEs) associated with T‐DXd, particularly interstitial lung disease (ILD)/pneumonitis, constituted a substantial clinical challenge, even resulting in fatal outcomes in some patients.
9
The phase 3 DESTINY‐Breast02 study (ClinicalTrials.gov identifier NCT03523585) reported ILD in 42 patients (10.3%), including two fatal (grade 5) events.
10
Notably, patients who had BCBM exhibited a higher incidence of ILD after T‐DXd treatment compared with those who did not have BM. The DESTINY‐Breast12 phase 3b/4 study (ClinicalTrials.gov identifier NCT04739761) reported an ILD/pneumonitis incidence of 16.0% (42 patients) in the BM cohort versus 12.9% (31 patients) in the non‐BM cohort. Grade 1 events predominated in both groups (BM group, 9.9%; non‐BM group, 9.1%). Notably, the BM cohort had more fatal (grade 5) events (six patients; 2.3%) than the non‐BM cohort (three patients; 1.2%).
11

However, the previous research predominantly evaluated the safety profile of T‐DXd in the general HER2‐positive ABC population. There was a notable lack of dedicated analyses specifically examining the incidence and severity of ILD/pneumonitis in patients with BCBM. Consequently, the objective of the current systematic review and meta‐analysis was to comprehensively assess TEAEs associated with T‐DXd in patients with BCBM, specifically determining the incidence, severity, and clinical implications of ILD and/or pneumonitis.

MATERIALS AND METHODS

Search strategy
This study was performed following the guidelines from the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta‐Analyses),
12
and it was registered in PROSPERO (the International Prospective Register of Systematic Reviews) under protocol number CRD420251130832. The PRISMA checklist for the abstract and the article are provided in the Supporting Materials. We systematically searched the PubMed, Embase, Web of Science, and Cochrane databases as well as American Society of Clinical Oncology, European Society for Medical Oncology, American Association for Cancer Research, and San Antonio Breast Cancer Symposium conference proceedings in July 2025. The following combination of medical subject headings terms and Boolean connectors were used: “breast cancer” AND “HER‐positive” OR “human epidermal growth factor receptor”; AND “T‐DXd” OR “trastuzumab deruxtecan.” A full description of the search strategy used on each database is provided in Table S1. We also searched the references of all included studies and relevant reviews about this topic.

Inclusion and exclusion criteria
The inclusion criteria were: (1) studies evaluating T‐DXd in patients with HER2‐positive and HER2‐low BCBS and (2) safety outcomes including ILD and/or pneumonia. HER2‐positive status was defined by an immunohistochemistry score of 3+ or 2+ with positive in situ hybridization results. HER2‐low status was defined by an immunohistochemistry score of 1+ or 2+ with negative in situ hybridization results.
The exclusion criteria were: (1) studies with overlapping populations, (2) not original studies (reviews, letters to the editor, and commentaries), (3) studies without the population of interest, (4) studies assessing exclusively patients with leptomeningeal metastases (no measurable brain disease available), and (5) case reports and case series.

Outcome indicators
Outcome indicators were the occurrences of various TEAEs, which were graded using the National Cancer Institute's Common Terminology Criteria for Adverse Events and coded according to the Medical Dictionary for Regulatory Activities. We performed a prespecified analysis of TEAEs according to the Common Terminology Criteria for Adverse Events (all grades, grades 1–3).

Data extraction
Two authors (Y.M. and X.L.) independently screened the studies by title and abstract, selected the articles for full‐text review, and extracted data from the included studies. All inconsistencies between the authors were resolved by consensus or by consulting a third author (Y.J.). We collected data from individual studies on the study design, study location, number of patients, and patients' baseline characteristics (e.g., BM status, median age, hormone receptor status, and previous lines of therapy). We extracted data for pooled analysis pertaining to TEAEs.

Quality evaluation
One multicohort study (nonrandomized) and eight single‐arm studies were assessed using the MINORS (methodological index for nonrandomized studies) index score.
13
A randomized effect model was applied to minimize the bias.

Statistical analysis
R software, version 4.2.2 (R Foundation for Statistical Computing) was used to perform all statistical analyses. We used the Cochrane Q χ2 test and the I2 statistic to examine heterogeneity across studies; p values < .10 and I2 values >50% were considered significant for heterogeneity. Leave‐one‐out sensitivity analysis was performed by removing each study one at a time and recalculating the study results. Funnel plot analysis of point estimates based on study weights and a regression test for funnel plot asymmetry (Egger test) were used to explore publication bias.

RESULTS

Baseline characteristics
In total, our initial search identified 3031 studies. After the removal of duplicates and exclusion by title and abstract, 135 studies were fully assessed. The majority of studies were excluded primarily because they did not include patients with BM, lacked reporting of TEAEs related to ILD/pneumonia, or were published study protocols only (Figure 1). Finally, this systematic review and meta‐analysis included nine studies involving 684 patients with BCBM. The characteristics of the included studies are described in Table 1

11
,
14
,
15
,
16
,
17
,
18
,
19
,
20
,
21

Most of the studies included patients with a median age of 51–55 years and hormone receptor‐positive tumors (63.0%). The mean follow‐up time across the studies was 16 months, ranging from 5.3 to 27.3 months. Table 2

11
,
14
,
15
,
16
,
17
,
18
,
19
,
20
,
21
details the baseline characteristics of patients with BCBM who received T‐DXd.

Quality assessment
Quality assessment demonstrated an acceptable risk of bias in all included trials (Table 3).
11
,
14
,
15
,
16
,
17
,
18
,
19
,
20
,
21
All of the studies had open‐label designs, and performance bias might have existed but did not affect the final outcome. Nine articles were assessed using the MINORS (methodological index for nonrandomized studies) index (16‐point scale). Three articles demonstrated high methodological quality, and six showed moderate quality. The assessment suggested that all included articles were appropriate for the current study.

Incidence of any TEAEs
The occurrence of any TEAEs was 97% (95% CI, 94%–100%; I2 = 57.9%), and the incidence of any TEAEs (grade ≥3) was 48% (95% CI, 42%–54%; I2 = 18.5%). The incidence of any TEAEs associated with dose reduction was 14% (95% CI, 11%–18%; I2 = 32.5%), and the incidence of those associated with treatment discontinuation was 28% (95% CI, 0%–57%; I2 = 88.8%). The rate of any TEAEs associated with death was 1% (95% CI, 0%–4%; I2 = 70.8%; Figure 2). In general, the pooled analysis results indicated that the safety outcomes of T‐DXd in patients with BCBM were acceptable. The types and incidences of all TEAEs included are listed in Table 4.

Incidence of ILD/pneumonia
Of the 684 patients included in the safety analysis of the studies, 79 cases of ILD/pneumonia were recorded. A pooled analysis revealed an overall ILD/pneumonia incidence rate of 10% (95% CI, 5%–16%; I2 = 75.8%). The incidence of ILD/pneumonia (grade ≥3) was 2% (95% CI, 1%–4%; I2 = 0%), and the incidence of ILD/pneumonia associated with discontinuation was 11% (95% CI, 0%–24%; I2 = 94.3%) (Figure 3).

Subgroup analysis of ILD and/or pneumonitis
Of the nine studies included in our analysis, eight reported the incidence rate of any‐grade ILD and/or pneumonitis as well as the rate of grade ≥3 ILD and/or pneumonitis. Therefore, we performed a subgroup analysis on the incidence of any‐grade ILD and/or pneumonitis as well as grade ≥3 ILD and/or pneumonitis (Figure 4). However, only three studies provided detailed reports on the discontinuation rates because of ILD and/or pneumonitis, making it impossible to conduct a subgroup analysis for this outcome. We observed a higher incidence of ILD/pneumonia (all grades) in studies published in 2024, particularly those conducted in Asia and retrospective studies. Moreover, a higher incidence of ILD/pneumonia (grade ≥3) was noted in studies published in 2024, particularly in Europe and in retrospective studies. Nevertheless, none of the differences between these subgroups were statistically significant.
FIGURE 4Subgroup meta‐analysis for the rates of ILD and/or pneumonitis according to year, region, and cohort: (A) all grades stratified by publication year, (B) grade ≥3 stratified by publication year, (C) all grades stratified by region, (D) grade ≥3 stratified by region, (E) all grades stratified by cohort, and (F) grade ≥3 stratified by cohort. CI indicates confidence interval; I2, between‐study variance used to indicate heterogeneity; ILD, interstitial lung disease.

Publication bias and sensitivity analysis
To assess publication bias and the sensitivity of our findings, we evaluated the effect of each included study on the pooled estimates for the incidence of ILD/pneumonia (all grade and grade ≥3). Publication bias was assessed using funnel plots and the Egger test, which incorporates heterogeneity estimates across studies. Because the number of studies included in the analysis was limited to nine, and one of those studies only reported the discontinuation rate, the exploratory Egger test and sensitivity analyses for ILD/pneumonia (all grades) and ILD/pneumonia (grade ≥3) were conducted using data from the remaining eight studies. An analysis of the all‐grade ILD/pneumonia incidence data revealed a symmetrical distribution in the funnel plot, with two studies falling outside the 95% CI (see Figure S1). The Egger test produced a significant p value (p = .003), indicating potential publication bias (Figure S1, Table S2). Sensitivity analyses through the sequential exclusion of individual studies demonstrated no substantial changes in the incidence rates or 95% CIs, supporting the reliability of the primary analysis results (Figure S1). For ILD/pneumonia incidence (grade ≥3), the funnel plot exhibited a symmetrical distribution with no studies outside the 95% CI, suggesting no significant publication bias (Figure S1). However, the Egger test produced a p value of .011, suggesting possible publication bias (Figure S1, Table S3). Sensitivity analyses demonstrated no significant alterations in incidence rates or 95% CIs, similarly confirming the robustness of the initial findings (Figure S1).

DISCUSSION
This systematic review and meta‐analysis provided a comprehensive assessment of TEAEs, with a specific focus on ILD/pneumonia, associated with T‐DXd therapy in patients with BCBM. By synthesizing data from nine studies involving 684 patients, our analysis elucidated critical insights into the safety of T‐DXd in this subpopulation with a poor prognosis.
A meta‐analysis that included 35 studies in patients who had solid tumors treated with T‐DXd indicated that the incidence of ILD and/or pneumonitis (all grades) was 13.2%, with a grade ≥3 incidence of 2.3%.
22
This result is similar to the data observed in BC. In a cohort of patients who had HER2‐positive BC treated with T‐DXd, irrespective of BM status, results from the DESTINY‐Breast04 study (ClinicalTrials.gov identifier NCT03734029) demonstrated an incidence of ILD and/or pneumonitis (all grades) of 12.1%, with a grade ≥3 incidence of 2.1%.
23
A meta‐analysis that included 1970 patients who had HER2‐positive BC treated with T‐DXd reported an all‐grade incidence of ILD and/or pneumonitis of 10.21%.
24
In the DESTINY‐Breast02 and DESTINY‐Breast12 studies, which specifically focused on populations with BM or subgroups, the incidence of ILD and/or pneumonitis was 10.3% and 16.0%, respectively.
10
,
11
Notably, in the DESTINY‐Breast12 study, the incidence of ILD/pneumonitis was significantly elevated compared with previous clinical trials that did not stratify by BM status, suggesting that BM may constitute a significant risk factor for the incidence of ILD/pneumonitis. The incidence of ILD and/or pneumonitis in our study was similar to that observed in these studies and was not significantly higher than the incidence in clinical trials that did not stratify by BM. This may have been because of the smaller sample sizes of the individual studies in our analysis as well as the relatively short follow‐up times in some of the included studies. It is worth noting that the DESTINY‐Breast06 study (ClinicalTrials.gov identifier NCT04494425), which mainly focused on HER2‐low and HER2‐ultralow populations, reported an incidence of ILD and/or pneumonitis (all grades) of 11.3%, similar to the incidence observed in HER2‐positive BC cohorts (12.1% and 10.21%). However, the incidence of ILD and/or pneumonitis (grade ≥3) was 1.4%, which is significantly lower than the 2.1% reported in the aforementioned studies.
25
This suggests that, compared with HER2‐positive patients, those with HER2‐low and/or HER2‐ultralow status may experience a lower incidence of severe adverse events, providing clinical evidence for future studies to explore the effect of HER2 expression status on the incidence of ILD and/or pneumonitis in patients with BM.
Beyond the specific ILD/pneumonitis rates, the effect of ILD/pneumonitis on treatment modification was notable. In the current study, the ILD/pneumonitis–associated discontinuation rate was 11%, demonstrating the critical role of ILD/pneumonitis in affecting treatment modification decisions. This high discontinuation rate presents a substantial clinical dilemma given the unprecedented intracranial efficacy demonstrated by T‐DXd in patients with HER2‐positive BCBM.
26
Therefore, effective management of ILD was critically important for constantly maintaining the intracranial efficacy of T‐DXd. It is noteworthy that no fatal outcomes were observed in this study. The complete absence of fatal ILD/pneumonitis events stands in particularly notable contrast to the 2.3% fatal ILD/pneumonitis rate documented in the BCBM cohort of DESTINY‐Breast12, underscoring the need for nuanced consideration of this seemingly positive outcome.
11

The elevated incidence of ILD/pneumonitis observed in patients with BCBM might be attributed to several molecular mechanisms. BM could induce significant disruption of the BBB, prompting activated microglia and astrocytes to release proinflammatory cytokines into the systemic circulation.
27
,
28
This process contributed to alterations in the immune microenvironment. Studies have demonstrated that the inflammatory environment could enhance the phagocytic activity of alveolar macrophages.
29
The alveolar macrophages resident in perivascular niches directly engulfed blood‐circulating T‐DXd by Fc–Fc gamma receptor (Fc‐FcγR) engagement. This Fc‐FcγR interaction with T‐DXd triggered a phenotypic shift in alveolar macrophages from an immunosuppressive to a pro‐ILD state.
30
Moreover, compared with patients who do not have BM, those who have BM might exhibit poorer outcomes secondary to ILD/pneumonitis, potentially mediated by systemic inflammation from ILD/pneumonitis exacerbating BBB disruption or dysregulation.
31
,
32
Meanwhile, ILD/pneumonitis‐induced dyspnea and hypoxia could elevate intracranial pressure, which, in turn, suppresses respiratory center function, adversely affecting survival.
33
,
34

In the current study, the overall incidence of TEAEs (all grades, 97%; grade ≥3, 48%) observed in patients with BCBM was consistent with the previously reported toxicity profile of T‐DXd in solid tumors (all grades, 97.2%; grade ≥3, 54.9%), highlighting the inherent toxicologic profile of T‐DXd.
35
In this study, fatigue and nausea were the most common all‐grade TEAEs, occurring in 67% and 60% of patients with BCBM, respectively, also consistent with the known effects of T‐DXd.
36
Neutropenia was the most frequent grade ≥3 event (15%), underscoring the necessity for vigilant hematologic monitoring.
37
The low incidence of fatal TEAEs (1%) offered reassurance regarding the safety of T‐DXd treatment.
Several limitations of this analysis warranted consideration. The included studies were predominantly retrospective or subgroup analyses of larger trials, introducing potential selection bias in ILD/pneumonitis monitoring, grading, and management protocols. Moderate‐to‐high heterogeneity (I2 = 68% for all‐grade ILD, 94.3% for ILD/pneumonitis‐associated discontinuation) was observed in some pooled estimates. The limited number of studies included in this analysis suggested that the findings from the publication bias and sensitivity analyses should be interpreted with caution. Furthermore, the relatively small sample sizes in some of the individual studies included may contribute to the observed high heterogeneity. In addition, the relatively short median follow‐up (8.6 months) in some studies might underestimate late‐onset ILD/pneumonitis events.
17
Also, in our meta‐analysis there are potential confounding factors, such as the activity of BM, the proportion of patients with lung metastasis, and the receipt of prior anti‐HER2 treatment strategies. However, because of the lack of standardized reporting of these clinical details in some of the included studies, we were unable to assess these factors through subgroup analysis. Despite these limitations, the current study confirmed that, although the absolute rate of severe or fatal ILD/pneumonitis in patients who had BCBM within the analyzed studies appears potentially lower than that in initial reports in unselected populations, ILD/pneumonitis remained a clinically significant toxicity in patients who had BCBM treated with T‐DXd, occurring in nearly one in 10 patients and leading to discontinuation in a substantial proportion. The absence of reported fatal ILD/pneumonitis events in this cohort was encouraging but requires validation in larger prospective studies with longer follow‐up specifically designed for patients who have BCBM.

Conclusion
T‐DXd represented a major breakthrough in the treatment of patients with HER2‐low and HER2‐positive BCBM, demonstrating unprecedented intracranial efficacy. This meta‐analysis provides essential safety data specific to this population, confirming that ILD/pneumonitis remains a prevalent and clinically significant TEAE, leading to treatment discontinuation in a fraction of patients. Although no fatal ILD/pneumonitis events were reported in this pooled BCBM cohort, the potential association between BM and heightened susceptibility to ILD/pneumonitis highlights the essential need for vigilant monitoring and proactive management to improve survival outcomes in these patients.

AUTHOR CONTRIBUTIONS

Yiwen Ma: Conceptualization; methodology; software; data curation; investigation; validation; formal analysis; funding acquisition; visualization; writing—original draft; writing—review and editing; resources. Xiaorui Li: Writing—review and editing; data curation; investigation; formal analysis; resources. Yujun Jiang: Data curation; investigation; formal analysis; writing—review and editing. Liping Xiao: Data curation; investigation; formal analysis; writing—review and editing. Tao Sun: Conceptualization; methodology; investigation; supervision; funding acquisition; project administration; resources; writing—review and editing.

CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.

Supporting information
Supporting Information S1

Supporting Information S2

Figure S1: Publication bias and sensitivity analysis.

Table S1: Full search strategy for each database.

Table S2: Egger’s test for the incidence of ILD (All Grades).

Table S3: Egger’s test for the incidence of ILD (Grade ≥ 3).