Introduction
Human epidermal growth factor receptor 2 (HER2) plays a crucial role in tumor growth, invasion, and development [
1,
2]. Extensive studies have shown that HER2 expression is closely related to the occurrence of various tumors and is one of the most important targets for developing anticancer therapies [
3,
4]. Antibody-drug conjugate (ADC), which is composed of a monoclonal antibody linked to a cytotoxic agent, is revolutionizing targeted cancer therapy [
5]. HER2, as a classic tumor target, has become an ideal target for the development of ADC drugs due to its high specific expression in tumor tissue and its high efficiency in mediating the endocytosis of ADC drugs. To date, the US Food and Drug Administration (FDA) has approved two ADC drugs (T-DM1 and T-DXd), and over 60 HER2-targeted ADC candidates are currently undergoing clinical trials [
6].
HER2-targeted ADCs have shown excellent efficacy and have had their indications expanded on the strength of their ingenious design of the molecular structure that delivers cytotoxic drugs specifically to cancer cells [
7‐
9]. However, fatal adverse events have been reported with HER2-targeted ADCs due to undesired uptake in healthy cells. Some of these adverse events can be life-threatening, such as pneumonitis, hematotoxicity, cardiotoxicity, and hepatotoxicity [
10‐
12]. The occurrence of fatal adverse events is difficult to avoid during anticancer treatments and causes great harm to patients and their families. In order to improve the treatment compliance of patients and ensure that patients can continue to benefit from the treatment of HER2-targeted ADCs, it is particularly important to clearly understand the profile of fatal adverse events and timely develop management measures. And the analysis of fatal adverse events will help to improve the guidelines and provide guidance for better guiding the clinical application of HER2-targeted ADC drugs.
Multiple clinical trials have been concerned with the fatal adverse events caused by HER2-targeted ADCs, but the limited number of patients in each trial has left the overall incidence and risk of such events unclear. Therefore, in this study, we conducted a systematic review and meta-analysis of the mortality profile of HER2-targeted ADCs. Utilizing a Bayesian hierarchical modeling approach, we quantitatively combined data from randomized controlled trials (RCTs) to address the incidence and risk of fatal adverse events in cancer patients treated with HER2-targeted ADCs. We aim to provide clinicians with a reference to use HER2-targeted ADCs appropriately and manage potential fatal adverse events related to these drugs.
Methods
Search methods and study selection
The present systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guideline (
eTable 1 in the Supplement) [
13,
14]. The protocol was registered in PROSPERO (No. CRD42022331627). We systematically searched the PubMed, Embase, Web of Science, and Scopus databases from inception to February 1, 2022, and updated the last search to July 1, 2023. The keywords, including “HER2 ADC”, “trastuzumab emtansine (T-DM1)”, and “trastuzumab deruxtecan (T-Dxd)”, are employed and full search strategy is detailed in
eTable 2 in Supplement. We also manually screened reference lists from relevant review articles to supplement the search. This study excluded non-randomized trials, editorials, correspondences, and reviews. We included only prospective RCTs of HER2-targeted ADC agents in the treatment of cancer patients. The inclusion criteria were based on the PICO-framework. In detail, Population (P): cancer patients; Intervention (I): treatments by HER2-targeted ADC agents. Comparison (C): The other therapeutics. Outcomes (O): any fatal adverse events. When publications reported the same trial, the most recent one was included. Two reviewers (JY Xie and C Zhang) were independently responsible for screening and reviewing the included literature and ZW Fu was involved to discuss and reach a consensus when there is a controversy.
The primary outcome was the mortality profile of HER2-targeted ADCs, including the incidence and relative risk (RR) of the fatal adverse event, which was defined as deaths caused most likely by the treated ADC drugs. Baseline information, such as authors, publication years, trial names, NCT numbers, trial phases, cancer types, and the used ADC therapies, were also extracted from each included trial.
Statistical analysis
The incidences of fatal adverse events were calculated by dividing the number of patients who experienced adverse events by the total number of patients. Since the incidence of fatal adverse events is typically low and some studies included zero events in the treatment and/or control groups, the variance for such studies approaches zero. Consequently, the weight of these studies would be overestimated in a classic meta-analysis. To address this issue, we utilized a Bayesian hierarchical modeling strategy to conduct the meta-analysis of fatal adverse events in this study.
For the ith study which reported the dichotomous outcomes, the number of patients with reported any adverse events in the ith study followed the binomial distribution: ri ~ binomial (𝑛𝑖, 𝑝𝑖), where 𝑛𝑖 was the total number of investigated populations and 𝑝𝑖 was the incidence of adverse events for the ith study. The logit transformation of 𝑝𝑖 followed a normal distribution among studies: 𝜃𝑖 = logit(𝑝𝑖) ~ normal (𝜇, 𝜎2), where 𝜇 was the mean of logit(𝑝𝑖) and 𝜎2 was the between-study variance. Then we could estimate the pooled incidence of adverse events and the corresponding 95% credible interval (CrI) through retransform: Incidence = exp (𝜇)/(1 + 𝑒𝑥𝑝 (𝜇)).
Bayesian inference was utilized to estimate the pooled effect sizes through the combination of prior information and observed sampling distribution [
15]. The Bayesian random effects model was applied to generate the estimates of the overall incidence and relative risk compared to the control arm, along with a 95% credible interval (CrI). The CrI represented the 2.5–97.5 percentiles of the posterior distribution of the estimation. For the mean parameters of normal distributions, a proper prior distribution with mean = 0 and sd = 4 was proposed. The between-study variances were assigned weakly informative normal prior distributions with the mode at 0 and the scale at 1. The posterior distribution of interest outcomes was estimated using the Markov Chain Monte Carlo (MCMC) algorithm and Gibbs sampling in all Bayesian hierarchical models [
16]. The statistical heterogeneity among the included studies was quantified using the between-study variances (τ) in this Bayesian meta-analysis, with lower values of τ indicating smaller heterogeneity. Significant heterogeneity was considered substantial if τ exceeded 1.5. The Bayesian forest plots of the meta-analysis provided both study estimates and shrinkage estimates, thereby allowing for a more comprehensive analysis of the pooled effect sizes.
The risk of bias in each included RCTs was evaluated using the revised Cochrane Risk of Bias tool (version 2.0) [
17]. In addition, a classic funnel plot was performed to detect any possible publication bias, as it can impact the validity of meta-analyses. All statistical analyses were conducted using the R program with packages “meta,“ “metafor,“ “bayesmeta,“ and “forestplot,“ which were used to extract and analyze the data in this study. The detailed code for the R program was provided in Supplementary File 1.
Discussion
Based on the 13 clinical RCTs, including 4,246 patients, the pooled incidence of fatal adverse events in patients treated with a HER2-targeted ADC was 0.078% compared with 0.095% in patients from control arms, and this risk was lower but not significantly than that with the control arm (RR = 0.8; 95% CrI, 0.5–1.26, τ = 0.17). These findings suggest that HER2-targeted ADCs may be a relatively safe and comparable alternative to standard conventional therapies for cancer patients, making them promising novel therapeutic options in clinical settings.
As an emerging biopharmaceutical drug, the HER2-targeted ADCs have provided promising alternative ways to fight against cancer [
29]. However, the information on fatal adverse events associated with HER2-targeted ADCs remains unclear. Cancer therapy using HER2-targeted ADCs is a double-edged sword. While focusing on efficacy, we should also pay close attention to the adverse event caused by drugs, especially the fatal ones, since it severely impacts patients and their families. Therefore, it is necessary to investigate the incidence of fatal adverse events to properly evaluate the benefit-risk ratio and make decisions in the oncology clinic. Our study showed that treatment with HER2-targeted ADCs results in 0.078% of patients dying due to adverse effects of ADC treatment alone. It also revealed a lower risk of fatal adverse events compared to the other standard therapies (RR = 0.8). These data should be essential in considering whether to use HER2-targeted ADCs treatment.
Among 43 reported deaths, the most common cause of death caused by HER2-targeted ADCs was respiratory toxicity, including six pneumonia, three pneumonitis, and one each of interstitial lung disease, pulmonary embolism, pneumonia aspiration, lung infection, bronchopneumonia, dyspnea, and atypical pneumonia. Besides, hematologic toxicity, infection, and hepatic toxicity accounted for the other leading cause of death. This meta-analysis demonstrates that the risk of fatal adverse events with HER2-targeted ADCs is comparable to conventional anticancer therapy. Therefore, it is crucial to closely monitor patients receiving HER2-targeted ADCs for symptoms related to the respiratory system, infection, and liver functions. Early recognition and management of toxic effects, including prompt initiation of dose reduction and other modulating agents like glucocorticoids, are essential for preventing fatalities.
Previous meta-analyses have estimated the incidence and risk of adverse events associated with T-DM1, but they did not analyze fatal events [
30‐
33]. There was also a meta-analysis that investigated the incidence of general adverse events related to antibody-drug conjugates in all clinical trials, including lymphopenia, nausea, neutropenia, peripheral neuropathy, and blurred vision [
34]. However, our study concentrated on fatal adverse events instead of general adverse events and also focused on the HER2-targeted ADCs, which represent the most common subclass of ADCs. Our meta-analysis is the largest to date, including 4,246 patients from high-quality RCTs, and provides a summary of HER2-targeted ADCs-related fatal adverse events in cancer patients. The results revealed that the toxicities of the respiratory system and myelosuppression attributed to the leading cause of death among patients receiving HER2-targeted ADCs. The reasonable interpretation of that was the high amounts of FcγR expressed in alveolar macrophages and myeloid cells and Fc-mediated non-specific uptake of HER2-targeted ADCs might contribute to these fatal adverse events [
35,
36]. Consequently, the next generation of HER2-targeted ADCs could consider optimizing the Fc fragments of the antibody part of ADCs, such as an increase in serum stability or improvement of binding specificity and affinity.
However, some limitations could be improved in our study. Firstly, our study relied on study-level data, and individual patient-level confounding factors could not be thoroughly assessed or included in the analysis. Secondly, since the primary outcomes of included RCTs were focused on the efficacy of HER2-targeted ADCs, the fatal adverse events were reported through different investigators and institutions, which might introduce potential bias in the assessment of whether fatal adverse events were associated with the treatment of HER2-targeted ADCs. Thirdly, probably because of the small sample size and potential reasons related to cancer, our analysis showed no evident risk difference between HER2-targeted ADCs and control-arm therapies. Finally, due to the scarcity of studies involving T-DXd in non-breast cancer patients, the pre-defined stratification factors (i.e., drug and cancer types) were insufficient to detect significant differences in the risk associated with distinct HER2-ADCs or cancer types.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.