Prognostic factors and treatment response in HER2-positive gastric cancer patients receiving trastuzumab deruxtecan: secondary analysis of the EN-DEAVOR study.

Introduction
Globally, Japan has the highest age-standardized incidence of gastric cancer for men and fourth highest for women.1 In 2022, the incidence of gastric cancer was 12.6%, and it was the third leading cause of cancer-related mortality in Japan.2 Human epidermal growth factor receptor 2 (HER2)-positive gastric cancer is one of the major subtypes of gastric cancer with a poor prognosis, and it constitutes ∼20% of all gastric cancer cases.3,4
Based on the results of the phase III Trastuzumab for Gastric Cancer (ToGA) trial, trastuzumab plus chemotherapy is recommended as the first-line treatment for patients with HER2-positive advanced gastric or gastroesophageal junction (GEJ) cancer5,6 as it demonstrated significantly longer median overall survival (OS) compared with chemotherapy alone.7 Ramucirumab plus paclitaxel is the preferred second-line treatment depending on patients’ prior therapy and performance status.5,6 The Japanese Gastric Cancer Association (JGCA) 2021 guidelines recommend third-line treatment with nivolumab, irinotecan, or trifluridine/tipiracil monotherapy in patients with unresectable advanced/recurrent gastric cancer who are in good general condition after second-line treatment.5 Based on the DESTINY-Gastric01 trial, which compared trastuzumab deruxtecan (T-DXd) versus physician’s choice of chemotherapy in patients with HER2-positive and HER2-low gastric or GEJ cancer who had progressed despite receiving previous therapies, including trastuzumab,8 the JGCA guidelines recommend that T-DXd should be prioritized for third-line treatment in patients with HER2-positive gastric cancer as it is the only drug for which significant survival prolongation has been confirmed versus chemotherapy regimens [objective response rate (ORR) 51% versus 14%, median OS 12.5 months versus 8.4 months].5,8
Before the approval of T-DXd for HER2-positive gastric cancer in Japan (September 2020), retrospectively analyzed data of a Japanese study [June 2015-July 2019 (N = 10 581)] in patients with advanced gastric cancer who started guideline-recommended first-line chemotherapy showed that only 50.6% and 22.6% of patients received second- and third-line chemotherapy, respectively.9 Similarly, another Japanese study (September 2017-March 2020) highlighted that nivolumab was the most commonly prescribed drug for third- or later-line treatment in patients with HER2-positive unresectable recurrent or metastatic gastric cancer, followed by irinotecan and trifluridine/tipiracil.10 Gastric cancer progresses more rapidly than other types of cancers, which may explain why only ∼50% and ∼22% of patients can proceed to second- and third-line treatments, respectively.9 Furthermore, patients with advanced gastric cancer have a poor general condition and are generally ineligible to receive third- or later-line treatments.5,11
EN-DEAVOR was a real-world study conducted in Japan to evaluate the effectiveness and safety of T-DXd in 312 patients with gastric cancer.12 The median real-world progression-free survival (rwPFS) was 4.6 months and ORR was 42.9%.12 Although studies have evaluated prognostic factors for the selection of nivolumab and irinotecan as third-line treatment for advanced gastric cancer,13, 14, 15, 16 data for T-DXd are limited.14,17 This is a secondary analysis of the EN-DEAVOR study to investigate the prognostic factors for T-DXd used as third- or later-line treatment in gastric cancer.

Patients and methods

Study design and eligibility criteria
The EN-DEAVOR study was a multi-institutional retrospective study conducted across 63 sites in Japan between 25 September 2020 and 30 September 2021 (study registration: UMIN000049032).12 The study was approved by the institutional ethical review board of the Nagoya University Graduate School of Medicine (No. 2022-0170). All patients provided informed consent to participate in this study. Patients aged ≥20 years with a histopathologically confirmed diagnosis of HER2-positive [immunohistochemistry (IHC) 3+ or IHC 2+ with in situ hybridization (ISH)+] unresectable advanced or recurrent gastric or GEJ adenocarcinoma that had worsened after chemotherapy and who had started treatment with T-DXd were included in the primary analysis.12 For this secondary analysis, patients who received T-DXd as third- or later-line therapy were included; those who received T-DXd as first- or second-line therapies were excluded.

Outcomes and assessments
The following details were collected from the medical records of participating patients and used as variables for a univariate analysis to determine the prognostic factors for rwPFS and ORR: sex, age, Eastern Cooperative Oncology Group performance status (ECOG PS), HER2 status (IHC and ISH) before T-DXd treatment, site of primary lesions, any surgeries for primary lesions, histological type of primary lesions, number of metastatic organs, metastasis sites (liver and peritoneum), ascites, modified Glasgow Prognostic Score (mGPS), biomarkers (lactate dehydrogenase, alkaline phosphatase, creatinine clearance, hemoglobin, and neutrophil-to-lymphocyte ratio), number of lines of previous therapy, nivolumab treatment history, and duration of trastuzumab treatment before T-DXd treatment. The median rwPFS according to the duration of prior trastuzumab treatment was calculated by the Kaplan–Meier method. The incidence of grade ≥3 adverse events (AEs) was recorded in patients with prior trastuzumab treatment ≥ median and < median.

Statistical analysis
For the univariate and multivariate analyses, P values were calculated for the Cox proportional hazards model and the logistic regression model. Factors with a significant P value (<0.05) in the univariate analysis were included in the multivariate model. The significance level for hypothesis testing was 5% two-sided. No adjustment for multiplicity was conducted. No extra complementing process was conducted if missing data were detected. Data are presented as hazard ratios (HRs) with 95% confidence interval (CI) for rwPFS and odds ratio (OR) with 95% CI for ORR. Statistical analysis was carried out using SAS ver. 9.4 (SAS Institute, Cary, NC).

Results

Patient disposition
In the primary analysis, 312 patients were included. After excluding 5 patients who had received T-DXd in the first- or second-line treatment of gastric cancer, 307 patients were included in the secondary analysis.

Demographics and baseline clinical characteristics
A total of 232 (75.6%) patients were male, 212 (69.1%) were aged ≥65 years, ∼90% of patients had ECOG PS 0-1 [0: 131 (42.7%) and 1: 137 (44.6%)], and 224 (73.0%) patients had target lesions (Table 1). According to the HER2 status survey items at the time of initial diagnosis of gastric cancer, 294/307 patients were enrolled as HER2-positive based on IHC and ISH findings, while 13/307 patients were either HER2-negative or lacked HER2 testing data. Sixteen patients underwent re-biopsy before receiving T-DXd. Among the 13 patients without documented HER2 status at the time of diagnosis, 3 were confirmed as HER2-positive before T-DXd dosing, making the total number of HER2-positive patients to 297 (IHC 3+ : 213, ISH 2+ and ISH+ : 84). The stomach was the site of the primary lesion in 261 (85.0%) patients; 103 (33.6%) patients had undergone surgery for the primary lesions, and the most common histological type of primary lesions was intestinal [169 (55.0%)]. Approximately 60% of patients had metastasis in two or more organs and ∼50% had liver metastasis. Ascites was present in 132 (43.0%) patients, and most patients had an mGPS score of 1 [134 (44.8%)]. Approximately 50% of patients had received three or more prior treatments for gastric cancer, with ∼40% having been treated with nivolumab. A total of 283 patients had received prior trastuzumab treatment, and the median treatment duration was 6.5 months (range 0-81.5 months).

Prognostic factors for rwPFS and ORR
In the univariate analysis, the following were positive prognostic factors for rwPFS: ECOG PS 0, HER2 status IHC 3+, any surgeries for primary lesions, intestinal type of primary lesions, no metastasis in the peritoneum, no ascites, mGPS 0 and 1, and duration of prior trastuzumab treatment ≥ median (≥6.5 months) (Table 2). In the multivariate analysis, HER2 status IHC 3+ (HR 0.65, 95% CI 0.49-0.86), intestinal type of primary lesions (HR 0.59, 95% CI 0.43-0.79), mGPS 0 and 1 (HR 0.71, 95% CI 0.53-0.95), and duration of prior trastuzumab treatment ≥ median (HR 0.75, 95% CI 0.58-0.97) were positive prognostic factors for rwPFS. ECOG PS, any surgeries for primary lesions, no metastasis in the peritoneum, and no ascites were not found to be prognostic factors for rwPFS in the multivariate analysis.
According to the Kaplan–Meier analysis, there was a trend for longer median rwPFS among patients with longer duration of prior trastuzumab treatment (Figure 1). The median rwPFS was 3.42 months (95% CI 2.46-3.94 months), 4.34 months (95% CI 2.83-5.09 months), 4.50 months (95% CI 3.42-6.24 months), and 6.01 months (95% CI 4.99-7.52 months) in patients with duration of prior trastuzumab treatment <25th percentile, ≥25th percentile to < median, ≥ median to <75th percentile, and ≥75th percentile, respectively.
In the univariate analysis, HER2 status IHC 3+, mGPS 0 and 1, and duration of prior trastuzumab treatment [≥ median (≥6.5 months)] were positive prognostic factors for ORR; however, only duration of prior trastuzumab treatment [≥ median (≥6.5 months)] was a positive prognostic factor for ORR in the multivariate analysis (OR 2.02, 95% CI 1.13-3.63) (Table 3).

Incidence rate of grade ≥3 AEs stratified by duration of prior trastuzumab treatment
Based on findings from the primary analysis, which showed a higher incidence of AEs in groups with greater efficacy, this sub-analysis investigated whether the frequency of grade ≥3 AEs differed according to trastuzumab treatment duration (≥median versus <median), a potential prognostic factor for both rwPFS and ORR in the multivariate analysis. The overall incidence rate of grade ≥3 AEs was similar in patients with duration of prior trastuzumab treatment [≥median (≥6.5 months)] and in those with T-DXd [<median (<6.5 months)] (48.2% versus 47.2%) (Table 4).

Discussion
This secondary analysis of the EN-DEAVOR study, which has the largest sample size included to date, highlights the clinical characteristics of patients with HER2-positive gastric or GEJ cancer who may show clinical benefits from T-DXd in the third- or later-line setting. The multivariate analysis identified that HER2 status IHC 3+, intestinal lesion type, mGPS 0 or 1, and prior trastuzumab treatment duration [≥ median (≥6.5 months)] were positive prognostic factors for rwPFS and prior trastuzumab treatment duration (≥ median) was the only positive prognostic factor for ORR. In the primary analysis,12 a trend was observed in which patients with greater treatment efficacy also had a higher incidence of grade ≥3 AEs. Based on this finding, the present sub-analysis examined whether there were differences in the occurrence of AEs according to the duration of prior trastuzumab treatment (≥median versus <median), one of the identified prognostic factors. The incidence of grade ≥3 AEs was comparable between patients with prior trastuzumab treatment duration <median and ≥median. The duration of prior trastuzumab treatment was a unique and strong prognostic factor in patients receiving T-DXd treatment. This finding may help in identifying the more appropriate patients with HER2-positive advanced gastric cancer for third- or later-line treatment with T-DXd in clinical practice.
Patients with intestinal type of primary lesions in the present study had a significantly longer rwPFS than those with diffuse type. The intestinal type of lesion is well differentiated, whereas the diffuse type is highly metastatic and characterized by rapid disease progression and a poor prognosis.18 A systematic review and meta-analysis suggested that patients with diffuse type of lesions had a worse prognosis than those with intestinal type of lesions for locoregional confined gastric cancer as well as advanced gastric cancer.19 These findings are consistent with a real-world study of nivolumab for third- or later-line treatment for advanced gastric cancer.16
In the present analysis, patients with HER2 status IHC 3+ had a significantly longer rwPFS compared with those whose HER2 status was IHC 2+ and ISH+. These findings are similar to those reported for OS in the DESTINY-Gastric01 subgroup analysis.8 Naturally, these findings underscore the importance of considering both tumor histology and HER2 status when selecting treatment strategies in later-line settings.
The multivariate analysis of ORR identified the duration of prior trastuzumab treatment as the only prognostic factor, highlighting that it could be a superior predictive marker for T-DXd effectiveness. HER2 status before T-DXd treatment was not found to be a prognostic factor, suggesting that HER2 heterogeneity or conversion to HER2-negative status due to prior therapy in gastric cancer is unlikely to significantly affect ORR. In the present study, patients with prior trastuzumab treatment duration [≥median (≥6.5 months)] had a significantly longer rwPFS and higher ORR than those with prior trastuzumab treatment duration (<median). Kaplan–Meier curves also showed a positive correlation between the duration of prior trastuzumab treatment and median rwPFS. Prior trastuzumab treatment duration was identified as a stronger prognostic factor than HER2 status as it was not influenced by HER2 expression heterogeneity. These findings suggest that prior trastuzumab treatment duration may help in identifying the more appropriate patients for T-DXd therapy, warranting further investigation in gastric cancer and other HER2-positive malignancies.
Patients with a duration of trastuzumab treatment before T-DXd therapy <25th percentile still achieved a median PFS of 3.42 months (although shorter than that of the other subgroups), thereby implying that T-DXd could be a viable treatment option even for patients with a shorter duration of prior trastuzumab treatment. The response to T-DXd in patients who failed to respond to trastuzumab may be attributed to the fact that T-DXd consists of the topoisomerase I inhibitor deruxtecan (payload) linked in an 8 : 1 ratio to trastuzumab,20 potentially overcoming resistance mechanisms. T-DXd penetrates adjacent tumor cells owing to its high membrane permeability, and thus it may offer a significant therapeutic benefit, the so-called bystander killing effect, and enhanced therapeutic potency on HER2-positive tumors, which tend to display intratumor heterogeneity.21 These findings suggest that even when prior trastuzumab treatment was not effective, T-DXd may still provide meaningful clinical benefits.
The Glasgow Prognostic Score (GPS) is an inflammation-based marker composed of elevated serum C-reactive protein (CRP) and decreased albumin concentration.22 The GPS is assigned as follows: 2 (patients with both elevated CRP and decreased albumin levels), 1 (patients with either an elevated CRP or decreased albumin level), and 0 (patients with normal CRP and albumin levels).22 The mGPS score is also based on serum albumin and CRP concentrations, but considers patients with hypoalbuminemia without elevated CRP as having low risk (mGPS = 0). In the present study, patients with mGPS 0 or 1 had a significantly longer rwPFS than those with mGPS 2. These findings are similar to those reported for nivolumab, where patients with a GPS of 0 had a significantly longer median OS than those with scores of 1 or 2,23 indicating that inflammation and nutritional impairment may lead to a reduction in tumor immunity and chemotherapy tolerance, thereby contributing to tumor progression.24 In later-line therapy, maintaining a patient’s good general condition is important, and the present study suggests that mGPS may be a better prognostic factor than ECOG PS. Therefore, measuring mGPS before starting T-DXd may be beneficial in determining the treatment course.
The choice of drugs after the failure of second-line treatment in patients with HER2-positive advanced gastric cancer should be made after a careful consideration of the patients’ clinical characteristics.13,15 Treatment with T-DXd reduced the tumor size in >80% of patients in the DESTINY-Gastric01 trial as well as in the real-world EN-DEAVOR study.8,12 In real-world Japanese studies of nivolumab used as third- or later-line treatment for patients with advanced gastric cancer, age ≥65 years, ECOG PS ≥1, presence of liver metastasis, HER2-positive status, and prior treatment with trastuzumab were reported as negative prognostic factors for survival.13,15 ECOG PS ≥1, presence of primary lesions, and presence of peritoneal metastasis were reported as negative prognostic factors for PFS in patients who received T-DXd in the third- or later-line setting for advanced gastric cancer.17 While certain trends were observed, age and the presence of liver and peritoneal metastases were not found to be statistically significant prognostic factors for rwPFS and ORR in the present study. The present study analyzed a large set of data collected across different centers in Japan and is one of the largest studies on T-DXd therapy. Considering the large dataset, it is hoped that the study has provided a clearer understanding of prognostic factors, which can be used to provide important guidance for future treatment strategies.
This study identified several clinical factors, including the duration of prior trastuzumab treatment, mGPS, HER2 expression intensity, and histological subtype, as potential predictors of treatment efficacy in patients with HER2-positive advanced gastric cancer receiving T-DXd in the third-line setting or beyond.
Meanwhile, T-DXd has already been introduced as a second-line therapy in western countries. With this shift toward earlier-line use, clinical attention to changes in HER2 status at the time of treatment (HER2 conversion) and the impact of molecular-level heterogeneity is expected to become increasingly important.25 Although this study did not evaluate dynamic changes in HER2 expression due to the low re-biopsy rate (<3%), the duration of prior trastuzumab treatment may serve as an objective prognostic indicator that is relatively unaffected by such biological variability.
Going forward, we hope that real-world clinical studies in western settings—where T-DXd is already established as a second-line therapy—will also explore the potential prognostic value of the duration of prior trastuzumab treatment.
However, there are a few limitations. Firstly, the study population was limited to a Japanese cohort, and caution is warranted when generalizing the findings to other populations or countries. In particular, the low rate of HER2 re-evaluation among patients with HER2-positive gastric cancer (<3%) and the lack of assessment of HER2 heterogeneity or conversion should be taken into consideration when comparing with western data. Secondly, the trastuzumab-free interval (TFI) was not included in this multivariate analysis, despite evidence suggesting that a longer TFI enhances the effectiveness of T-DXd.12 This decision was made based on the likelihood that TFI is confounded by various factors, such as patient condition, number of prior treatment lines, and response to second-line therapy. Additionally, although a significant difference in rwPFS was observed in the primary analysis, no such difference was found for ORR, which also contributed to the decision. The present results similarly suggest that patients with a shorter TFI tended to have a higher prevalence of background factors associated with poor prognosis, such as poor PS and ascites, which may have influenced the PFS outcomes.
The association between TFI and prognosis is an important subject for further investigation and will be explored in a future sub-analysis. Thirdly, as this was a retrospective analysis of real-world clinical data, there are potential confounding factors for which data could not be collected, such as comorbidities, tumor burden, HER2 heterogeneity, and HER2 status before T-DXd administration, and tumor markers.

Conclusions
HER2 status IHC 3+, intestinal type primary lesions, mGPS 0 and 1, and longer duration of prior trastuzumab treatment [≥median (≥6.5 months)] were identified as positive prognostic factors for rwPFS in patients with advanced gastric or GEJ cancer treated with T-DXd in the third- or later-line settings. The longer duration of prior trastuzumab treatment (≥median) was the only positive prognostic factor for ORR. These findings provide insights into the treatment strategies for patients with advanced gastric cancer, and patients with the stated prognostic factors may be recommended for T-DXd. Patients with favorable prognostic factors, such as low mGPS, long prior trastuzumab exposure, HER2 status IHC 3+, and intestinal type primary lesions should be recommended for T-DXd to maximize outcomes before patient condition worsens.