Trastuzumab deruxtecan in HER2-positive advanced gastric cancer: exploratory biomarker analysis of the randomized, phase 2 DESTINY-Gastric01 trial

Abstract

Trastuzumab deruxtecan (T-DXd) showed statistically significant clinical improvement in patients with human epidermal growth factor receptor 2-positive (HER2⁺) gastric cancer in the DESTINY-Gastric01 trial. Exploratory results from DESTINY-Gastric01 suggested a potential benefit in patients with HER2-low gastric cancer. Spatial and temporal heterogeneity in HER2 expression or gene alteration, an inherent characteristic of gastric cancer tumors, presents a challenge in identifying patients who may respond to T-DXd. Specific biomarkers related to therapeutic response have not been explored extensively. Exploratory analyses were conducted to assess baseline HER2-associated biomarkers in circulating tumor DNA and tissue samples, and to investigate mechanisms of resistance to T-DXd. Baseline HER2-associated biomarkers were correlated with objective response rate (ORR) in the primary cohort of patients with HER2⁺ gastric cancer. The primary cohort had 64% concordance between HER2 positivity and HER2 (ERBB2) plasma gene amplification. Other key driver gene amplifications, specifically MET, EGFR and FGFR2, in circulating tumor DNA were associated with numerically lower ORR. Among 12 patients with HER2 gain-of-function mutations, ORR was 58.3% (7 of 12). ORR was consistent regardless of timing of immunohistochemistry sample collection. Further investigations are required in larger studies.

Fig. 1. Sample collection scheme.

Exploratory biomarker data were collected from primary and exploratory cohorts at baseline and EoT. Tumor tissue was selected for HER2 status and RNA-seq, and liquid biopsy was used to examine plasma and serum biomarkers. Plasma analysis examined ctDNA and serum biomarkers included HER2ECD. There were 91 patients in the HER2 IHC 3⁺ cohort and 28 in the HER2 IHC 2⁺/ISH⁺ cohort. ^aChanges in plasma HER2 amplification were categorized as either baseline (alterations detected at baseline) or acquired (alterations detected only in EoT samples). ^bGene alterations concomitant with T-DXd resistance. Amp., amplification; CGEP, comprehensive gene expression profile.

Fig. 2. Baseline HER2 expression in primary and exploratory cohorts.

a, Univariate analysis of baseline HER2 biomarker status and ORR in the primary cohort T-DXd arm. b, ORR based on biomarker status in the exploratory cohorts (HER2-low). c, OS by adjusted plasma HER2 copy number, HER2ECD in the primary cohort and HER2ECD in the exploratory cohorts. d, Forest plot of OS. a, HER2 mRNA gene expression median 9.72; serum HER2ECD median 9.72. b, Dashed vertical line represents overall ORR. a,b, Error bars represent 95% CI. d, HR < 1 favors the biomarker-selected group. Error bars represent 95% CI. ^aFor OS by HER2 ApCN, an exploratory cutoff (apCN 18.2) value was determined that minimized the P value estimated by log-rank test. Patients with values below 18.2 included those with no amplification. ^bFor OS according to HER2ECD, an exploratory cutoff value (14.4 ng ml⁻¹ in the primary cohort, 11.6 ng ml⁻¹ in the exploratory cohorts) was determined that minimized the P value estimated by log-rank test. Mut, mutant.

Fig. 3. HER2 status in ctDNA plasma and tumor samples.

a–c, Samples were assessed by HER2 apCN (a), HER2 (b) and serum HER2ECD (c). Minimum and maximum are represented by the whiskers, the box represents the first–third quarters, the center represents the median and dots represent individual samples. ^aOnly HER2 plasma-detected samples are shown; all cohorts are included except for two patients from the exploratory cohorts whose HER2 test results were missing. ^bBaseline only (48 samples); all cohorts are included. ^cBaseline only (166 samples); all cohorts are included except for two patients from the exploratory cohorts whose HER2 test results were missing. CPM, counts per million.

Fig. 4. Clinical outcomes in the primary cohort according to timing of tissue collection with respect to first Tmab treatment.

a, ORR in the study according to tissue collected before or after/during first Tmab treatment. b, OS analyzed over the course of the study according to tissue collected before or after/during first Tmab treatment. a, Error bars represent 95% CI; the center of error bars indicates ORR. ^aIncludes data for the response-evaluable set (all randomized patients who received at least one dose of study drug and had measurable tumors based on independent central review at baseline).

Fig. 5. Genomic alterations as determined by ctDNA.

a, Landscape of genomic alterations in baseline ctDNA^a. b, Relationships among ctDNA-detected alterations in key signal transduction genes (MET, EGFR, FGFR2, PIK3CA GoF, KRAS/NRAS), tumor mutation burden and ORR. ^aExcluding germline, putative clonal hematopoiesis of indeterminate potential and synonymous variants without annotations using OncoKB. Included any amplification type (focal and aneuploidy). Excluding two patients in exploratory cohorts because of missing HER2 test result. Highly frequent genes (≥15%) are shown. Patients were ordered by cohort, HER2 status, BOR and OS. ^bbTMB cutoff value was 20 mutations per megabase according to the Guardant Health OMNI platform. b, Error bars represent 95% CI. BOR, best overall response; MSI, microsatellite instability; PCHG, percentage change from baseline (tumor size by computed tomography scan).

Fig. 6. Analysis of patterns of gene variants at EoT.

a,b, Acquired/lost gene alteration from baseline to EoT by comparison of the change in gene amplification status (a) and by comparison of the change of SNV/Indel status (b). c, SNVs/Indels and amplifications from EoT samples. ^aTop ten acquired/loss of amplification from baseline to EoT, including both arms, ordered by P value. The McNemar test was used to compare change in gene amplification status from baseline to EoT regardless of amplification type. Shown is high-frequency amplification at either baseline or EoT in at least five patients among whom HER2 was the only significant amplification change (P = 0.0064). Multiple testing correction was not performed. ^bTop ten acquired/loss of SNVs/Indels from baseline to EoT, including both arms, ordered by P value. The McNemar test was used to compare change in SNV/Indel status from baseline to EoT regardless of function. Shown are high-frequency SNVs/Indels at either baseline or EoT in at least five patients. There were no significant SNVs/Indels. Multiple testing correction was not performed. ^cThe order of data is based on an algorithm aimed at determination of mutual exclusivity. The heatmap shows potential acquired SNVs/Indels detected only at EoT, regardless of function. C1D1, cycle 1, day 1.

Extended Data Fig. 1. HER2 status between matched archive and fresh biopsy samples.

Freq, frequency; HER2, human epidermal growth factor receptor 2; IHC, immunohistochemistry; ISH, in situ hybridization; T-DXd, trastuzumab deruxtecan. All samples (fresh and archival) were taken before T-DXd treatment. There were 41 samples included in this analysis. Two of the 41 paired samples were collected from non-matched sites. One archive sample was collected from ‘other’ as the site; this sample had a HER2 IHC status change from HER2 1+ to HER2 2 + /ISH−. Another archive sample was collected from liver as the site; this sample did not have a HER2 IHC status change (HER2 IHC 3+ to HER2 IHC 3+).

Extended Data Fig. 2. TOP1 acquired mutations identified in 3 patients at end of treatment.

BOR, best overall response; mo, months; PFS, progression-free survival; PR, partial response; SD, stable disease; TOP1, topoisomerase 1.

Extended Data Fig. 3. Differential expression of various genes in responders and nonresponders.

chr.loc, chromosome location; HER2, human epidermal growth factor receptor 2; RECIST, Response Evaluation Criteria in Solid Tumors. Results are ordered by statistical value, and genes in Chr 17q12 region are shown as triangles. To identify differentially expressed genes between responder (complete response or partial response) and nonresponder (progressive disease, stable disease, not evaluable) defined by RECIST, moderated t-test was used without multiple testing correction (P ≤ 0.01, absolute log2 FC ≥ 1).