Efficacy and safety of anamorelin for cancer cachexia in patients with unresectable or recurrent gastric cancer: a multicentre, open-label, randomised controlled trial.

Introduction
Cancer cachexia is a multifactorial syndrome characterized by a progressive involuntary skeletal muscle mass loss, with or without fat loss, which cannot be fully reversed by conventional nutritional support, leading to significant functional impairment and poor prognosis.1 Previously reported incidence of cancer cachexia rates ranging from 50% to 80%, with highest incidence reported in patients with lung and gastrointestinal cancer.2 The pathophysiology of cancer cachexia involves a complex interplay between tumor-derived factors and host inflammatory responses, leading to metabolic alterations and catabolic processes.3,4 Despite its profound impact on prognosis and QOL, effective therapeutic options for cancer cachexia remain limited, leaving a significant unmet clinical need.
Anamorelin, a novel ghrelin receptor agonist, has emerged as a promising candidate for the treatment of cancer cachexia.5 Ghrelin, was first reported as endogenous ligand of orphan receptor,6 that was Growth Hormone Secretagogue Receptor, extracted from rat and human stomachs, has been reported to have various bioactivity such as appetite stimulation,7 weight gain, increase of muscle mass,8 and suppression of excessive inflammation9 The ROMANA 1 and 2 trials demonstrated the efficacy of anamorelin to improve appetite, lean body mass (LBM), and total weight in patients with non-small cell lung cancer-associated cancer cachexia.10 A smaller scale phase II trial11 targeting gastrointestinal cancer cachexia including colon cancer, gastric cancer, pancreatic cancer, demonstrated efficacy, however only five cases of gastric cancer were included, leaving limited information regarding the effect of anamorelin in gastric cancer patients. Given that ghrelin is a gastric-derived hormone6 and that plasma ghrelin levels are known to persistently decreased in patients with gastric cancer those who underwent gastrectomy,12 combined with the high prevalence of severe malnutrition observed both after surgery and at recurrence in these patients, anamorelin might exert greater therapeutic effects in this population. On the other hand, a previously reported small-scale Phase II study showed a lower population of responders to anamorelin among patients with gastric cancer compared to those with colorectal or pancreatic cancer,11,13 and its efficacy in gastric cancer patients remains controversial. To address this unresolved question, we conducted a first randomized controlled trial to evaluate the safety and efficacy of anamorelin in patients with unresectable/recurrent gastric cancer patients complicated by cancer cachexia.

Methods

Study design and population
This study was a multicenter, open-label, randomized controlled trial to evaluate anamorelin administration on change of LBM for unresectable advance or recurrent gastric cancer patients with cachexia who were receiving first to third-line chemotherapy. Participating hospitals were 10 hospitals in Japan. This study was conducted and reported in accordance with the CONSORT 2025 guidelines for randomized controlled trials. Statistical analysis advice, protocol development, electronic data capture (EDC) construction support, and data management are provided by the Department of Future Medical Development at Osaka University, while monitoring is conducted by P-Pro Japan Co., Ltd. Enrolled patients meet all of the following eligibility criteria below and not violate any of the exclusion criteria. Eligibility criteria; (1) Recurrent gastric cancer over 3 months after gastrectomy or unresectable advanced gastric cancer, (2) Receiving first to third-line chemotherapy, (3) Involuntary weight loss of ≥5% observed within the previous 6 months, (4) Fulfilling at least 2 items of following criteria; (1. Fatigue, 2. Decreased general muscle strength, 3. Fulfilling at least one item of following criteria (i. C-Reactive Protein (CRP) > 0.5 mg/dL, ii. Hemoglobin<12 g/dL, iii. Albumin<3.2 g/dL)), (5) Aged≥20 years at the time of informed consent, (6) Life expectancy ≥ 4 months from the time of informed consent, (7) Eastern Cooperative Oncology Group (ECOG) performance status of 0–2,14 (8) Adequate organ (Bone marrow, hepatic, renal) function, (9) with written informed consent. Exclusion criteria; (1) Previous history of administration of anamorelin, (2) Previous history of allergy for ingredients of anamorelin, (3) Previous history of congestive heart failure, (4) Myocardial infarction or cardiac angina, (5) Severe dysfunction of impulse conducting system (complete AV block etc.), (6) Taking prescription medications with: Clarithromycin, Indinavir, Itraconazole, Nelfinavir, Saquinavir, Telaprevir, Voriconazole, Ritonavir, Cobicistat, (7) Moderate liver dysfunction (Child-Pugh classification B or C), (8) Inability of taking orally due to bowel abnormality (e.g., bowel obstruction), (9) Poor control of diabetes mellitus, (10) Pregnant or possibility of pregnancy, (11) Active synchronous or metachronous malignancies (excluding carcinoma in situ), (12) Self-contained mechanical device such as pacemaker, (13) Patients considered inappropriate for inclusion by the attending physician due to concerns regarding patient safety and protocol compliance. Sex was self-reported by participants. The study was approved by the institutional review board at each participating site (Supplementary Table S1) and conducted in accordance with the Japanese Ethical Guidelines for Clinical Research and the Declaration of Helsinki. Written informed consent was obtained from all participants before study enrolment. The trial was registered with the Japan Registry of Clinical Trials (jRCT) prior to initiation. (jRCTs051210108; registered on October 13, 2021.).

Randomization and masking
The patients enrolled in the trial were randomly assigned to either the anamorelin group (Group A) or the non-anamorelin group (Group N) in a 1:1 ratio. Allocation to the treatment group was performed using a stratified permuted block randomization method, with stratification factors including the participating institution and the type of surgical procedure (No gastrectomy, partial gastrectomy (non-TG), or total gastrectomy (TG)). The random sequence was generated by an independent statistician using a computer-based random number generator. Allocation concealment was ensured by centralized web-based registration operated by an independent data center. This was an open-label trial, and no masking was applied. All participants, investigators, outcome assessors, and statisticians were aware of the assigned treatment groups.

Procedures
Patients assigned to Group A were orally administered 100 mg of anamorelin once daily in the morning on an empty stomach, starting on the day of allocation or the following day, for a duration of 12 weeks. Anamorelin was provided as a commercial formulation (Adlumiz®, manufactured by Ono Pharmaceutical Co., Ltd., Osaka, Japan). Patients in the Group N did not receive anamorelin, and no placebo was employed. Anamorelin was discontinued in cases of adverse events, disease progression, deterioration in general condition, or withdrawal of consent. Treatment duration was recorded from initiation to discontinuation. Although body composition can be assessed using CT or DXA, this study was a multicenter trial, and it was ethically and practically difficult to perform repeated CT or DXA scans every four weeks for all participants. Therefore, total weight and body composition, including LBM and fat mass were measured using the TANITA body composition analyzer DC-430A-N with the bioelectrical impedance analysis (BIA) method,15 which was prepared for this trial across all participating hospitals, at least 2 h of fasting. The LBM values were automatically calculated by the device using the manufacturer’s proprietary algorithm. The exact formula used in this device is not publicly disclosed. Measurements were conducted at baseline, weeks 4, 8, and 12s. The allowable measurement window was within 4-weeks prior to registration for baseline, within ±2 weeks for the 4 weeks and 8 weeks measurement, and within 2 weeks after the 12 weeks measurement. Handgrip strength was measured using a grip dynamometer from N-FORCE Co., which was standardized across all participating hospitals. Measurements of handgrip strength were taken twice on each hand, and the maximum value of the non-dominant hand was adopted. The timing of measurements and the allowable time window were same as body composition assessment. Adverse events during study period and the evaluation of the three selected symptoms (anorexia, nausea, and vomiting) at 8 weeks were assessed, regardless of chemotherapy regimen or schedule. Blood tests and biochemical analyses were conducted at the same time points as body composition measurements. Abdominal computed tomography (CT) scans were performed at baseline and 8 weeks. Quality of life (QOL) questionnaires, including the Cancer Fatigue Scale16 and Questionnaire for Cancer Patients treated with Anti-Cancer Drugs (QOL-ACD),17,18 were completed by patients at baseline, 4 weeks, and 8 weeks. For safety monitoring, electrocardiograms (ECGs) were performed at baseline and, for Group A only, at 8 weeks and 12 weeks.

Outcomes
The primary, secondary, and safety analyses were performed in the modified per-protocol population, which included all patients who received at least one dose of the assigned treatment and did not meet major protocol violations. The primary endpoint was the change in LBM at 8 weeks from baseline. Although anamorelin was administered for 12 weeks, previous studies such as ROMAN 1 and ROMANA 2 have demonstrated its efficacy as early as week 3 and week 6, with these effects sustained over a prolonged period.10 Therefore, in order to minimize dropout due to disease progression, the primary endpoint in this trial was set at 8 weeks after randomization.
The secondary endpoints were as follows: (1) Changes in total weight, fat mass, and handgrip strength at 8 weeks from baseline; (2) Serial changes in LBM, total weight, fat mass, and handgrip strength at baseline, weeks 4, 8, and 12; (3) Change from baseline to week 8 in nutritional blood biomarkers, including serum albumin, prealbumin, total protein, total cholesterol, lymphocyte count, and CRP levels; (4) QOL assessment, changes from baseline over 8 weeks in the Cancer Fatigue Scale and QOL-ACD scores, as well as 2 specific items related to appetite from the QOL-ACD questionnaire: Q8 “Did you have an appetite?” and Q9 “Did you find food to be tasty?”; (4) Tumor response evaluation based on RECIST version 1.1 criteria19 using CT imaging taken at week 8; (5) Comparisons of 3 selected symptoms at week 8: anorexia, nausea, and vomiting; (6) Incidence and severity of all adverse events during study period. In addition, for the primary endpoint, subgroup analyses were conducted to explore factors that may influence the efficacy of anamorelin. Patients who have never received anamorelin in Group A were excluded from the safety and efficacy populations, in accordance with the per-protocol analysis defined in the trial protocol.

Statistical analysis
All statistical methods were pre-specified in the study protocol. To ensure the reliability of this study, central and on-site monitoring were implemented by P-Pro Japan Co., Ltd. in accordance with the monitoring manual, with monitors appointed by the principal investigator. All statistical methods were pre-specified in the study protocol. Data handling procedures for all enrolled patients were determined by the principal investigator and lead statistician and documented in the final report prior to data lock.
As the primary analysis, the change in LBM at 8 weeks from baseline between the two groups was compared using an unpaired t-test. The significance level for the test is set at 5%, and a two-sided 95% confidence interval for the mean difference in change of LBM was calculated. In addition, an analysis of covariance (ANCOVA) adjusting the type of surgical procedure and baseline LBM, is pre-specified. The sample size for this study was determined based on the results of the previously reported ROMANA-1 and ROMAN-2 trials.10 Assuming that the mean increase in LBM in the Group A compared to the Group N was 1.49 kg, with a standard deviation of 3.44 kg, a statistical significance level of α = 0.05 (two-sided), and a statistical power of 1-β = 0.80, the required sample size to detect a statistically significant difference is estimated to be 85 patients per group. Considering a dropout rate of 15%, the final sample size is set at 100 patients per group. Secondary endpoints included, changes in body weight, fat mass, handgrip strength, nutritional blood biomarkers, Cancer fatigue scale,16 and QOL-ACD scores17,18 from baseline to week 8 were compared between groups using an unpaired t-test. Serial changes in LBM, total weight, fat mass, and handgrip strength from baseline are analyzed using a mixed-effects model. In this model, the change in LBM from baseline is set as the dependent variable, with treatment group, time point, and the type of surgical procedure as fixed effects, and patient as a random effect. The interaction between treatment group and time point is also included as a fixed effect to compare trends between groups. All available outcome data at each time point, including those from patients with incomplete follow-up, were included in the model under the assumption of missing at random. Tumor response to chemotherapy was assessed according to RECIST version 1.1 criteria19 separately for each line of treatment (first-line, second-line, and third-line). Objective response was defined as the proportion of patients achieving complete response (CR) or partial response (PR). The objective response rate (ORR) and its 95% confidence interval were calculated for each treatment line. Between-group comparisons were performed using Fisher’s exact test. The frequency of three selected chemotherapy-related adverse events at week 8 were categorized as Grade 0–1 and Grade≥2 according to CTCAE version 5.0, and compared between groups using Fisher’s exact test. The analysis of adverse events is conducted on the safety populations, with the number and proportion of cases reported by event and severity for each treatment group. For the primary endpoint, subgroup analyses are presented using forest plots to demonstrate confidence intervals for following subgroups: (1) Sex (male, and female), (2) body mass index (BMI) (≥20kg/m2, and <20kg/m2), (3) CRP (≥1.0 mg, and <1.0 mg), (4) The type of surgical procedure (no gastrectomy, non-TG, and TG), (5) Line of chemotherapy (first-line, and second or third-line), (6) ECOG performance status (PS 0–1, and PS 2), and (7) Usage of ONS (present, and absent). While Subgroup analyses were pre-specified in the study protocol, the inclusion of sex, BMI, and CRP as specific subgroup factors was not pre-specified in the protocol and should therefore be considered post hoc exploratory analyses. Although multiple analyses were planned as secondary endpoints and the issue of multiple comparisons exists, no adjustment for multiplicity was made regarding the significance level. In the study protocol, changes in skeletal muscle indices assessed by CT imaging and changes in serum ghrelin levels from baseline to week 8 were originally planned as secondary endpoints. However, as the data for these outcomes were not finalized at the time of submission, they are not included in the present manuscript and will be reported separately in a future publication. Categorical variables are presented as the number of cases and corresponding percentages. Continuous variables are expressed as median with interquartile ranges (IQR) for background characteristics, and as means with 95% confidence intervals for primary and secondary endpoints. Background characteristics were compared between groups using Fisher’s exact test for categorical variables and Wilcoxon rank sum test for continuous variables. All statistical tests were two-sided and P values of 0.05 or less were deemed statistically significant. Statistical analyses were conducted with JMP Pro 17.2.0.

Role of the funding source
This study was funded by Ono Pharmaceutical Co., Ltd. The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. KY and TY had full access to all the data in the study. KY, YK, YM, and YD had final responsibility for the decision to submit the manuscript for publication.

Results
From November 17, 2021 to July 4, 2024, informed consent was obtained from 217 patients. Among them, 14 patients did not meet the eligibility criteria. Of 203 patients who were subsequently randomized, 104 were allocated to Group A and 99 to Group N. There were no major protocol deviations in either group. In Group A, two patients did not receive any dose of anamorelin after randomization and were therefore excluded from both the safety and efficacy analyses. In each group, one patient had no safety or efficacy data available, resulting in 101 patients in Group A and 98 patients in Group N being included in the safety analysis. For efficacy analysis, one patient in Group N had no efficacy data collected and excluded from the efficacy population.
Within 12 weeks of randomization, treatment discontinuation occurred due to disease progression in 17 patients in Group A and 12 patients in Group N, and due to withdrawal of consent in 2 patients in Group A. As a result, 82 patients in Group A and 85 patients in Group N completed the trial (Fig. 1).
Table 1 shows the comparison of baseline characteristics of the efficacy population between the two groups. In both groups, approximately 40% of patients had not undergone gastrectomy, another 40% had undergone partial gastrectomy, and 20% had undergone total gastrectomy. Regarding the line of chemotherapy, 70% were receiving first-line, 20% second line, and 10% third-line treatment. Details of the chemotherapy regimens are described in Supplementary Table S2. No significant differences were observed between groups in terms of the use of ONS, present or volume of ascites, or metastatic site. As for body composition data, LBM tended to be higher in Group A, possible reflecting the slightly greater proportion of male patients, although the difference was not statistically significant (42.4 versus 39.8 kg, P = 0.092). No significant differences were found between groups in total weight, fat mass, handgrip strength, Cancer Fatigue Scale scores, or QOL-ACD scores. The median (IQR) duration of anamorelin administration was 82 (63–103) days, approximately 12 weeks, with relatively few patients receiving the drug for less than 4 weeks (11 cases) or for 4–8 weeks (13 cases).
The Primary endpoint of mean change in LBM at 8 weeks from baseline was +0.99 (95% CI: +0.34 to +1.64) kg in Group A and +0.14 (95% CI: −0.49 to +0.77) kg in Group N, with a between-group difference of 0.85 kg (95% CI: −0.05 to +1.75; P = 0.063) by unadjusted t-test (Table 2). Although this did not reach statistical significance, the pre-specified ANCOVA model adjusting for baseline LBM and the type of surgical procedure showed a significant between-group difference (estimated marginal mean: 1.077 kg; 95% CI: 0.184 to 1.969; P = 0.018).
The change in total weight, fat weight, handgrip strength, and blood-based nutritional parameters at 8 weeks from baseline was indicated in Table 2. The mean changes at 8 weeks from baseline in total body weight (+1.17 kg [95% CI, +0.56 to +1.78] versus −0.59 kg [95% CI, −1.22 to +0.04]; P < 0.0001) and fat mass (+0.15 kg [95% CI, −0.40 to +0.70] versus −0.73 kg [95% CI, −1.37 to −0.08]; P = 0.040) was significantly greater in Group A compared with Group N. Notably, the increase in total weight was prominent despite similar baseline rates of severe malnutrition as defined by the World Health Organization (WHO) (BMI <16 kg/m2) between the groups (12% versus 11%) (Table 1). At 8 weeks, the proportion of patients with BMI < 16 kg/m2 was significantly lower in Group A (2% versus 13%; P = 0.0057). No significant difference was observed between the groups in the changes in handgrip strength. Regarding blood nutritional markers, change at 8 weeks from baseline in prealbumin and total cholesterol were significantly greater in Group A than in Group N, whereas no significant differences were observed for other laboratory parameters.
Fig. 2 illustrates the serial changes (mean [95%CI] from baseline at 4, 8, and 12 weeks) in LBM, total weight, fat mass, and handgrip strength, along with the number of patients assessed at each time point and P values from a mixed-effects model. The model included treatment group, time point, the type of surgical procedure (No gastrectomy, non-TG, or TG), and the interaction between treatment group and time point as fixed effects, with patients as random effects. A significant difference in LBM was observed at week 8 but not at weeks 4 or 12. In contrast, for total weight, significant differences in the change from baseline were observed at all time points. No significant differences were observed in handgrip strength at any time point.
The changes from baseline to week 8 in the Cancer Fatigue Scale16 and QOL-ACD scores17,18 are presented in Supplementary Table S3. The Cancer Fatigue Scale was divided into physical fatigue, emotional fatigue, cognitive fatigue, and total fatigue domains, but no significant differences were observed between Group A and Group N in any of these domains. The QOL-ACD was analyzed across the domains of activity, physical condition, mental and psychological condition, social functioning, facial scale, and total score. Additionally, within the physical condition domain, which includes items related to appetite and enjoyment of meals, changes in Q8 and Q9 were analyzed separately. Among the domains, only physical condition showed a significant improvement in the Group A compared to the Group N (+1.8 [95% CI +0.9 to +2.7] versus 0.0 [−1.0 to +1.0]; P = 0.0091). Additionally, both Q8 (+0.9 [+0.5 to +1.2] versus +0.2 [−0.2 to +0.5]; P = 0.0037) and Q9 (+0.7 [+0.3 to +1.0] versus 0.0 [−0.3 to +0.4]; P = 0.011) within the physical condition domain showed significant improvements in the Group A.
The efficacy of chemotherapy and the frequency of three selected chemotherapy-related adverse events at week 8 is presented in Supplementary Table S4. The efficacy of chemotherapy was compared by treatment line (first-, second-, and third-line therapy). According to the RECIST criteria, the objective response rate in first-line therapy was 31.7% (21.3–44.2) in Group A versus 28.1% (18.1–40.8) in Group N; in second-line therapy, 12.5% (3.5–36.0) versus 21.1% (8.5–43.3); and in third-line therapy, 14.3% (2.5–51.3) versus 0%. No significant differences were observed between the two groups at any treatment line. Regarding three selected chemotherapy-related adverse events at 8 weeks (anorexia, nausea, and vomiting), the incidence of Grade 2 or higher anorexia (10.6% versus 23.2%, P = 0.022) and nausea (0% versus 6.3%, P = 0.013) was significantly lower in Group A. The incidence of vomiting was low in both groups (3.2% versus 4.2%), with no significant difference.
Supplementary Table S5 summarizes the incidence and severity of all adverse events observed in both groups during the study period. With respect to hyperglycemia, a known adverse event associated with anamorelin, no cases were observed in Group N. In Group A, one patient (1%) experienced Grade 3 hyperglycemia, and four patients (4%) experienced Grade 1–2 hyperglycemia.
Although no clinically apparent conduction disorders were reported, new electrocardiographic abnormalities emerged at week 8 in 12 of 77 patients in Group A. These included T-wave abnormalities (n = 4), supraventricular premature contractions (n = 3), first-degree atrioventricular block (n = 2), and one case each of ventricular premature contraction, right bundle branch block, and left anterior fascicular block. Resolution of baseline abnormalities was observed in seven patients (data not shown).
To identify patient subgroups that may derive greater benefit from anamorelin, we performed prespecified subgroup analyses (Fig. 3). In the subgroup analyses, the type of surgical procedure, line of chemotherapy, ECOG performance status, and the use of ONS were pre-specified in the study protocol. In contrast, sex, BMI, and CRP were additionally included as exploratory factors and were not pre-specified in the protocol. Although we initially hypothesized that patients who had undergone total gastrectomy would experience a greater treatment effect, no significant differences in the efficacy of anamorelin were observed based on the type of surgical procedure. The efficacy of anamorelin did not differ according to baseline CRP levels. Notably, greater treatment effects were observed among patients receiving first-line chemotherapy (mean difference −1.08 [95% CI −2.16 to −0.001]) and those with an ECOG performance status of 0–1 (mean difference −0.90 [95% CI −1.80 to +0.001]) (Fig. 3).

Discussion
In this RCT, anamorelin did not achieve a statistically significant improvement in LBM at 8 weeks compared with control (Group N), with a mean difference of +0.85 kg (95%CI −0.05 to +1.75; P = 0.063, unadjusted t test). Although the primary endpoint was not met, the magnitude of LBM change in the anamorelin group was consistent with that observed in the ROMANA 1 and 2 trials (approximately +1.0 kg),10 suggesting a reproducible trend across studies. The lack of statistical significance may be partly attributable to the smaller sample size compared to the ROMANA studies, as well as baseline imbalances, such as a slightly higher proportion of male patients and greater baseline LBM in anamorelin group, and potential differences in concomitant anticancer therapies. Notably, a pre-specified analysis of covariance (ANCOVA), adjusting for baseline LBM and the type of surgical procedure, demonstrated a statistically significant treatment effect. These findings support a potential anabolic benefit of anamorelin after accounting for baseline variability.
Compared with prehabilitation interventions combining nutritional and resistance training -which typically report LBM gains of +0.2 to +0.9 kg over 3–4 weeks20,21-the LBM increase observed with anamorelin, though not statistically significant, may represent a clinically meaningful anabolic effect. Beyond LBM, our findings were consistent with the ROMANA studies in demonstrating early (from week 4) and sustained improvements in LBM and total weight, enhanced food intake-related QOL, and no significant differences in handgrip strength, fatigue, or most chemotherapy-related adverse events. The absence of improvement in handgrip strength, consistent with the ROMANA studies, suggests that anamorelin alone may be insufficient to improve muscle function, and combination with resistance exercise may be required. The incidence of hyperglycemia in this study was comparable to the rate reported in a large post-marketing surveillance involving over 6000 cases,22 and no severe treatment-related adverse events were observed. A unique observation in this study was a reduction in anorexia and nausea among three selected chemotherapy-related symptoms in the anamorelin group, which may have contributed to improved dietary intake and subsequent gains in LBM and total weight.
To our knowledge, this is the first multicenter RCT evaluating anamorelin in gastrointestinal malignancy-associated cachexia. Previous studies have shown that plasma ghrelin levels are markedly reduced after gastrectomy—approximately 50% after distal gastrectomy and nearly 90% after total gastrectomy.12 Based on this, we hypothesized that anamorelin, as a ghrelin receptor agonist, might be particularly effective in post-gastrectomy patients due to their endogenous ghrelin deficiency. Conversely, in patients without gastrectomy, plasma ghrelin levels tend to remain elevated in response to cancer-related weight loss,23 potentially limiting the additional effect of anamorelin. However, our findings showed no clear association between the procedures of gastrectomy and the efficacy of anamorelin. This observation is consistent with previous gastric cancer–specific observational study24 and suggests that circulating ghrelin levels may not be a major determinant of anamorelin’s clinical effectiveness. Subgroup analyses further suggested that patients receiving first-line therapy or with ECOG PS 0–1 may derive greater benefit from anamorelin, consistent with prior observational studies in gastrointestinal cancer.13 A recent retrospective observational study has reported that the therapeutic efficacy of anamorelin was limited in patients with systemic inflammation.25 In contrast, post-hoc analyses of the ROMANA 1 and 2 trials suggested that patients with systemic inflammation and those with a low BMI (<20 kg/m2) may derive greater benefit from anamorelin.26 However, in our subgroup analysis using RCT data neither BMI nor CRP levels were predictive of anamorelin efficacy. This discrepancy may be partly explained by the lower proportion of patients with elevated CRP in our cohort compared to the ROMANA trials. This difference likely reflects the distinct study populations: our trial enrolled patients with unresectable or recurrent gastric cancer undergoing first-to third-line chemotherapy, whereas ROMANA-1 and 2 focused on patients with non-small cell lung cancer receiving either chemotherapy or best supportive care. Our findings highlight the need for further evidence to accurately identify patient populations most likely to benefit from anamorelin therapy.
Emerging evidence suggests that cancer cachexia not only impairs treatment compliance but may also diminish therapeutic efficacy, as demonstrated by recent findings linking cachexia to desensitization of immunotherapy in lung cancer.27 Pharmacologic interventions targeting cachexia aim to prevent further deterioration, preserve or increase muscle mass, improve QOL, and ultimately enhance survival. However, to date, no prospective trial has demonstrated that improving cachexia leads to better treatment compliance or improved survival outcomes.
Several pharmacologic options have been explored for the management of cachexia, although the supporting evidence remains heterogeneous. In addition to anamorelin, corticosteroids28 have traditionally been employed, though their effects are limited. Recently, other promising agents have emerged, including megestrol acetate,29 and antibodies targeting growth differentiation factor 15 (GDF-15),30 a cytokine increasingly recognized as a key driver of cancer cachexia. GDF-15 has been implicated in appetite suppression and altered energy metabolism and may contribute to cachexia progression. Moreover, it appears to inhibit PFA-1α–dependent T-cell recruitment, contributing to desensitization to anti–PD-1 therapy.31 GDF-15 thus represents not only a promising biomarker but also a potential therapeutic target; anti–GDF-15 antibodies under development may offer future strategies to improve appetite and preserve muscle mass.32,33 Currently, however, anamorelin remains the only pharmacologic agent approved for human use in clinical practice. Our findings, consistent with previous RCTs, suggest that anamorelin contributes to modest gains in LBM and total weight. Although exercise intervention was not implemented in this study, its integration into multimodal interventions, including nutritional and exercise interventions,20,34 offer a promising strategy to counter cancer cachexia and improve treatment outcomes, warranting further investigation.
This study has several limitations. First, due to disease progression and missing data, the number of evaluable patients at 8 weeks (n = 160) and 12 weeks (n = 141) was lower than the prespecified target (n = 170 at 8 weeks), reducing statistical power. Second, although improvements in appetite and food intake-related QOL were observed, dietary intake (total caloric and protein intake) was not measured. Therefore, whether the observed gains in body weight and fat mass were mediated by increased intake—potentially due to improved appetite or relief of nausea—remains uncertain. Future studies should incorporate dietary assessment methods, such as food diaries or 24-h dietary recalls, to clarify the relationship between symptom improvement and nutritional intake. Third, the trial was open-label and lacked a placebo control. Given the open-label design of the study, there was a potential risk of imbalance in the provision of nutritional counseling, exercise support, psychosocial care and bias in the assessment of patient-reported outcomes, particularly those of a subjective nature such as quality of life and appetite. Fourth, long-term outcomes, including overall survival, were not assessed. However, the proportion of patients with severe malnutrition (BMI <16 kg/m2), a WHO-defined predictor of mortality, was significantly lower in Group A. A prospective cohort study to assess survival in this population is in preparation. Fifth, the heterogeneity of the study population, including patients receiving various lines of therapy, regimens, and treatment schedules, may have influenced the results for the three selected symptoms—anorexia, nausea, and vomiting. Conversely, such heterogeneity might have attenuated the ability to detect the therapeutic effects of anamorelin. Finally, while our study included patients from multiple high-volume cancer centers across Japan, thereby enhancing external validity, caution should be exercised when generalising these findings to non-Asian populations or community-based settings. Differences in ethnicity, clinical practice patterns, and access to supportive care may limit the broader applicability of the results. Nevertheless, the consistency of our findings with those from the ROMANA trials conducted in Western populations supports the potential generalisability of anamorelin’s effects across diverse patient groups.
In conclusion, this study represents the first randomized controlled trial to evaluate the role of anamorelin in patients with unresectable advanced or recurrent gastric cancer complicated by cancer cachexia. Anamorelin was safely administered alongside ongoing chemotherapy, resulting in a significant increase in body weight, although no statistically significant gain in LBM was observed. These findings highlight the potential role of anamorelin as part of a multimodal therapeutic strategy for cancer cachexia in gastric cancer and provide a foundation for future research to optimize its clinical application.

Contributors
KY, YK, YM, and YD were responsible for study design and concept of this study. KY and TY directly accessed and verified the underlying data reported in the manuscript. KY, YK, HE and YD contributed to data interpretation and critical revision of the manuscript for important intellectual content. KY conducted data collection and wrote the initial draft of the manuscript. TO, YY, NS, AT, RK, YA, KS, JM, YK, KY, TS, and TT contributed to data collection, interpretation, and review of the manuscript. TY was responsible for statistical analysis in this study. All authors read and approved the final version of the manuscript and have agreed to the accountability of all aspects of the study.

Data sharing statement
A de-identified, fixed dataset used for this study is securely stored and will not be publicly available, as further analysis for secondary studies are currently ongoing. However, data may be made available for individual participant data meta-analyses or guideline development upon reasonable request. Access will be granted at the discretion of the corresponding author after review and approval of a methodologically sound proposal, and upon signing a data access agreement.

Declaration of interests
All authors declare no competing interests.