跳到正文
原文
Nature 健康护理主题·· 2026-10-12AI 评分50

GLP-1受体激动剂与2型糖尿病患者血液肿瘤风险的多中心倾向评分匹配队列分析

Glucagon-like peptide-1 receptor agonists and risk for hematologic malignancies: a multicenter propensity score-matched analysis

AI 导读

一项基于TriNetX电子健康记录的回顾性倾向评分匹配队列研究显示,2型糖尿病患者使用GLP-1受体激动剂后,与DPP-4抑制剂相比所评估的血液肿瘤发生率均无显著差异,与SGLT2抑制剂相比则观察到非霍奇金淋巴瘤(HR 0.82)、骨髓增殖性肿瘤(HR 0.78)和霍奇金淋巴瘤(HR 0.52)发生率较低。

正文

To the Editor:

Glucagon-like peptide-1 receptor agonists (GLP-1RAs) have become central to the management of type 2 diabetes mellitus (T2DM) and obesity, with indications now extending to cardiovascular risk reduction, renal protection, and liver disease [1]. As utilization of these agents continues to grow, thorough characterization of their long-term safety profile, including potential effects on malignancy risk, remains a priority [2]. Preclinical evidence indicates that GLP-1RAs exert anti-inflammatory and immunomodulatory effects, including suppression of NF-κB signaling and modulation of immune cell function, which could theoretically influence hematopoiesis and lymphomagenesis [2, 3]. Nevertheless, clinical data evaluating the relationship between GLP-1RA use and hematologic malignancy risk are sparse.

We performed a retrospective propensity score-matched cohort study using the TriNetX federated research network, a global platform comprising over 100 million de-identified electronic health records. Qualifying GLP-1RAs included semaglutide, dulaglutide, liraglutide, and exenatide. Adults ( ≥ 18 years) with type 2 diabetes mellitus who initiated GLP-1RAs between January 1, 2015, and March 31, 2021, were identified. GLP-1RA users were compared with users of two clinically relevant active comparators, dipeptidyl peptidase-4 (DPP-4) inhibitors and sodium-glucose cotransporter 2 (SGLT2) inhibitors (Supplementary Table 1), agents commonly prescribed as second-line therapies for T2DM during the study period [4, 5]. Eligible patients were required to have a diagnosis of type 2 diabetes mellitus within one year before the index prescription, followed by both a subsequent prescription for the same medication class and a documented type 2 diabetes diagnosis between one month and one year after the index date.

Two separate analyses were conducted: (1) GLP-1RA users without prior DPP-4 inhibitor exposure versus DPP-4 inhibitor users without prior GLP-1RA exposure, and (2) GLP-1RA users without prior SGLT2 inhibitor exposure versus SGLT2 inhibitor users without prior GLP-1RA exposure. For each outcome-specific analysis, patients with a documented history of the corresponding hematologic malignancy before the beginning of the outcome-assessment window were excluded. The index date was defined as the date of first prescription for the respective drug class. Propensity-score matching (1:1) was performed using demographics, comorbidities, concomitant medications, laboratory values, and additional hematologic malignancy-related risk factors, including HIV infection, tobacco exposure, prior irradiation, H. pylori infection, and selected autoimmune and inflammatory disorders. Outcomes of interest included incident non-Hodgkin lymphoma (NHL), Hodgkin lymphoma, myeloid leukemia, lymphoid leukemia, plasma cell neoplasms, myelodysplastic syndromes (MDS), and myeloproliferative neoplasms (MPN). Time-to-event analyses were performed using Kaplan–Meier methods and Cox proportional hazards models, with outcomes assessed from 180 days through 5 years after the index date (Supplementary Table 2). To assess whether the matched cohorts reproduced known class-related safety profiles, we also evaluated gastroesophageal reflux disease (GERD), acute pancreatitis, diabetic ketoacidosis, diarrhea, and constipation [6].

Following matching, 68,949 pairs were included in the GLP-1RA versus DPP-4 inhibitor analysis and 71,391 pairs in the GLP-1RA versus SGLT2 inhibitor analysis, with standardized differences below 0.10 for the matched covariates (Supplementary Tables 3, 4). In the DPP-4 inhibitor comparison, mean follow-up was 1580 days (SD 461) in the GLP-1RA cohort and 1534 days (SD 514) in the DPP-4 inhibitor cohort, with a median of 1825 days in both cohorts (IQR 240 and 370 days, respectively). In the SGLT2 inhibitor comparison, mean follow-up was 1592 days (SD 452) in the GLP-1RA cohort and 1579 days (SD 468) in the SGLT2 inhibitor cohort, with a median of 1825 days in both cohorts (IQR 204 and 209 days, respectively).

In the GLP-1RA versus DPP-4 inhibitor comparison, no significant difference was observed for NHL (HR 0.99; 95% CI 0.82–1.20; p = 0.93; 208 vs. 203 events) (Fig. 1). No significant differences were observed for MDS (HR 0.84; 95% CI 0.65–1.09; p = 0.19; 105 vs. 121 events), plasma cell neoplasms (HR 0.92; 95% CI 0.73–1.15; p = 0.46; 140 vs. 148 events), myeloid leukemia (HR 1.16; 95% CI 0.86–1.55; p = 0.33; 98 vs. 82 events), lymphoid leukemia (HR 1.07; 95% CI 0.81–1.41; p = 0.64; 104 vs. 94 events), Hodgkin lymphoma (HR 0.97; 95% CI 0.60–1.56; p = 0.89; 34 vs. 34 events), or MPN (HR 1.01; 95% CI 0.86–1.18; p = 0.95; 298 vs. 288 events). Cumulative incidence curves for each hematologic malignancy outcome in the GLP-1RA versus DPP-4 inhibitor comparison are shown in Supplemental Fig. 1.

Fig. 1

Risk of hematologic malignancies among GLP-1RA users versus DPP-4 inhibitor users.

In the GLP-1RA versus SGLT2 inhibitor comparison, GLP-1RA use was associated with lower incidences of NHL (HR 0.82; 95% CI 0.68–0.99; p = 0.04; 200 vs. 241 events), MPN (HR 0.78; 95% CI 0.67–0.91; p = 0.002; 300 vs. 379 events), and Hodgkin lymphoma (HR 0.52; 95% CI 0.30–0.91; p = 0.02; 19 vs. 36 events) (Fig. 2). No significant differences were found for myeloid leukemia (HR 0.83; 95% CI 0.63–1.11; p = 0.22; 85 vs. 101 events), plasma cell neoplasms (HR 0.90; 95% CI 0.70–1.15; p = 0.41; 120 vs. 132 events), lymphoid leukemia (HR 0.90; 95% CI 0.68–1.18; p = 0.45; 97 vs. 107 events), or MDS (HR 1.20; 95% CI 0.89–1.60; p = 0.23; 100 vs. 83 events). Cumulative incidence curves for each hematologic malignancy outcome in the GLP-1RA versus SGLT2 inhibitor comparison are shown in Supplemental Fig. 2.

Fig. 2

Risk of hematologic malignancies among GLP-1RA users versus SGLT2 inhibitor users.

In exploratory analyses of selected NHL subtypes, no statistically significant differences were observed for follicular lymphoma, non-follicular lymphoma, mature T/NK-cell lymphoma, or other specified or unspecified NHL for both comparisons (Supplementary Fig. 3, 4).

In the adverse event analyses, there was a higher incidence of GERD in patients who received GLP-1RA compared with DPP-4 inhibitors (HR 1.04; 95% CI 1.00–1.07; p = 0.03). Meanwhile, the risk of acute pancreatitis, diabetic ketoacidosis, diarrhea, and constipation did not differ significantly (Supplementary Fig. 5). Compared with SGLT2 inhibitors, GLP-1RA use was associated with higher incidences of GERD (HR 1.14; 95% CI 1.10–1.18; p < 0.001), diarrhea (HR 1.14; 95% CI 1.10–1.18; p < 0.001), and constipation (HR 1.10; 95% CI 1.07–1.14; p < 0.001), no significant difference in acute pancreatitis (HR 0.98; 95% CI 0.89–1.08; p = 0.72), and a lower incidence of diabetic ketoacidosis (HR 0.83; 95% CI 0.76–0.91; p < 0.001) (Supplementary Fig. 6).

The principal finding of this study is that the observed associations between GLP-1RA use and hematologic malignancy incidence differed according to the active comparator selected: lower incidences of NHL, MPN, and Hodgkin lymphoma were observed relative to SGLT2 inhibitors but not DPP-4 inhibitors. However, these findings were not reproduced across comparators and should be interpreted cautiously in the context of multiple outcomes, relatively low event counts, and the absence of adjustment for multiple comparisons. Analyses of follicular lymphoma, non-follicular lymphoma, mature T/NK-cell lymphoma, and other specified or unspecified NHL did not demonstrate significant differences in either comparator analysis.

This observation carries clinical relevance given that NHL is the most prevalent hematologic malignancy and that both obesity and T2DM are established risk factors for its development [7,8,9]. If confirmed, potential mechanisms that could contribute to the lower observed NHL incidence include GLP-1RA-mediated attenuation of chronic inflammation, enhancement of immune surveillance, and direct effects on lymphocyte proliferation [2, 3].

These findings contribute to a growing body of evidence examining GLP-1RAs and cancer risk. In a large observational study using the TriNetX database in patients with obesity, Levy et al. reported significant reductions in lymphoid and hematopoietic cancers among GLP-1RA users (HR 0.69; 95% CI 0.60–0.80) compared with matched controls [10]. Ashruf et al. found that GLP-1RA use was associated with markedly lower risk of several hematologic malignancies compared with insulin, including NHL (HR 0.42; 95% CI 0.30–0.58), MDS (HR 0.19; 95% CI 0.11–0.35), and MPN (HR 0.50; 95% CI 0.41–0.61) [11]; however, no significant differences emerged when GLP-1RAs were compared with metformin. A recent Japanese claims database study by Kawashiri et al. similarly demonstrated that SGLT2 inhibitor use was associated with reduced risk of hematologic malignancies, including lymphoma (HR 0.71; 95% CI 0.57–0.89), relative to non-SGLT2 inhibitor antidiabetic agents [12]. Importantly, unlike previous studies comparing GLP-1RAs with insulin, metformin, or nonusers, the present analysis demonstrates that the observed NHL association is highly dependent on comparator selection. Meta-analyses of randomized controlled trials have generally identified no increased cancer risk with GLP-1RAs, although these trials were not designed to evaluate cancer outcomes and had limited follow-up duration [13].

In this study, DPP-4 inhibitors and SGLT2 inhibitors were selected as active comparators because they are commonly considered alongside GLP-1RAs as non-insulin second-line therapies for type 2 diabetes [5, 6]. Previous large observational studies compared GLP-1RAs with insulin, metformin, or matched nonusers; however, metformin is typically first-line therapy and frequently continued throughout the disease course, while insulin is more often initiated in patients with longer-standing or more advanced diabetes and greater comorbidity burden. Comparisons with these agents may therefore be susceptible to confounding related to treatment stage and disease severity. DPP-4 inhibitors and SGLT2 inhibitors are not free from similar concerns, as DPP-4 inhibitors are often preferred in older or frailer patients and when hypoglycemia avoidance or medication cost are priorities, while SGLT2 inhibitors are frequently favored in patients with chronic kidney disease, heart failure, or established cardiovascular disease. These class-specific prescribing patterns may introduce residual confounding that propensity score matching cannot fully address.

Across both comparator analyses, GLP-1RA use was not associated with a consistent increase in NHL, Hodgkin lymphoma, plasma cell neoplasms, myeloid leukemia, lymphoid leukemia, MDS, or MPN. The lower incidences of NHL, MPN, and Hodgkin lymphoma observed relative to SGLT2 inhibitors were not replicated in the DPP-4 inhibitor comparison. These findings do not suggest an increased risk of the hematologic malignancies evaluated but do not establish a protective effect. DPP-4 inhibitors and GLP-1RAs both act through incretin signaling but differ in magnitude and duration of GLP-1 receptor activation, effects on body weight, and broader metabolic consequences. The observation that the lower incidences emerged only in the SGLT2 inhibitor comparison and not in the DPP-4 inhibitor comparison indicates these associations are sensitive to comparator selection and may reflect differences in the populations receiving each drug class rather than a direct effect of GLP-1RA use.

Our analysis did not identify a statistically significant difference in incidence of plasma cell neoplasms between GLP-1RA users and either active comparator, a finding that contrasts with recent reports suggesting potential associations between GLP-1RA use and myeloma-related outcomes. Grandhi et al. found that GLP-1RA use was associated with reduced MGUS progression to multiple myeloma among veterans with diabetes (HR 0.45, 95% CI 0.22–0.93) [14], and Abidi et al. reported similar findings in a large real-world cohort [15]. Importantly, these studies evaluated MGUS progression rather than incident disease. Irons et al., also using the TriNetX database, reported a lower incidence of multiple myeloma among GLP-1RA users compared with those prescribed SGLT2 inhibitors or neither drug class (HR 0.64, p = 0.01) [16]. However, they enrolled patients from 2019 to 2024 without reporting follow-up duration; for malignancies with prolonged latency, this limited observation period may be insufficient, whereas our study achieved a mean follow-up of approximately 4.3 years. Further investigation with larger sample sizes and analyses of myeloma precursor states is warranted.

Several limitations warrant consideration. These include potential residual confounding despite propensity score matching, reliance on ICD-10 codes for outcome ascertainment without histologic confirmation, inability to evaluate dose-response relationships, and the possibility of detection bias. Notably, GLP-1RAs are associated with clinically significant weight loss, substantially greater than that observed with DPP-4 inhibitors or SGLT2 inhibitors. Because obesity is an independent risk factor for NHL, the observed lower incidence of NHL may be partly or wholly mediated by weight reduction rather than direct immunomodulatory effects of GLP-1 receptor activation. The current study design cannot distinguish between these mechanisms. Furthermore, the relatively low event rates for individual hematologic malignancy subtypes limit statistical power for subtype-specific analyses. In our study, we required a repeat prescription within one year of the index date to strengthen exposure ascertainment; however, TriNetX does not capture dispensing, administration, or adherence data, so misclassification of drug exposure cannot be excluded. BMI and HbA1c values were available for approximately 65–72% and 80–82% of matched patients, respectively, and no multiple imputation was performed; residual confounding from incomplete anthropometric and glycemic data therefore cannot be excluded. In addition, GLP-1RAs, DPP-4 inhibitors, and SGLT2 inhibitors entered clinical practice at different times and were adopted under evolving treatment guidelines, which may have contributed to residual confounding despite propensityscore matching.

In this large propensity score-matched cohort study of adults with type 2 diabetes mellitus, GLP-1RA use was not associated with a consistent increase in the incidence of the hematologic malignancies evaluated compared with DPP-4 or SGLT2 inhibitors. Lower incidences of NHL, MPN, and Hodgkin lymphoma were observed relative to SGLT2 inhibitors but not DPP-4 inhibitors. The comparator-dependent nature of these findings and the potential for residual confounding preclude causal interpretation. These findings do not identify a consistent increase in risk of the hematologic malignancies evaluated but require confirmation in prospective studies with detailed longitudinal medication exposure data.

References

  1. Drucker DJ. GLP-1-based therapies for diabetes, obesity and beyond. Nat Rev Drug Discov. 2025;24:631–50. https://doi.org/10.1038/s41573-025-01183-8.

    Article  PubMed  Google Scholar 

  2. Valencia-Rincón E, Rai R, Chandra V, Wellberg EA. GLP-1 receptor agonists and cancer: current clinical evidence and translational opportunities for preclinical research. J Clin Invest. 2025;135:e194743. https://doi.org/10.1172/jci194743.

    Article  PubMed  PubMed Central  Google Scholar 

  3. Wawrzak-Pienkowska K, Pienkowski T, Golonko A, Krupa A, Warpechowski M, Plonski AF, et al. Incretin signaling at the crossroads of metabolism, inflammation, and tumorigenesis: implications for obesity patients. Eur J Pharm. 2025;1007:178290. https://doi.org/10.1016/j.ejphar.2025.178290.

    Article  Google Scholar 

  4. Davies MJ, D’Alessio DA, Fradkin J, Kernan WN, Mathieu C, Mingrone G, et al. Management of Hyperglycemia in Type 2 Diabetes, 2018. A Consensus Report by the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diab Care. 2018;41:2669–701. https://doi.org/10.2337/dci18-0033.

    Article  Google Scholar 

  5. Khera R, Aminorroaya A, Dhingra LS, Thangaraj PM, Pedroso Camargos A, Bu F, et al. Comparative effectiveness of second-line antihyperglycemic agents for cardiovascular outcomes: a multinational, federated analysis of LEGEND-T2DM. J Am Coll Cardiol. 2024;84:904–17. https://doi.org/10.1016/j.jacc.2024.05.069.

    Article  PubMed  PubMed Central  Google Scholar 

  6. Chiang CH, Jaroenlapnopparat A, Colak SC, Yu CC, Xanthavanij N, Wang TH, et al. Glucagon-Like Peptide-1 receptor agonists and gastrointestinal adverse events: a systematic review and meta-analysis. Gastroenterology. 2025;169:1268–81. https://doi.org/10.1053/j.gastro.2025.06.003.

    Article  PubMed  Google Scholar 

  7. Zhang N, Wu J, Wang Q, Liang Y, Li X, Chen G, et al. Global burden of hematologic malignancies and evolution patterns over the past 30 years. Blood Cancer J. 2023;13:82. https://doi.org/10.1038/s41408-023-00853-3.

    Article  PubMed  PubMed Central  Google Scholar 

  8. Larsson SC, Wolk A. Obesity and risk of non-Hodgkin’s lymphoma: a meta-analysis. Int J Cancer. 2007;121:1564–70. https://doi.org/10.1002/ijc.22762.

    Article  PubMed  Google Scholar 

  9. Castillo JJ, Mull N, Reagan JL, Nemr S, Mitri J. Increased incidence of non-Hodgkin lymphoma, leukemia, and myeloma in patients with diabetes mellitus type 2: a meta-analysis of observational studies. Blood. 2012;119:4845–50. https://doi.org/10.1182/blood-2011-06-362830.

    Article  PubMed  PubMed Central  Google Scholar 

  10. Levy S, Attia A, Elshazli RM, Abdelmaksoud A, Tatum D, Aiash H, et al. Differential effects of GLP-1 receptor agonists on cancer risk in obesity: a nationwide analysis of 1.1 million patients. Cancers. 2024;17:78 https://doi.org/10.3390/cancers17010078.

    Article  PubMed  PubMed Central  Google Scholar 

  11. Ashruf OS, Hundal J, Mushtaq A, Kaelber DC, Anwer F, Singh A. Hematologic cancers among patients with Type 2 diabetes prescribed GLP-1 receptor agonists. JAMA Netw Open. 2025;8:e250802. https://doi.org/10.1001/jamanetworkopen.2025.0802.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Kawashiri T, Yamamoto N, Fujiwara M, Shimizu T, Kobayashi D, Fujita S, et al. Impact of sodium-glucose cotransporter 2 (SGLT2) inhibitors on hematologic malignancy risk: Insights from Japanese claims data. J Natl Cancer Inst. 2026;https://doi.org/10.1093/jnci/djag038.

  13. Ko A, Chang YC, Bahar F, Wang TH, Xanthavanij N, Yu CC, et al. Risk for cancer with glucagon-like Peptide-1 receptor agonists and dual agonists: a systematic review and meta-analysis. Ann Intern Med. 2025;https://doi.org/10.7326/annals-25-02237.

  14. Grandhi N, Liu L, Wang M, Thomas T, Schoen M, Sanfilippo K, et al. Association between glucagon-like peptide-1 receptor agonist use and progression of monoclonal gammopathy of uncertain significance to multiple myeloma among patients with diabetes. JNCI Cancer Spectr. 2024;8:pkae095. https://doi.org/10.1093/jncics/pkae095.

    Article  PubMed  PubMed Central  Google Scholar 

  15. Abidi A, Mann H, Purvey S, Kolipakkam B, Cassel B, Yazbeck V. GLP-1 receptor agonist use is associated with lower risk of progression from MGUS to SMM or MM in a large real-world cohort. Blood. 2025;146:404. https://doi.org/10.1182/blood-2025-404.

    Article  Google Scholar 

  16. Irons EE, Pfeil KA, Perez JA, van Besien K. Incidence of hematologic malignancies and mortality associated with GLP-1 receptor agonist and SGLT2 inhibitor use in type 2 diabetes mellitus: results of a retrospective cohort study of electronic health records. EClinicalMedicine. 2026;91:103749. https://doi.org/10.1016/j.eclinm.2025.103749.

    Article  PubMed  PubMed Central  Google Scholar 

Download references

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Author information

Author notes

  1. These authors contributed equally: Pratik Shah, Adit Singhal.

Authors and Affiliations

  1. Northwell Health Cancer Institute, Lake Success, NY, USA

    Pratik Shah, Adit Singhal & Cho-Han Chiang

  2. Department of Medicine, University at Buffalo-Catholic Health System, Buffalo, NY, USA

    Tsu Hsien Wang

  3. Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA

    Albert Ko

  4. Department of Medicine, Danbury Hospital, Danbury, CT, USA

    Yu-Cheng Chang

  5. Feinstein Institutes for Medical Research, Manhasset, NY, USA

    Anita Ng & Andrew Shih

  6. Department of Family Medicine, Taipei Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation, New Taipei City, Taiwan

    Yuan Ping Hsia

  7. Department of Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan

    Cho-Hung Chiang

Authors

  1. Pratik Shah
  2. Adit Singhal
  3. Tsu Hsien Wang
  4. Albert Ko
  5. Yu-Cheng Chang
  6. Anita Ng
  7. Andrew Shih
  8. Yuan Ping Hsia
  9. Cho-Hung Chiang
  10. Cho-Han Chiang

Contributions

Study design: P Shah, A Singhal, YP Hsia, CHa Chiang. Acquisition of data: P Shah, A Singhal, YC Chang, A Ko, TH Wang. Data analysis: P Shah, A Singhal, YC Chang, TH Wang, A Ng, A Shih. Interpretation of data: P Shah, A Singhal, YC Chang, A Ko, TH Wang, A Ng, A Shih. Manuscript drafting and writing: P Shah, A Singhal, CHa Chiang. Critical revision of the manuscript for important intellectual content: P Shah, A Singhal, YC Chang, A Ko, TH Wang, A Ng, A Shih, YP Hsia, CHu Chiang, CHa Chiang. Final approval of the manuscript: All authors.

Corresponding authors

Correspondence to Yuan Ping Hsia or Cho-Han Chiang.

Ethics declarations

Competing interests

The authors declare no competing interests.

Ethical approval statement

Not applicable.

Artificial intelligence statement

Artificial intelligence was used solely to assist with language editing, including grammar, sentence structure, clarity, and overall manuscript flow. The AI tool was not used to generate scientific content, perform data analysis, influence study design, or contribute to data interpretation or study conclusions. All content was reviewed and approved by the authors, who take full responsibility for the accuracy and integrity of the manuscript. Editing assistance was provided using Claude, based on a Claude 4 family large language model (Anthropic, San Francisco, CA, USA), accessed via the Claude web interface.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Prior publication/presentation: There have been no prior publications or presentations of this study.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Shah, P., Singhal, A., Wang, T.H. et al. Glucagon-like peptide-1 receptor agonists and risk for hematologic malignancies: a multicenter propensity score-matched analysis. Blood Cancer J. 16, 178 (2026). https://doi.org/10.1038/s41408-026-01618-4

Download citation

  • Received: 21 April 2026

  • Revised: 14 August 2026

  • Accepted: 27 August 2026

  • Published: 12 October 2026

  • Version of record: 12 October 2026

  • DOI: https://doi.org/10.1038/s41408-026-01618-4

来源:Nature 健康护理主题 · nature.com