Electric Acupoint Stimulation and Self-Acupressure Both Steer Breast Cancer Patients Toward Milder Chemotherapy Side-Effect Trajectories

Chemotherapy for breast cancer rarely produces a single side effect in isolation. Nausea and vomiting, anxiety, depression, and disturbed sleep tend to arrive together, feed into one another, and evolve over the course of treatment in ways that standard single-symptom trials fail to capture. A new secondary analysis of a randomized controlled trial, published in Supportive Care in Cancer, tackles this complexity head-on by combining a trajectory-modeling technique borrowed from criminology and developmental psychology with a machine-learning method for estimating individualized treatment effects. The result is one of the clearest pictures yet of how two non-drug interventions—transcutaneous electrical acupoint stimulation (TEAS) and self-acupressure (SA)—reshape the entire course of chemotherapy-related symptom burden rather than just one symptom at a time.

The research team, led by investigators at Chengdu University of Traditional Chinese Medicine and collaborating institutions across Sichuan, China, analyzed data from 189 breast cancer patients receiving chemotherapy. Rather than asking whether an intervention lowered a symptom score at a single time point, the researchers applied the Group-Based Multi-Trajectory Model (GBMTM), a statistical framework originally developed by Nagin and colleagues that identifies latent subgroups of individuals who share similar developmental paths across multiple outcomes simultaneously. In this case, the outcomes were four distinct but interrelated dimensions of chemotherapy toxicity: acute and delayed chemotherapy-induced nausea and vomiting (CINV), anxiety, depression, and sleep quality, each measured repeatedly across the chemotherapy cycle.

The GBMTM analysis revealed that the 189 patients did not respond to chemotherapy as a homogeneous population. Instead, the model sorted them into two sharply different trajectory classes. The first, labeled High Symptom Burden-Persistent (HSBP), comprised 101 patients whose nausea, psychological distress, and sleep disruption remained elevated across treatment. The second, labeled Low Symptom Burden-Relieving (LSBR), included 88 patients whose anxiety, depression, and sleep problems improved substantially over time. Strikingly, even within this more favorable group, the incidence of acute CINV remained persistently high, contrasting with the comprehensive relief seen in the other three symptom domains. This dissociation suggests that acute nausea may be governed by physiological mechanisms—serotonin, substance P, and dopamine signaling in the emetic pathways—that are less responsive to the psychological and behavioral improvements that accompany the other symptoms.

With trajectory classes established, the team turned to causal forest analysis, a machine-learning approach rooted in the work of Wager and Athey that estimates heterogeneous treatment effects across individuals rather than assuming a single average effect applies to everyone. Causal forests build ensembles of decision trees that partition patients based on baseline characteristics, allowing the algorithm to estimate how the probability of belonging to the favorable LSBR trajectory changes when a patient receives TEAS or SA compared with a control condition. The average treatment effect (ATE) for TEAS was 0.147 (95% CI: 0.143, 0.151), meaning that receiving TEAS increased the probability of landing in the LSBR class by roughly 14.7 percentage points. SA performed slightly better, with an ATE of 0.176 (95% CI: 0.162, 0.190), a difference that reached statistical significance at P < 0.05.

That self-acupressure edged out transcutaneous electrical acupoint stimulation is noteworthy given the practical implications. TEAS involves placing electrodes at classical acupuncture points—typically P6 (Neiguan) on the forearm, ST36 (Zusanli) on the leg, and other meridian points standardized in the Chinese national nomenclature GB/T 12346-2021—and delivering low-frequency electrical pulses through the skin. It requires equipment, trained personnel, and scheduled sessions. Self-acupressure, by contrast, requires only that patients learn to apply manual pressure to the same or similar points themselves, making it essentially free, endlessly portable, and scalable to any oncology ward or home setting. A systematic review by Cheng and colleagues published in the Journal of Pain and Symptom Management in 2023 had already suggested that self-acupressure holds promise for symptom management in cancer patients; the new causal forest analysis adds quantitative trajectory-level evidence that a simple self-administered technique can shift a patient’s entire symptom course toward a more favorable pattern.

The mechanistic rationale for both interventions rests on the neurophysiology of acupuncture points. Stimulation of P6, located approximately three finger-widths proximal to the wrist crease between the tendons of flexor carpi radialis and palmaris longus, is thought to modulate vagal afferent signaling and influence the chemoreceptor trigger zone and vomiting center in the brainstem. Meta-analyses of TEAS for postoperative nausea and vomiting, including a 2020 analysis in the International Journal of Surgery and a 2023 systematic review in PLoS ONE, have consistently shown reductions in emetic episodes comparable to standard antiemetic prophylaxis in surgical settings. A 2023 non-inferiority randomized controlled trial in Surgery found TEAS comparable to dexamethasone for preventing postoperative nausea and vomiting in breast surgery. Extending this evidence base to the chemotherapy setting—and, crucially, to the joint trajectory of nausea alongside anxiety, depression, and sleep—represents a meaningful conceptual advance.

The study’s methodological architecture deserves attention from anyone following the intersection of machine learning and clinical research. Group-based trajectory modeling has been used extensively in developmental psychopathology and criminology to identify latent classes of behavioral trajectories, but its application to multidimensional cancer symptom data is relatively recent. A 2023 study by Grayson and colleagues in the same journal examined psychoneurological symptom cluster trajectories in early-stage breast cancer, and a 2022 study by Li and colleagues in Frontiers in Public Health identified trajectory classes of anxiety and depression among Chinese breast cancer patients. The current study goes further by modeling four symptom dimensions simultaneously and then layering causal inference on top, using the honest causal forest approach described by Jawadekar and colleagues in the American Journal of Epidemiology in 2023 and by Sverdrup and colleagues in a 2025 tutorial in the International Journal of Methods in Psychiatric Research. This combination—trajectory identification followed by treatment-effect estimation—offers a template for future longitudinal oncology research that moves beyond simple mean comparisons.

The clinical stakes are substantial. Breast cancer is the most commonly diagnosed cancer worldwide, with GLOBOCAN 2022 estimates placing global incidence at over 2.3 million new cases annually. Chemotherapy remains a mainstay of treatment, but its toxicity profile—particularly the interplay between physical symptoms like CINV and psychological symptoms like anxiety and depression—directly affects treatment adherence, quality of life, and even long-term outcomes. A 2020 meta-analysis in Molecular Psychiatry covering 282,203 patients demonstrated that depression and anxiety are associated with increased breast cancer recurrence and mortality, making effective management of psychological symptoms not merely a comfort measure but a potential prognostic intervention. Sleep disturbances, meanwhile, have been linked to elevated stroke risk in breast cancer patients with sleep disorders, according to a 2021 study in the Journal of Cancer. The finding that a single non-pharmacological intervention can shift patients across multiple symptom domains simultaneously is therefore clinically meaningful in a way that single-symptom trials cannot fully convey.

Several caveats temper the enthusiasm. This was a secondary analysis of a randomized controlled trial registered with the Chinese Clinical Trial Registry (ChiCTR2300077667, registered 15 November 2023), and secondary analyses, however sophisticated, inherit the constraints of the original trial design. The sample of 189 patients, while adequate for trajectory modeling, limits the precision with which causal forests can estimate heterogeneous effects across subgroups defined by age, chemotherapy regimen, or baseline symptom severity. The ATE estimates, though statistically significant, are modest in absolute terms—a 14.7 to 17.6 percentage-point increase in the probability of favorable trajectory classification. And the finding that acute CINV remained persistently high even in the LSBR group suggests that neither intervention fully addresses the physiological drivers of acute emesis, which may require optimization of standard antiemetic prophylaxis in parallel with non-drug approaches.

Nevertheless, the study’s central contribution—demonstrating that TEAS and SA both increase the likelihood that a breast cancer patient will follow a low-burden, self-resolving symptom trajectory across nausea, anxiety, depression, and sleep simultaneously—offers oncology teams a practical, evidence-based option that complements pharmacological antiemetic guidelines from MASCC, ESMO, NCCN, and ASCO. As machine-learning methods for individualized treatment effect estimation continue to mature, and as trajectory modeling becomes more accessible through open-source statistical software, the analytical framework pioneered here may well become standard practice in supportive oncology research. For now, the message for patients and clinicians is concrete: a simple self-administered acupressure routine, taught once and practiced at home, may be as effective as a device-based electrical stimulation protocol in steering the entire course of chemotherapy toxicity toward recovery.

Subject of Research: Effects of transcutaneous electrical acupoint stimulation and self-acupressure on multidimensional chemotherapy side-effect trajectories in breast cancer patients

Article Title: Effects of transcutaneous electrical acupoint stimulation versus acupressure on the trajectories of multidimensional adverse reactions to chemotherapy in breast cancer patients: a secondary analysis of a randomized controlled trial

Article References: Yuan, J., Zhao, X., Li, H., Li, B., Tian, C., Chu, J., Kayishaer, T., Tao, Z., Jiang, H., Hu, X., Zhu, B., & Yang, Y. (2026). Effects of transcutaneous electrical acupoint stimulation versus acupressure on the trajectories of multidimensional adverse reactions to chemotherapy in breast cancer patients: a secondary analysis of a randomized controlled trial. Supportive Care in Cancer, 34(9), Article 917. https://doi.org/10.1007/s00520-026-11168-x

Image Credits: AI Generated

DOI: 10.1007/s00520-026-11168-x

Keywords: breast cancer, chemotherapy, transcutaneous electrical acupoint stimulation, self-acupressure, chemotherapy-induced nausea and vomiting, anxiety, depression, sleep quality, group-based multi-trajectory model, causal forest, symptom trajectories, supportive care