Investigating the symptoms and sentinel symptoms of patients with lymphoma during chemotherapy: a cross-sectional survey Authors: Yang Gao, Lihua Wu, Jiani Ji, Xingyu Chen, Jingjing Lu, Moli Cao Published: 2025 ABSTRACT AbstractObjectivesTo investigate symptom clusters (SCs) and sentinel symptoms in patients with lymphoma during chemotherapy treatment and examine the associations between sentinel symptoms and laboratory parameters.DesignThis was a single-centre cross-sectional study.SettingData collection was conducted through the General Information Questionnaire and the Chinese version of the Memorial Symptom Assessment Scale. Exploratory factor analysis was used to extract SCs, and association rule mining employing the Apriori algorithm was implemented to identify sentinel symptoms. Correlation analysis was used to explore the correlation between sentinel symptoms and laboratory indicators.ParticipantsA total of 268 valid questionnaires were collected from patients.Outcome measuresExplore SCs and sentinel symptoms in patients with lymphoma during chemotherapy and analyse the correlation between sentinel symptoms and laboratory indicators.ResultsFour distinct SCs were identified among the participants. Drowsiness, nausea, sweating and skin changes were determined as sentinel symptoms of emotional–fatigue–sleep, nutritional metabolism, lymphoma-specific and chemotherapy–body SCs, respectively. Among these symptoms, drowsiness demonstrated positive associations with white blood cell count and glutamic pyruvic transaminase levels, while exhibiting negative associations with haemoglobin, platelet and albumin levels. Sweating exhibited negative associations with platelet count. Skin changes demonstrated negative associations with D-dimer levels. There were no significant associations detected between nausea and laboratory parameters.ConclusionsPatients receiving lymphoma chemotherapy exhibited various symptoms manifesting as four distinct SCs. The identified sentinel symptoms comprised drowsiness, nausea, sweating and skin changes. With the exception of nausea, all variables demonstrated significant associations with multiple laboratory parameters.Trial registration numberChiCTR2500096487. STRENGTHS AND LIMITATIONS OF THIS STUDY The Apriori algorithm model was applied to identify sentinel symptoms within the established symptom clusters. The Apriori algorithm model was applied to identify sentinel symptoms within the established symptom clusters. A sentinel symptom was defined as an indicator within a cluster that may predict the onset of other symptoms in that cluster. A sentinel symptom was defined as an indicator within a cluster that may predict the onset of other symptoms in that cluster. Cross-sectional designs cannot establish causal relationships between symptoms. Cross-sectional designs cannot establish causal relationships between symptoms. The use of self-reported data introduces the potential for recall and reporting biases. The use of self-reported data introduces the potential for recall and reporting biases. Introduction Lymphomas constitute a collection of remarkably heterogeneous malignancies derived from the lymphoid haematopoietic system, with 635 400 new cases and 291 300 deaths documented globally by 2022, demonstrating escalating patterns in both incidence and mortality rates.1 Research developments focused on individual patient characteristics and personalised therapeutic approaches, encompassing chemotherapy, radiotherapy, targeted therapy and haematopoietic stem cell transplantation, have substantially enhanced patient survival outcomes. Chemotherapy continues to be regarded as the principal method for achieving remission or cure in patients with lymphoma.24 Although chemotherapeutic agents demonstrate efficacy in cancer treatment, patients with lymphoma frequently encounter a spectrum of chemotherapy-associated complications attributed to their toxic adverse effects.5 Approximately 31.6–72.6% of patients experience ‘clustered’ manifestations, including fatigue, sleep disturbances, anxiety, alopecia, oral pain and skin alterations,6 with certain specific symptoms also being observed, such as localised swelling, fever, night sweats and pruritus.7 The manifestations within the symptom clusters (SCs) are not present individually or merely ‘additive’; instead, they interact through a ‘multiplicative’ mechanism, mutually intensifying one another. This phenomenon imposes a substantial burden on patients both physiologically and psychologically and may potentially contribute to disability and irreversible sequelae. Consequently, such interactions might be associated with delayed or inadequate response to therapeutic protocols, potentially compromising the prognosis and quality of life of patients.5811 When addressing a SC, comprehensive consideration should be given to how the management approach for one symptom influences other manifestations in the cluster and the interactions among various management strategies. The symptom management model was initially introduced by Larson et al,12while Dodd et al13 subsequently enhanced and refined this framework in 2001, later designating it as symptom management theory in 2008. This theoretical framework encompasses three interconnected dimensions: symptom experience, management strategies and clinical outcomes. Symptom experience is characterised by a patient’s perception, evaluation and behavioural response to presenting symptoms. Consequently, effective symptom management approaches should be grounded in comprehensive patient symptom assessment and evaluation. Dodd et al14 and Kim et al15 identified intricate relationships among two or more coexisting symptoms that demonstrate specific characteristics, including correlation patterns, temporal simultaneity, potential multidimensional properties and stability profiles, which may be linked to shared physiological and pathological pathways. The underlying mechanisms of symptom manifestation can be elucidated through the identification of key symptomatic indicators. Sentinel symptoms function as predictive markers that are associated with the initiation or advancement of additional symptoms within designated SCs, demonstrating substantial correlations with other cluster components and potentially contributing to their emergence or intensification.1618 Within symptom network theory, core symptoms are understood to reflect central interactions within the network, while bridge symptoms act as conduits for symptom propagation across different clusters. Furthermore, sentinel symptoms are defined by their ability to predict the onset of other symptoms within their own cluster—including both core and bridge symptoms.19 Therefore, by predicting the emergence of key symptoms, a clinical focus on sentinel symptoms could advance the understanding of underlying cluster mechanisms, thus facilitating targeted, pre-emptive interventions that alleviate the symptom burden, optimise management efficacy and improve patient quality of life.16 19 Therapeutic approaches and interventions targeting sentinel symptoms may contribute to reducing the overall burden of cancer SCs, potentially offering clinical guidance in lung cancer management16 19 20 and breast cancer.21 22 Therefore, investigation of sentinel symptoms represents a significant area of research for developing targeted interventions addressing SCs. Based on the aforementioned background, as research into SCs continues to advance, current evidence suggests that toxic side effects associated with cancer progression and chemotherapy protocols may be linked to the development of sentinel symptoms within these clusters, with such associations reflected through modifications in laboratory parameters.2325 In investigations examining lymphoma SCs, prior research has employed instruments such as the Memorial Symptom Assessment Scale (MSAS) combined with exploratory factor analysis (EFA) to identify three to five SCs throughout the prechemotherapy, chemotherapy and survivorship periods.4 6 26 27 These observations consistently reveal the clustered presentation of symptoms among patients with lymphoma. Nevertheless, the composition and heterogeneity of these clusters remain inadequately characterised, with unclear interactions between individual symptoms within a cluster. Moreover, chemotherapy-induced toxicities are not restricted to subjective symptoms but are also expressed as modifications in laboratory parameters. Although previous investigations have examined the distribution and influencing factors of SCs in chemotherapy-treated patients with lymphoma, significant research gaps continue to exist: (1) sentinel symptoms within clusters have not been systematically explored using association rule mining techniques, (2) potential correlations between sentinel symptoms and laboratory indicators remain unexamined. These limitations impede the advancement of precise symptom management strategies and reduce the effectiveness of individualised interventions for this population. Consequently, this investigation employed the MSAS combined with EFA to assess symptoms in patients with lymphoma to identify SCs and determine sentinel symptoms within these clusters. Furthermore, correlation analyses were performed between laboratory indicators and sentinel symptoms in patients receiving chemotherapy. These observations provide a theoretical foundation for developing individualised nursing interventions customised to sentinel symptoms and SCs in patients with lymphoma receiving chemotherapy. Methods Study design and participants This was a single-centre cross-sectional study. The Strengthening the Reporting of Observational Studies in Epidemiology statement guidelines were employed to report this investigation. A convenience sampling approach was employed to recruit patients receiving chemotherapy for lymphoma who attended the Shanxi Bethune Hospital between August 2024 and January 2025. The inclusion criteria were established as follows: (1) individuals aged 18 years or above; (2) fulfilment of diagnostic standards outlined in the 2024 lymphoma diagnosis and treatment guidelines with a pathologically confirmed lymphoma diagnosis; and (3) adequate cognitive awareness and capacity for routine verbal communication. The exclusion criteria were defined as follows: (1) current or prior psychiatric disorders/cognitive impairment (determined through a review of patient medical records and interviews with caregivers); (2) severe systemic infections, concurrent malignancies or serious physical conditions affecting cardiac, cerebral or pulmonary systems; (3) receipt of psychological intervention or exercise-based therapy within the preceding 3 months; (4) enrolment in alternative clinical investigations; and (5) inability to complete a minimum of one chemotherapy cycle due to various factors. This study was registered with the China Clinical Trial Center (ChiCTR2500096487). Sample size The sample size for this investigation was determined using the following formula for prevalence studies28: N= (Zα/2/δ)2p (1 – p). The significance level was established at α=0.05 (Zα/2=1.96), with acceptable error δ=0.05. Based on prior research, the prevalence of symptoms among patients with lymphoma undergoing chemotherapy was documented at 80.5% (p=0.805),29 yielding a minimum required sample size of 242 participants. To ensure adequate numbers of valid questionnaires, the sample size was expanded by 10% throughout the survey period. Therefore, a minimum of 267 patients with lymphoma receiving chemotherapy and satisfying enrolment criteria were recruited. Data collection Data collection was conducted through paper-based questionnaires comprising two distinct sections: the Demographic Questionnaire and the Chinese version of the MSAS (MSAS-CH). This assessment was conducted on day 7 of the chemotherapy cycle, with the day of drug infusion defined as day 1. Prior to data collection, eligible participants were informed of the study’s purpose and significance. Following the provision of consent, they were instructed on how to complete the questionnaires. The demographic questionnaire was completed by investigators through the electronic medical records system, while the MSAS-CH was self-administered by study participants. Instruments General information questionnaire The General Information Questionnaire encompassed two components. The first component concentrated on demographic characteristics, including age (years), sex, education level, medical payment method and additional demographic information. The second component incorporated clinical disease parameters, such as pathological staging and chemotherapy cycles, along with laboratory indicators, including white blood cell (WBC) count, haemoglobin (Hb) level, platelet (PLT) count, plasma D-dimer level, glutamic pyruvic transaminase level (ALT), glutamic oxaloacetic transaminase and albumin (ALB) level, which were retrieved by the investigators from the hospital’s electronic medical record system. The Chinese version of the Memorial Symptom Assessment Scale Symptom evaluation was performed using the MSAS-CH, initially developed by Portenoy et al30 and subsequently adapted for the Chinese cultural context by Cheng et al.31 This instrument was constructed to evaluate patients’ self-reported symptoms during the preceding week. It comprises three subscales: the Physical Symptom Subscale Score, which contains 32 items; the Psychological Symptom Subscale Score; and the Global Distress Index. All items were evaluated using a Likert scale. Specifically, the prevalence, frequency and severity of symptoms were assessed on a 4-point scale ranging from 1 to 4. The mean prevalence frequency, severity and distress scores were considered as the scores for each symptom. The distress level was evaluated on a 5-point scale ranging from 1 to 5, with the score for each symptom representing the mean of its severity and distress scores. Elevated scores are associated with more severe distress. The MSAS-CH exhibited satisfactory reliability and validity, with Cronbach’s α=0.79 to −0.87 and structural validity=0.68 to −0.79. It is extensively used for assessing cancer symptoms in China.31 Statistical analysis Initially, research data were compiled in Excel spreadsheets by two researchers and subsequently subjected to statistical analysis using SPSS V.26.0 and SPSS Modeler V.18.0. Count data were characterised through frequency and composition ratios. Normally distributed data are denoted as means±SD. Symptoms exhibiting a prevalence rate ≥25%32 33 among patients with lymphoma undergoing chemotherapy were selected for subsequent analysis. Subsequently, EFA was employed for SC identification, with symptom severity and factor analysis loadings being determined through principal component analysis and maximum variance factor rotation. The number of factors to retain was determined using the eigenvalue-greater-than-one criterion. Factors with an eigenvalue >1 and factor loadings >0.3 were extracted as common factors. The criteria for symptom group extraction were established as follows: (1) factor loading >0.3, whereby when identical symptom loadings exceeded 0.3 across multiple factors, the maximum loading determined symptom attribution; (2) each factor encompassed a minimum of two symptoms; and (3) Cronbach’s α coefficient exceeded 0.7. Prior to analysis, data validity was evaluated through Kaiser–Meyer–Olkin (KMO) measurement and Bartlett’s spherical test. When KMO values exceeded 0.6 and Bartlett test results were <0.05, the data were deemed appropriate for factor analysis.19 After SC extraction, association rule learning algorithms are commonly used for sentinel symptom identification within SCs. Apriori represents an association rule learning algorithm that has been employed to explore potential relationships across numerous medical field studies in recent years. Apriori can establish the frequency of item set occurrence within databases and association rule significance based on specific metrics, including support, confidence and lift. This algorithm can be used to generate valid symptom connections within SCs, thereby facilitating sentinel symptom identification. Support and confidence constitute standard metrics employed to assess symptom relationships. Support quantifies the probability of antecedent and consequent symptom co-occurrence, while confidence measures the conditional probability of simultaneous antecedent symptom presence and consequent symptom manifestation. Lift represents the ratio of consequent symptom appearance probability given antecedent symptom presence to consequent symptom appearance probability under any conditions.34 According to Apriori algorithm principles, intracluster symptom correlation analysis employs thresholds established as support >40%, confidence >60%,22 confidence>support and lift >1, indicating valid symptom relationships and thereby confirming sentinel symptoms. Finally, correlations between sentinel symptom severity and laboratory indicators were evaluated using Spearman’s correlation analysis. p<0.05 was considered statistically significant. Results Participant characteristics A total of 285 questionnaires were distributed throughout this investigation, with 17 invalid questionnaires being excluded (6 containing regular responses and 11 with incomplete information). This process yielded 268 valid questionnaires, representing an effective response rate of 94.03%. The mean age was recorded as 58.34±14.72 years, with the majority of participants (99.25%) having access to medical payment coverage. Approximately 80% of patients with lymphoma were classified within pathological stages III and IV, the mean body mass index was documented as 23.58±3.21 kg/m² and Eastern Cooperative Oncology Group scores ranged between 1 and 2 (66.1%). The majority of patients (69.4%) were not diagnosed with common chronic conditions such as diabetes and hypertension (onlinesupplemental file 1 online supplemental table 1). Scores of symptoms in patients with lymphoma undergoing chemotherapy The prevalence and severity of symptoms among patients with lymphoma who received chemotherapy demonstrated considerable heterogeneity, with the most commonly reported symptoms being energy deficiency/fatigue (92.2%), followed by appetite reduction (69.0%) and alopecia (66.4%) (online supplemental table 2). Composition of each SC in patients with lymphoma undergoing chemotherapy The KMO statistic for the factor analysis was determined to be 0.748, while Bartlett’s test of sphericity revealed χ² = 1543.363 (p<0.01), indicating that the factor analysis outcomes were appropriate. The Cronbach’s alpha coefficients for the four SC components were all observed to exceed 0.8, demonstrating elevated internal consistency. Based on the factor analysis findings (online supplemental table 3) combined with the varimax orthogonal rotation methodology, only symptoms exhibiting an eigenvalue >1, factor loading >0.3 and symptom prevalence rate ≥25% were incorporated into the analysis. 12 symptoms, encompassing difficulty concentrating/mind, nervousness, difficulty urinating, vomiting, difficulty/shortness/tightness in breath, diarrhoea, problems with sexual interest/activity, skin itching, difficulty swallowing, weight loss, constipation and swelling of the arms or legs, were eliminated. Based on the symptom characteristics, four SCs were identified: emotion–fatigue–sleep SCs (anxiety/worrying, feeling sad, feeling irritable, feeling drowsy, difficulty sleeping/not good sleep, dizziness and lack of energy/fatigue), nutritional metabolism SCs (nausea, lack of appetite and change in taste/food taste), lymphoma-specific SCs (feeling swelling/abdominal bloating, sweating, dry mouth, oral sores, pain and numbness/tingling in hands/feet) and chemotherapy–body SCs (skin changes, hair loss, coughing and feeling ‘I do not look like myself’). The Cronbach’s α coefficients for these SCs were observed to be 0.891, 0.892, 0.820 and 0.854, respectively. Sentinel symptoms within four SCs among patients receiving chemotherapy for lymphoma Based on the Apriori algorithm, association analysis was performed on different SCs to identify the sentinel symptoms in each group along with their support, confidence and lift (online supplemental table 4). In the emotion–fatigue–sleep SC, feeling drowsy was identified as a sentinel symptom (support: 44.40%, confidence: 97.48%, lift: 1.06) and demonstrated a significant predictive capacity for other symptoms in the group, including lack of energy/fatigue, difficulty sleeping/poor sleep and irritability. Nausea represented the sentinel symptom of nutritional metabolism SCs (support: 54.10%, confidence: 85.52%, lift: 1.24), sweating was associated with the sentinel symptom of lymphoma-specific SCs (support: 57.09%, confidence: 74.51%, lift: 1.19) and skin change corresponded to the sentinel symptom of chemotherapy body SCs (support: 52.99%, confidence: 77.46%, lift: 1.17). Correlation between the severity of different sentinel symptoms and laboratory indicators in the symptom group of patients with lymphoma undergoing chemotherapy Drowsiness manifestations were found to be positively associated with WBC (r=0.125, p=0.04) and ALT (r=0.133, p=0.03) while demonstrating negative associations with Hb (r=−0.213, p=0.01), ALB (r=−0.125, p=0.04) and PLT (r=−0.138, p=0.02). Perspiration symptoms exhibited negative associations with PLT (r=−0.121, p=0.04). Dermatological manifestations were inversely associated with D-dimer concentrations (r=−0.121, p=0.04). Nausea, recognised as a sentinel manifestation, demonstrated no significant associations with the laboratory parameters examined (p>0.05) (online supplemental table 5). Discussion Prevalence and severity of symptoms in patients with lymphoma undergoing chemotherapy In this investigation, symptom assessment was conducted among 268 patients with lymphoma receiving chemotherapy. The findings indicate the coexistence of multiple symptoms in this patient population. Patients with lymphoma undergoing chemotherapy demonstrated symptom clustering patterns, with prodromal symptoms being closely correlated with other manifestations within the cluster. The investigation revealed that the most frequently reported symptom following chemotherapy for lymphoma was lack of energy/fatigue, which remained consistent with the most prevalent symptoms observed in patients with cancer, as documented by Bolukbas and Kutluturkan35 and Qiu et al.6 Fatigue symptoms among patients with cancer are recognised as being among the most commonly encountered36 and are associated with survival outcomes.37 Additionally, chemotherapy is used in lymphoma treatment,38 with patients experiencing fatigue symptoms linked to myelosuppression following chemotherapy administration.39 Consequently, regular fatigue screening should be implemented for all patients at initial diagnosis, throughout cancer treatment and post-treatment based on clinical indications.40 Moreover, the prevalence of appetite loss was documented at 69%, which corresponds with findings reported in previous investigations.41 42 In the management of patient with cancer, appetite loss is frequently recognised as part of the gastrointestinal symptom constellation and is closely linked with nausea and vomiting.43 Considering the symptom interaction within the symptom group,14 nausea and vomiting may influence patients’ appetite, subsequently reducing nutritional intake, which might contribute to eating-related distress44 and potentially affect physical function. The study findings suggest that non-pharmacological interventions, including corresponding drug interventions or relaxation training, may be implemented to ensure adequate nutritional intake and alleviate symptom burden.9 Finally, hair loss was associated with a higher prevalence (66.4%) among patients with lymphoma receiving chemotherapy, which aligns with findings documented by Qiu.7 Most chemotherapeutic agents are associated with effects on hair follicle cell division and growth, potentially contributing to hair loss. Compared with chemotherapeutic drugs, targeted agents, such as everolimus, may be linked with reduced hair loss occurrence.45 The study findings suggest that hair loss interventions are clinically significant for potentially improving patients’ psychological and physical well-being. Composition of SCs in patients with lymphoma undergoing chemotherapy This study used EFA to define the composition of SCs and demonstrated that patients with lymphoma could be categorised into four SCs during chemotherapy. The first SC was identified as the emotional–fatigue–sleep SC, which encompassed anxiety/worrying, feeling sad, feeling irritable, feeling drowsy, difficulty sleeping/not good sleep, dizziness, lack of energy/fatigue and stable SCs throughout the course of chemotherapy. This phenomenon might be correlated with biological mechanisms involving neurotransmission and excessive release of proinflammatory cytokines.46 47 Current evidence suggests that inflammation and neuroendocrine–immune dysfunction are associated with the onset and maintenance of cancer-related SCs.48 Cancer treatments trigger neurotransmission and inflammatory–immune responses,47 49 50 subsequently influencing the hypothalamic‒pituitary‒adrenal (HPA) axis system,51 52 potentially contributing to patient-reported symptoms, such as fatigue, sleep disturbances and anxiety.5355 Studies have demonstrated that chemotherapy physically and mentally impacts patients with lymphoma, being associated with a cumulative pain effect that continues to compromise their emotional and physical functions.56 Based on the findings of this study, healthcare administrators should comprehensively monitor patients’ symptom experiences and psychological needs throughout their treatment trajectories. Considering the dynamic evolution of chemotherapy-related symptoms in patients with lymphoma across cycles, a dynamic symptom management model should be implemented to deliver proactive interventions,5 including non-pharmacological approaches such as progressive muscle relaxation, massage therapy, aerobic exercise, mindfulness meditation and yoga.57 These interventions are designed to mitigate symptom burden and enhance patients’ quality of life. The second SC was identified as nutritional metabolism SCs, encompassing nausea, lack of appetite and changes in taste, which aligns with the results of Rha and Sezgin.27 56 Damage to the digestive tract mucosa, glands and taste buds by chemotherapeutic drugs may contribute to symptoms, such as decreased appetite, nausea and vomiting, which are linked to the formation of a symptomatic group.58 Nausea can increase sympathetic excitability, which, in turn, increases duodenal tension and is associated with a lack of appetite. Combined with the results of this study, patients should be informed of the relevant knowledge about nausea, and standardised management should be carried out before, during and after chemotherapy. The severity of symptoms should be assessed to explore multidimensional management methods suitable for patients, such as psychological counselling, exercise, physical therapy, traditional Chinese medicine and the use of appropriate ginger and probiotics to alleviate nausea symptoms.5861 The third SC was characterised as lymphoma-specific SC, encompassing feelings of swelling/abdominal bloating, sweating, dry mouth, oral sores, pain and numbness or tingling of the hands or feet. These symptoms are comparable to those reported in studies on other haematological malignancies and solid tumours. Clinically, most patients with lymphoma present with enlarged lymph nodes or compressive effects caused by these enlarged nodes, which are associated with a high prevalence of perceived body swelling. Bolukbas and Kutluturkan35 reported that mouth sores and dry mouth are induced by chemotherapeutic drugs and are potentially linked with the active proliferation of oral mucosal cells, which are more susceptible to the effects of these drugs.62 Chemotherapy-induced oral mucositis (CIOM) represents a common adverse reaction in patients undergoing chemotherapy for haematological malignancies, with a prevalence rate of 20–80% .63 64 CIOM compromises the integrity of the oral mucosa, being associated with hypogeusia, oral pain, dysphagia, malnutrition and communication difficulties, severely undermining the quality of life of patients.65 Studies have demonstrated that oral ulcer protectants can prevent damage to the oral mucosa and alleviate pain.66 Furthermore, chemotherapeutic agents have low selectivity and damage normal cells, tissues and organs, which is correlated with peripheral neuropathy and varying degrees of numbness or tingling in the hands and feet.67 Additionally, drugs such as alkaloids can contribute to chemotherapy-induced peripheral neurotoxicity (CIPN), being linked to hand/foot numbness or tingling.68 Therefore, this study suggests the necessity of screening for CIOM and CIPN risks and implementing early protective measures in patients undergoing chemotherapy. The fourth SC was designated as chemotherapy–body SC, encompassing skin changes, hair loss, coughing and the sensation of ‘I do not look like myself’. Chemotherapeutic agents, such as lenalidomide, asparaginase and gemcitabine, often induce skin itching. One of the adverse effects of doxorubicin is haemorrhagic erythematosus, whereas methotrexate and bleomycin can be associated with skin changes owing to pigment accumulation.7 Patients directly experience skin changes and hair loss, along with perceived bodily swelling and various burdens associated with the treatment, thus feeling that ‘I do not look like myself”. This psychological and social impact is particularly pronounced among women with cancer.69 Due to the disease itself and prolonged treatment, the immune system is suppressed in patients with lymphoma, thereby increasing the risk of infections, particularly respiratory tract infections, which may contribute to coughing.70 71 The mechanisms underlying chemotherapy-induced somatic SC require further study, suggesting that a holistic approach should be adopted for patients and early intervention measures tailored to their needs should be developed. Sentinel symptoms in the symptom group of patients with lymphoma undergoing chemotherapy and their correlation with laboratory indicators This study identified four sentinel symptoms, namely, drowsiness, nausea, sweating and skin changes. Drowsiness was characterised as a sentinel symptom for the emotion–fatigue–sleep SC. It was positively correlated with WBC and ALT levels and negatively correlated with Hb, ALB and PLT levels. The observed laboratory abnormalities, notably reduced Hb and ALB levels, are likely linked to multifactorial interactions involving lymphoma pathophysiology, glucocorticoid therapy and chemotherapy-related toxicities. In this cohort, reduced Hb and ALB levels are associated with immune dysfunction and cancer-related anaemia and may be connected to symptoms such as lethargy, fatigue and dizziness. 72 Glucocorticoids, a cornerstone of lymphoma treatment regimens,2 73 substantially enhance susceptibility to opportunistic infections,74 whereas lymphoma-associated immune dysregulation and concurrent infections synergistically elevate WBC counts.75 Chemotherapy-induced bone marrow suppression diminishes PLT production, and severe infections worsen thrombocytopenia through coagulation pathway activation and PLT consumption.76 Concurrently, drug-induced liver injury is manifested as elevated ALT, intensifying systemic weakness and malaise.77 The interplay between chemotherapy toxicity, haematological derangements (WBC/PLT/ALT), environmental stressors and treatment-related disruptions predisposes patients to sleep disturbances. These sleep deficits dysregulate the HPA axis, amplify systemic inflammation via proinflammatory cytokine cascades and worsen psychological comorbidities, including anxiety and depressive symptoms.78 A recent study demonstrated that Chan-Chuang qigong practice modulates hypothalamus-mediated homeostatic responses by stabilising autonomic nervous system activity, thereby ameliorating cancer-related fatigue and improving sleep quality in patients with non-Hodgkin’s lymphoma.79 Furthermore, this mind–body intervention positively modulates haematological parameters, including erythrocyte and leucocyte counts and elevated Hb levels.80 Mechanistically, these effects were associated with enhanced immune reactivity, suggesting a potential neuroimmune regulatory role of Chan-Chuang qigong in oncological care. Critically, the temporal primacy of drowsiness within the sleep–fatigue–sleep disturbance SC remains unclear. Furthermore, the expert consensus on nutritional therapy for haematological malignancies indicates that chemotherapy-induced myelosuppression can adversely affect patient metabolism and nutritional status; routine nutritional screening and assessment for patients with lymphoma is therefore recommended. Given the weak correlations observed in our study between drowsiness and levels of ALB and WBC, this symptom warrants careful clinical interpretation when present.81 These findings highlight the importance of integrated monitoring of laboratory biomarkers, psychological states and sleep quality in the early identification of sentinel symptoms. Proactive interventions targeting these symptoms may disrupt the self-reinforcing mechanisms of symptom clustering, thereby attenuating downstream cascading effects and improving patient-centred outcomes. Nausea was identified as a sentinel symptom of nutrition–metabolism SC, which is not markedly associated with laboratory indicators and may be driven by the direct toxicity of chemotherapy drugs rather than by systemic pathophysiological changes. Longitudinal research on patients with lymphoma during chemotherapy revealed that while the incidence of nausea progressively increased throughout the treatment cycles, the severity of the symptom remained stable.82 This may be attributed to the comprehensive management strategies advocated in clinical guidelines and expert consensus, which emphasise risk assessment, graded management and personalised interventions (including pharmacological, dietary and lifestyle modifications).8385 These proactive measures likely mitigated severe symptom burden, which may explain the lack of correlation observed with laboratory parameters. Chemotherapy nausea in patients with cancer may be associated with the activation of a series of inflammatory biological processes in the gastrointestinal tract, contributing to more perturbation of signalling pathways, including the activation of the NF-κB signalling pathway and other pathways, such as apoptosis and cytokine–cytokine signalling.86 These intricate processes complicate gastrointestinal inflammation and contribute to the development of other gastrointestinal symptoms, such as vomiting and a lack of appetite.62 87 88 Sweating has been identified as a sentinel symptom of the lymphoma-specific SC and is negatively associated with PLT levels. In a study by Zheng et al,89 sweating and dry mouth were prominent symptoms, accompanied by varying degrees of pain and numbness. These symptoms may be associated with the fact that patients with lymphoma often have a weak constitution or deficiency secondary to pathogenic excess in traditional Chinese medicine. In this study, approximately 80% of the patients were diagnosed with stage III or IV lymphoma, indicating extensive lymph node infiltration and a clinical state of prolonged illness-induced physical depletion. However, in patients with relapsed or refractory disease, the use of chemotherapeutic agents and glucocorticoids further compromises immune function and enhances the risk of infection. PLTs, which are recognised as first-line sentinels against bacterial infections,90 are rapidly depleted during infection control through mechanisms such as pathogen neutralisation and inflammatory modulation. Despite the weak correlation between sweating and PLT count, which may be related to disease complexity, clinical practice should incorporate comprehensive symptom assessment in patients with lymphoma. A management strategy focusing on sentinel symptoms is advocated to address SCs effectively. Skin changes were characterised as sentinel symptoms for chemotherapy–body SC and are negatively associated with D-dimer levels. Studies have shown that changes in D-dimer levels are related to the occurrence, treatment and prognosis of lymphoma.91 Additionally, drugs used to treat lymphoma accelerate keratinocyte apoptosis, thereby weakening the patient’s skin resistance and contributing to a series of adverse skin reactions, such as dry skin, pruritus (lenalidomide) and pigmentation (methotrexate and bleomycin pigment deposition contribute to darkening of the skin and doxorubicin haemorrhagic erythema).2 44 85 86 92 Skin changes are both severe and disturbing and may affect the composition of the SC. A weak correlation was observed between skin changes and D-dimer levels in this study. As such, a cautious interpretation of its clinical significance is required, which calls for an understanding of the specific chemotherapeutic agents used and early changes in laboratory parameters. Consequently, patient management should involve monitoring for skin changes and adverse reactions, coupled with guidance on skin cleansing, moisturising and symptomatic intervention to reduce the prevalence of these symptoms.7 86 93 Limitations This study has several limitations. Its cross-sectional design precludes causal inference regarding symptoms and prevents analysis of their temporal dynamics. Additionally, the reliance on patient self-reported data may introduce recall bias. Finally, the symptom identification and analysis relied on exploratory methods (factor analysis, Apriori algorithm and correlation analysis) without confirmatory factor analysis to verify the model’s stability and generalisability. Future longitudinal studies integrating diverse statistical methods are needed to dynamically monitor symptom changes in patients undergoing chemotherapy and to facilitate a more profound investigation into SC mechanisms. Conclusions This investigation used the Apriori algorithm to identify corresponding sentinel symptoms and evaluate their associations with laboratory indicators. Multiple SCs were observed in patients receiving chemotherapy for lymphoma, and correlations were found between sentinel symptoms and laboratory parameters. These findings are associated with the predictive and indicative properties of these symptoms. Sentinel symptoms may serve as a focal point for symptom management. Emphasis on screening and early intervention targeting sentinel symptoms might enhance the effectiveness of interventions on SCs and reduce patient discomfort, thereby ensuring consistent and comprehensive treatment. Future interventional research will examine the efficacy of sentinel symptom-based interventions in enhancing SC management and patient outcomes. Supplementary material