Characterizing the distribution of Chinese medicine syndrome of cancer-related fatigue in childhood lymphoma based on latent structure model

Abstract

OBJECTIVE:

To explore the distribution patterns of Traditional Chinese Medicine (TCM) syndromes in children with lymphoma-related cancer-related fatigue (CRF) using latent structure model analysis.

METHODS: This was a single-center cross-sectional study in which Chinese medicine diagnostic information was collected from 153 children diagnosed with childhood lymphoma CRF from July 2022 to December 2023. The latent structure model of the CRF for childhood lymphoma was constructed using the latent tree model-extension adjustment simplification until termination algorithm, and the population was classified using comprehensive clustering. Combined with the existing guidelines, we explored the distribution characteristics and dynamic progression of syndrome types in patients with childhood lymphoma CRF.

RESULTS:

Initially, a latent structural model incorporating 72 manifest variables was constructed. Through comprehensive clustering, 6 distinct TCM syndrome types were identified, and corresponding quantitative identification rules were formulated. The 153 pediatric cases were then reclassified according to these rules, with the syndromes ranked in descending order of percentage as follows: blood deficiency and stasis, vigorous fire due to Yindeficiency, deficiency of both the lung and spleen, liver constraint and Qi stagnation, dampness-heat retention, and heart-spleen heat accumulation. Preliminary findings suggest that the shifting patterns of TCM syndromes in childhood lymphoma CRF may exhibit clinically meaningful relationships with tumor staging and treatment time courses.

CONCLUSION:

Using latent structure analysis, 6 TCM syndromes and their quantitative identification rules were identified. A comparison with existing guidelines revealed that CRF in patients with childhood lymphoma is characterized primarily by a syndrome of mixed deficiency and excess. Therefore, in the treatment of pediatric CRF, it is recommended not only that deficiency patterns be addressed but also that the clinical stage and treatment phase of the tumor be considered. Additionally, therapeutic strategies should emphasize promoting digestion, resolving food stagnation, regulating Qi, and clearing heat to formulate a tailored treatment plan.

1. INTRODUCTION

The National Comprehensive Cancer Network of the USA (NCCN) defined cancer-related fatigue (CRF) as “a distressing, persistent, subjective sense of physical, emotional and/or cognitive tiredness or exhaustion related to cancer or cancer treatment that is not proportional to recent activity and interferes with usual functioning”.¹ In a report from the International Late Effects of Childhood Cancer Guideline Harmonization Group, the prevalence of CRF was found to range from 10.2% to 85.0% in childhood, adolescent, and young adult cancer survivors.² The report highlighted that young adult survivors faced a greater risk of fatigue and experienced more severe fatigue compared with healthy controls.²

As a common pediatric malignancy, the primary treatment for lymphoma is chemotherapy. Chemotherapy cycles can last up to 1-2 years or even longer. During treatment, children with tumors experience varying degrees of fatigue,³ which negatively impacts their learning, mood, and daily activities.⁴ In some cases, treatment may be interrupted because of severe fatigue.⁵ Prolonged and severe fatigue can even affect survival outcomes of patients with cancer.⁶ Therefore, addressing CRF in these patients is clinically important. Alleviating fatigue following chemotherapy is also a priority for parents seeking practical solutions during treatment.

In addition to chemotherapy, the occurrence of CRF is also associated with various factors, such as anemia, cancer pain, sleep disorders, depression and anxiety, highlighting the complexity of CRF pathogenesis.^7,8 The exact mechanisms underlying CRF remain unclear, and specific early diagnostic markers are lacking. Consequently, there is no established gold-standard intervention for managing fatigue in patients with CRF. The NCCN guidelines recommend nonpharmacological treatments, such as exercise interventions, psychosocial interventions, and nutritional support, to improve fatigue in children.¹ However, the level of evidence for these recommendations varies from moderate to low.¹ To date, there is insufficient evidence to demonstrate the efficacy and safety of pharmacological interventions for fatigue in cancer patients and survivors.⁹

Since ancient times, Traditional Chinese Medicine (TCM) has been effective in the treatment of chronic fatigue disorders.¹⁰ In recent years, TCM therapy for CRF has become a significant area of focus.¹¹ The evidence suggests that TCM therapy works well against CRF in adult tumor patients.¹² For children with cancer, research on CRF has focused primarily on nonpharmacological interventions such as exercise and psychological therapies, although the quality of these studies varies.² Meta-analyses have shown that TCM combined with chemotherapy can improve the overall response rate to treatment and reduce cancer-related symptoms such as nausea and vomiting, anemia and fatigue.¹³ However, there is a lack of systematic research data on the use of TCM for CRF in children.

The process of TCM treatment begins with classifying patients into different positive subtypes on the basis of their symptoms and signs, tongue features and pulse. This process is known as syndrome differentiation. It guides the selection of formulas tailored to each syndrome, which in turn affects the effectiveness of TCM treatment. On the basis of clinical data on adult tumors, the Clinical Practice Guidelines for Cancer-related Fatigue in China (2021 edition) established syndrome differentiation criteria for CRF.¹⁴ Studies have shown that pediatric and adult tumors differ in terms of pathological subtypes, body constitution, immune status, and treatment protocols, leading to variations in TCM syndromes between children and adults.^15-17 Therefore, it is essential to conduct a TCM syndrome study on the CRF in childhood lymphomas. To address this point, our investigation provided reliable clinical data and compared the similarities and differences in CRF between adults and children. This study also provides a syndrome basis for TCM treatment of CRF in patients with childhood lymphoma.

Common data analysis methods in the study of TCM syndrome element research include cluster analysis, factor analysis, and regression analysis.¹⁸ Each of these methods has its own strengths and limitations in clinical application. However, none of them can completely capture the fundamental characteristics and complex internal relationships of syndrome elements. The latent tree structural model method we used is a new method to study TCM syndromes.¹⁹ This method not only maximizes the advantages of other analytical methods but also enables quantitative research on the elements of syndrome and the construction of complex relationships among symptoms, syndrome elements and syndrome. This objective, multidimensional clustering approach to CRF in patients with childhood lymphoma is more closely aligned with the TCM system of diagnosis and treatment.¹⁶

2. METHODS

2.1. Participants

The present study used data from a cross-sectional survey performed in Beijing. Participants were recruited from July 2022 to December 2023 from Beijing Children's Hospital, a Chinese tertiary general pediatric hospital and national pediatric tumor detection center. The diagnosis of childhood lymphoma was confirmed by attending physicians in both inpatient and outpatient settings, based on the WHO Classification of Tumors of Haematopoietic and Lymphoid Tissues (Revised 4th ed).²⁰ Children with cancer were recruited if they were between 8 and 18 years of age and met the diagnostic criteria for CRF according to the International Classification of Diseases, 10th edition.²¹ Participants who failed to complete the questionnaire or who provided poor-quality responses were excluded. Additionally, children with other malignant tumors or metastases, as well as those with serious underlying conditions such as heart, liver, or kidney disease, were excluded. To avoid interference from TCM treatment on the study results, children who had used Chinese herbs or proprietary Chinese medicines within four weeks prior to enrollment were also excluded. The children received standard chemotherapy on the basis of the 2017 edition of the China Net-Childhood Lymphoma treatment protocol.^22-24 For children with coinfections or severe myelosuppression, appropriate anti-infective treatments, colony-stimulating factor infusion, or blood transfusions are administered. Additionally, these children receive systemic supportive care, including hepatoprotective therapy, gastric protection with proton pump inhibitors, and other symptomatic treatments as necessary. The study was conducted ethically, and care was taken to ensure that the clinical care of the children will not be affected in any way. During the 18-month period, a total of 153 questionnaires were distributed in the survey.

2.2. Procedure

Ethics approval was obtained from the Ethics Committee of Beijing Children's Hospital of Capital Medical University under Grant No. [2022]-E-095-Y.

The survey involved face-to-face interviews conducted by pediatricians. Participation was voluntary, with both the children and their guardians agreeing to answer the interview questions and providing written informed consent. The questionnaire consisted of two parts: sociodemographic data and the TCM Syndrome Scale for CRF in children. The TCM Syndrome Scale was developed by members of the Department of TCM and the Lymphoma Specialty Group, with reference to the Chinese Medicine Clinical Terminology Part 2: Syndrome from the National Standards Network,²⁵ relevant textbooks,^26-28 and the clinical characteristics and practices specific to children. The scale includes 84 items, covering symptoms, signs, and details about the patient’s tongue and pulse (supplementary Tables 1, 2). Given the limitations in children's language abilities, the terminology used in the questionnaire was optimized to help children respond quickly and accurately. Parents were able to assist with the questionnaire at any time during the survey to ensure the accuracy and reliability of the responses. All clinical signs documented in the assessment scale were systematically recorded by certified TCM practitioners following standardized training protocols. On average, it took approximately 10 to 15 min for a child to complete the interview.

On the basis of the questionnaire responses, the data were standardized according to uniform criteria. Symptoms and signs were consistently coded as either present or absent. Upon completion of clinical information collection, SPSS Statistics 22.0 (IBM Corp., Armonk, NY, USA) was used to create a dataset to record all the data, with multiple double-entry procedures employed to ensure data quality.

2.3. Statistical analysis

The frequency of indicators was statistically analyzed using SPSS Statistics 22.0 (IBM Corp., Armonk, NY, USA) to identify the top 10 high-frequency symptoms and signs. A total of 72 symptoms and signs with a positive rate of 10% or higher were selected for inclusion in the subsequent study.

Latent structure analysis was employed to construct a model of TCM syndrome in three steps. This process was performed using Lantern 5.0 software from Professor Nevin Lianwen of the Hong Kong University of Science and Technology (Hong Kong, China), which was sourced from the official website.¹⁹ The first step is to establish a latent structural model of symptoms and signs of CRF with childhood lymphoma using the software's Latent Tree Model-Extension Adjustment Simplification until Termination (LTM-EAST) algorithm. This study aimed to provide a preliminary classification of symptoms and signs. Each latent variable in the model was interpreted on the basis of TCM knowledge, allowing for the identification of syndrome elements and target organs of the disease. The quality of the model was evaluated using the Bayesian information criterion (BIC) score, which was -6477.8, indicating that the latent structure model constructed from the current explicit data was the best.

The symptoms and signs included in the model were the manifest variables, and the data were analyzed to derive the elliptical latent variable Y. The latent variable is a high-level summary of the logical connections among the manifest variables, reflecting the combination patterns of the associated symptoms and signs. Each latent variable further divides the data into two or three latent categories. The numbers in parentheses of the elliptical latent variables represent the number of latent class states. The latent categories are "yes" or "no" relationships, indicating the likelihood of each listed symptom and sign being present or not present, respectively.

The thickness of the line between the latent variable and the manifest variable represents the strength of the association between the variables. For example, thicker lines between Y0 and poor appetite, Y3 and heavy body and drowsiness, Y8 and timidity, and Y10 and greasy fur indicate a stronger association between these variables. In contrast, thinner lines between Y2 and petechia, Y7 and rapid pulse, Y8 and hematochezia, and Y9 and red tongue suggest a weaker association in defining the latent variables. This may be due to these manifest variables being common symptoms with less specificity for distinguishing appropriate latent variables.

To address potential local independence in the algorithm, we adopted a comprehensive cluster analysis using a human-computer collaboration approach to further clarify the complex relationships among syndrome elements, symptoms and signs. The expert team comprised nine senior-level specialists from the Malignant Tumor and Hematological Disease Group at the Beijing Institute of Integrative Chinese and Western Medicine (BICWM). The team developed 17 common syndrome elements and their respective characteristics for lymphoma in children, with reference to the Chinese Medicine Clinical Terminology Part 2: Syndrome from the National Standards Network,²⁵ relevant textbooks,^26-28 and clinical practice. Accordingly, the symptoms and signs of the latent variables in the model were identified by TCM syndrome elements, including both disease location syndrome elements and disease nature syndrome elements. Latent variables with the same TCM syndrome elements or those intrinsically linked by disease mechanisms were grouped into a single classification. This step involved initial categorization through expert discussions and online voting. Next, different latent variables that could reveal various aspects of the same syndrome or reflect the underlying connotations of the same disease mechanism were extracted and reincorporated into the software for further comprehensive cluster analysis. As a result, we developed a latent structural model representing the different syndrome types of CRF in children, evaluated the model's quality using Bayesian information criterion scores, and plotted the corresponding information curves. Through manual interpretation, we preliminarily identified the TCM syndrome types for CRF in patients with childhood lymphoma.

Finally, using the naive Bayes classifier, we established quantitative identification rules for different syndromes, including symptom entries, corresponding scores, determination thresholds, and accuracy measures. These rules provide a more objective basis for assisting in clinical diagnosis than human judgment alone. We applied these identification rules to assess the 153 children with lymphoma included in the study and compared the distributions of TCM syndrome types between pediatric lymphoma CRF and adult CRF.

3. RESULTS

3.1. Analysis of demographic characteristics and tumor composition

A total of 153 children with lymphoma participated in the survey. Among these children, 107 (69.9%) were male, with a mean age of (10.8 ± 2.7) years, and 46 (30.1%) were female, with a mean age of (11.0 ± 2.0) years. Lymphomas are classified into Hodgkin's lymphoma (HL) and non-Hodgkin's lymphoma (NHL) on the basis of morphology and immunohistochemistry. In this study, 14 patients (9.15%) were diagnosed with HL, whereas 139 (90.85%) were diagnosed with NHL. Among the NHL patients, 52 (33.99%) had T-lymphoblastic lymphoma, 20 (13.07%) had B-lymphoblastic lymphoma, 22 (14.38%) had Burkitt's lymphoma, 9 (5.88%) had diffuse large B-cell lymphoma, 28 (18.30%) had anaplastic large-cell lymphoma, and 8 (5.23%) had other types of lymphoma (Table 1).

Table 1.

Statistical tables of basic information on demographic characteristics

Item		Frequency	Frequency distribution (%)
Sex	Male	107	69.93
	Female	46	30.07
Age (years)	8-9	58	37.91
	10-12	60	39.22
	13-15	24	15.69
	16-17	11	7.19
Lymphoma immunophenotyping	Hodgkin's lymphoma	14	9.15
	T-lymphoblastic lymphoma	52	33.99
	B-lymphoblastic lymphoma	20	13.07
	Burkitt's lymphoma	22	14.38
	Diffuse large B-cell lymphoma	9	5.88
	Anaplastic large cell lymphoma	28	18.30
	Others	8	5.23
Lymphoma clinical staging	Ⅰ	4	2.61
	Ⅱ	17	11.11
	Ⅲ	57	37.25
	Ⅳ	75	49.02
Disease course (months)	0-3	12	7.84
	4-6	109	71.24
	7-12	25	16.34
	over 12	7	4.58
Associated diseases/symptoms	bone marrow suppression	58	37.91
	dyspepsia	74	48.37
	vomiting	70	45.75
	sleep disorders	20	13.07
	pain	41	26.80
	mood disorders	81	52.94
	mucositis of the anal orifice	58	37.91

Notes: Hodgkin's lymphoma staging based on Ann Arbor staging, non-Hodgkin's lymphoma staging based on St Jude staging.

3.2. Statistics on high-frequency symptoms and signs

A total of 72 symptoms and signs (with a positivity rate of more than 10%) were included in this study, with 4669 occurrences, accounting for 98.65% of the total number of occurrences. The top 10 high-frequency symptoms and signs were deep-colored urine (74.51%), weakness (71.90%), spiritlessness (68.63%), irritability (67.97%), poor appetite (65.36%), pale lips and nails (64.05%), thick and greasy tongue coating (61.44%), cancer-related pain (60.78%), craving for cold (58.82%), and constipation (57.52%).

3.3. Cluster analysis and model interpretation of latent structure

As shown in Figure 1, the 72 symptoms and signs were taken as manifest variables, and the LTM-EAST algorithm was used to construct the latent structural model of CRF symptoms and signs of childhood lymphoma. The study identified 18 latent variables, labeled Y0 to Y17, which illustrated the connections between symptoms and reflected both pathological and etiological evidence in the children. Except for Y13, which contains 3 latent categories, the remaining latent variables had only 2 latent categories, indicating different classifications within the population.

Figure 1. Latent structural modeling of symptoms and signs in childhood lymphoma with cancer-related fatigue.

Take the latent variable Y4 as an example (supplementary Table 3). When the cumulative information coverage reached the specified threshold, no new manifest variables were added. With respect to shortness of breath, the cumulative coverage of Y4 reached 99%, indicating that these symptoms adequately explained the latent variables. Y4 contained 2 latent classes. This means that there were two states for the latent variable Y4: Y4 = S0 (P = 0.35) and Y4 = S1 (P = 0.65). This indicates a 35% probability of not exhibiting symptoms (S0) and a 65% probability of exhibiting symptoms (S1). In other words, the latent category S0 represented 35% of the population, whereas S1 accounted for 65%. The probabilities of having symptoms of weakness, spiritlessness and shortness of breath in the population of this latent category S0 were 0.2, 0.15, and 0.02, respectively, whereas they were 1, 0.97, and 0.59 for S1, respectively. The latent variables were interpreted in TCM through expert consultation. For example, if Y4 is interpreted as Qi deficiency symptoms, then patients in the S0 category have milder symptoms and patients, whereas those in S1 exhibit more severe symptoms. The interpretation of the other latent variables followed the same approach as Y4.

3.4. Comprehensive cluster analysis and model interpretation

To further explore the intrinsic links between the latent variables, we employed a comprehensive cluster analysis using a human-computer collaboration approach to cocluster the 18 latent variables into six categories, labeled Z1 through Z6. The comprehensive clustering model is shown in Figure 2, with detailed clustering information provided in Table 2 (supplementary Table 4). The BIC values ranged from -2199.66 to -688.25, indicating good model fit. The cumulative information coverage of 99% suggests that these symptoms and signs effectively capture the characteristics of the clusters. Each composite cluster contains 2 latent classes, denoted as S0 and S1. In accordance with the diagnostic standards outlined in the National Standard Network of TCM Clinical Terminology²⁵ and supplemented by multiple established textbooks,^26-28 the expert panel systematically performed TCM-specific interpretations of the symptoms and signs associated with the latent variables, which are shown in Table 2. For example, take Z2 for model interpretation:

Figure 2. Comprehensive clustering model diagram of children's CRF latent structure.

A: comprehensive clustering model diagram Z1; B: comprehensive clustering model diagram Z2; C: comprehensive clustering model diagram Z3; D: comprehensive clustering model diagram Z4; E: comprehensive clustering model diagram Z5; F: comprehensive clustering model diagram Z6. CRF: cancer-related fatigue.

Table 2.

Comprehensive clustering model symptoms, signs and symptoms distribution

CAT	Latent variable	Signs and symptoms	BIC	Pathogenetic syndrome elements	Pathological location syndrome elements
Z1	Y0, Y3, Y10, Y13, Y15, Y17	Greasy mouth without thirst, dry and uncomfortable eyes, loose stools, amnesia, skin oiliness, heavy body and drowsiness, dysuria, deep-colored urine, tastelessness without thirst, bitter taste in mouth, susceptibility, polydipsia, acne, abdominal bloating, hypersomnia, fear of cold	-2199.66	Dampness, heat	Liver, spleen, kidneys
Z2	Y4, Y8, Y11, Y16	Spiritlessness, weakness, timidity, night sweating, hypothymic depression, spontaneous sweating, shortness of breath, feverishness in palms and soles, pale tongue	-1246.09	Qi deficiency, Yin deficiency	Spleen, lungs
Z3	Y9, Y10, Y11, Y12,	Squamous and dry skin, plump tongue, blue or purplish tongue, weak pulse, thready pulse, pale lip and nail colour, pink tongue, red tongue, slippery fur, greasy fur	-1259.01	Blood stasis, blood deficiency	Heart, spleen
Z4	Y15, Y16	Night sweating, spontaneous sweating, dry and uncomfortable eyes, greasy mouth without thirst, dysuria, feverishness in palms and soles, amnesia	-866.83	Yin deficiency, heat	Liver, kidney
Z5	Y13, Y14	Sensitization, anxiety, emaciation, sighing, irritability, obesity, skin dryness, stringy pulse	-688.25	Qi stagnation, heat	Liver, heart
Z6	Y1, Y2, Y5, Y6, Y7	Poor appetite, nausea and vomiting, cold limbs, sore in mouth, light sleep, localized stabbing pain, sore throat, difficulty initiating sleep, difficulty with sleep consolidation, stink of excrement, halitosis	-2022.45	Heat, food retention, blood stasis	Spleen, heart

Notes: CAT: category; BIC: bayesian information criterion.

Category Z2: this category incorporated four latent variables — Y4, Y8, Y11, and Y16 — which included 14 symptoms and signs such as spiritlessness, weakness, and timidity. Among these, spiritlessness had the highest degree of mutual information with Z2, indicating its major contribution. When the symptoms and signs were sequentially up to the "pale tongue", the cumulative information coverage reached 99%. These findings indicate that spiritlessness, weakness, timidity, night sweating, hypothymic depression, spontaneous sweating, shortness of breath, feverishness in the palms and soles, and a pale tongue can adequately describe the characteristics of Z2.

Z2 was divided into two latent classes: S0 and S1. The a priori probability of Z2 = S0 is 0.62, whereas Z2 = S1 is 0.38, meaning that 62% of the population falls into the S0 category and 38% falls into S1. With the exception of a pale tongue, the probability of having other symptoms and signs was greater in the S1 population than in the S0 population. In the TCM interpretation, the Z2 population was characterized by both lung and spleen deficiencies, with pathognomonic elements involving Qi and Yin deficiencies. S0 represents individuals with milder symptoms, leaning toward spleen-Qi deficiency, whereas S1 represents those with more severe symptoms, indicating a combination of lung-spleen Qi and Yin deficiencies. Similarly, the remaining integrated clustering models were sequentially analyzed for data analysis and TCM interpretation.

Category Z1: Z1 was characterized by 16 symptoms and signs, including greasy mouth without thirst, dry and uncomfortable eyes, loose stools, amnesia, etc. The probability of these symptoms and signs in the Z1 = S1 group (P = 0.59) was greater than that in the Z1 = S0 group (P = 0.41). These findings suggest that the Z1 population corresponds to damp-heat condensation syndrome, with milder symptoms in the S0 population and more severe symptoms in the S1 population. Category Z3: 10 symptoms and signs adequately described the characteristics of Z3, including squamous and dry skin, plump tongue, blue or purplish tongue, weak pulse, etc. Except for a pink tongue, the probabilities of all these symptoms and signs were greater in the Z3 = S1 (P = 0.46) population than in the Z3 = S0 (P = 0.54) population. These findings suggested that the Z3 population exhibited blood deficiency and stasis syndrome. In the S0 group, symptoms were milder and leaned toward blood deficiency, whereas in the S1 group, symptoms were more severe, indicating a combination of blood deficiency and blood stasis. Category Z4: 8 symptoms and signs adequately characterized Z4, including night sweating, spontaneous sweating, dry and uncomfortable eyes, etc. The probability of these symptoms and signs was greater in the Z4 = S1 group (P = 0.36) than in the Z4 = S0 group (P = 0.64). This result indicated that the Z4 population experienced vigorous fire due to Yin deficiency syndrome. The S0 population presented mild Yin deficiency and fire exuberance syndrome, whereas the S1 population presented severe Yin deficiency and fire exuberance syndrome.

Category Z5: 8 symptoms and signs adequately characterized Z5, including sensitization, anxiety, emaciation, ect. The probabilities of obesity and a stringy pulse were greater in the Z5 = S0 group (P = 0.63), whereas the probabilities of sensitization, upset and irritability, sighing, and emaciation were greater in the Z5 = S1 group (P = 0.37). In TCM, the Z5 population is interpreted as having liver constraints and Qi stagnation syndrome. Owing to Qi blockage in the body, water and fluid metabolism are affected. Thus, the S0 group was more inclined toward Qi stagnation and dampness obstruction, whereas the S1 group exhibited more emotional abnormalities, leaning toward liver Qi stagnation. Both groups also tended toward internal heat.

Category Z6: 11 symptoms and signs adequately characterized Z6, including poor appetite, nausea and vomiting, cold limbs, etc. The probability of these symptoms and signs was greater in the Z6 = S1 group (P = 0.64) than in the Z6 = S0 group (P = 0.36). In the TCM interpretation, Z6 = S1 (P = 0.64) was interpreted as heart-spleen heat accumulation syndrome.

3.5. Establishing rules for quantitative identification

The comprehensive clustering results were reimported into the Lantern 5.0 software naive Bayes classifier to establish the quantitative identification rules for the Z1 to Z6 syndromes, as shown in Tables 3 to 8. The children's positive symptoms were scored according to the table, whereas negative symptoms were recorded as 0 points. A total score was then calculated on the basis of the groupings in the table. If the total score exceeded the identification threshold for a particular syndrome type, the chil d was categorized into that syndrome type. Combined with expert knowledge, the clinical information of 153 pediatric CRF patients was fully integrated for secondary identification. The study ultimately identified the CRF syndrome types in children, ranked in descending order as follows: blood deficiency and stasis syndrome (54.2%), vigorous fire due to Yin deficiency (47.7%), deficiency of both the lung and spleen syndrome (46.4%), liver constraint and Qi stagnation (42.4%), damp-heat condensation syndrome (41.8%), and heart-spleen heat accumulation syndrome (39.2%).

Table 3.

Rules for recognizing damp-heat condensation syndrome

No.	Symptom	Score	No.	Symptom	Score
1	Greasy mouth without thirst	3.6	9	Tastelessness without thirst	2.2
2	Dry and uncomfortable eyes	3.6	10	Bitter taste in mouth	2.1
3	Loose stools	3.1	11	Susceptibility	1.8
4	Amnesia	3.2	12	Polydipsia	1.8
5	Skin oiliness	2.9	13	Acne	1.7
6	Heavy body and drowsiness	2.8	14	Abdominal bloating	1.5
7	Dysuria	2.8	15	Hypersomnia	1.5
8	Deep-colored urine	3.4	16	Fear of cold	1.5

Notes: recognition threshold 18.4 points. Identification accuracy 95.4%.

Table 8.

Rules for recognizing heart-spleen heat accumulation syndrome

No.	Symptom	Score	No.	Symptom	Score
1	Poor appetite	4	7	Stink of excrement	2
2	Light sleep	3	8	Difficulty with sleep consolidation	2
3	Nausea and vomiting	3	9	Halitosis	2
4	Localized stabbing pain	2	10	Cold limbs	2
5	Difficulty initiating sleep	2	11	Sore in mouth	2
6	Sore throat	3

Notes: recognition threshold 14.5 points. Identification accuracy 90.8%.

Table 4.

Rules for recognizing deficiency of both the lung and spleen syndrome

No.	Symptom	Score	No.	Symptom	Score
1	Spiritlessness	5.5	6	Spontaneous sweating	2.2
2	Weakness	5.7	7	Shortness of breath	1.9
3	Timidity	2.9	8	Feverishness in palms and soles	1.5
4	Night sweating	2.4	9	Pale tongue	1
5	Hypothymic depression	2.1

Notes: recognition threshold 17.0 points. Identification accuracy 90.2%.

Table 5.

Rules for recognizing blood deficiency and stasis syndrome

No.	Symptom	Score	No.	Symptom	Score
1	Squamous and dry skin	4	6	Pale lip and nail colour	2
2	Plump tongue	3.6	7	Pink tongue	2
3	Blue or purplish tongue	4.3	8	Red tongue	1.6
4	Weak pulse	3.3	9	Slippery fur	1.7
5	Thready pulse	2.8	10	Greasy fur	1.1

Notes: recognition threshold 9.7 points. Identification accuracy 94.1%.

Table 6.

Rules for recognizing vigorous fire due to Yin deficiency syndrome

No.	Symptom	Score	No.	Symptom	Score
1	Night sweating	4	5	Dysuria	2.6
2	Spontaneous sweating	3.4	6	Feverishness in palms and soles	2.5
3	Dry and uncomfortable eyes	2.9	7	Amnesia	2.5
4	Greasy mouth without thirst	2.6

Notes: Recognition threshold 9.2 points. Identification accuracy 96.7%.

Table 7.

Rules for recognizing liver constraint and Qi stagnation syndrome

No.	Symptom	Score	No.	Symptom	Score
1	Sensitization	5.2	5	Irritability	2
2	Anxiety	4.2	6	Obesity	1.1
3	Emaciation	3.3	7	Skin dryness	1.1
4	Sighing	2.7	8	Stringy pulse	1

Notes: recognition threshold 4.9 points. Identification accuracy 92.8%.

3.6. Visualization analysis of CRF syndrome in children

To further investigate the dynamic evolution of TCM syndromes, Sankey diagram visualization was used to analyze the interplay between syndrome type, clinical stage, and treatment duration in patients with childhood lymphoma (Figure 3). The results revealed that TCM syndrome progression is correlated with lymphoma clinical stage and treatment duration. In particular, blood deficiency and stasis syndrome displayed the highest degree of association with advanced-stage (stage IV) lymphoma and the middle therapeutic period (4-6 months), as demonstrated by its most prominent flow trajectory in the Sankey diagram. In contrast, weaker associations were observed between syndromes and early-stage disease (stage I), initial treatment (0-3 months), and prolonged treatment (over 12 months). This may be attributed to the limited representation of these subgroups in the study cohort (stage I: 2.61%; 0-3 months: 7.84%; over 12 months: 4.58%).

Figure 3. Sankey diagram analysis of TCM syndrome-clinical stage-treatment duration correlations in children's CRF.

TCM: Traditional Chinese Medicine; CRF: cancer-related fatigue.

4. DISCUSSION

4.1. Pathogenesis

The pathogenesis of CRF is complex, with the current understanding suggesting that both central and peripheral mechanisms may be involved. The central mechanisms of fatigue include hypothalamic-pituitary-adrenal (HPA) axis disorders, cytokine dysregulation, circadian rhythm disturbances, 5-hydroxytryptamine dysregulation, and activation of vagal nerve conduction.^1,14 Peripheral fatigue is attributed primarily to metabolic dysregulation in muscle.^1,14

According to TCM, children with lymphoma suffer from persistent malignant tumor cell proliferation and adverse reactions to chemotherapeutic drugs. This leads to imbalances in Qi and the physiological functions of internal organs, as well as the accumulation of pathological products such as phlegm and blood stasis. These factors further exacerbate the dysfunction of the Qi and physiology of internal organs, creating a vicious cycle. Over time, this results in a deficiency of kidney essence. When kidney essence is deficient, the marrow sea becomes insufficient, resulting in a complex pathology characterized by the loss of both body and essence. This dysfunction is closely linked to dysregulation of the HPA axis, indicating that the core issue lies in the abnormal functioning of the energy regulation center, leading to disturbances in the stress response and energy metabolism. The concept is further supported by other theories. Mitochondrial dysfunction in myocytes aligns with the theory that the spleen governs muscles and is responsible for the transportation of water and grain essence, highlighting the role of energy production disorders in the pathogenesis of CRF. The clinical manifestations of CRF present multidimensional signs of deficiency, such as physical fatigue, cognitive impairment, and a depressed mood, which align with the concept of 'deficiency labor' disease in TCM, reflecting abnormalities in the functioning of various internal organs.²⁹ The incidence of CRF in children after cancer treatment can reach 61.7%.³⁰ However, clinical studies on this topic are limited, and no relevant syndrome-based research currently exists. The latent structure model, a variant of the Bayesian network, represents a novel method for syndrome analysis.¹⁹

In this study, we evaluated the symptoms and signs of CRF in patients with childhood lymphoma and identified 10 high-frequency symptoms. Using latent structural model analysis, we constructed a TCM syndrome model of CRF in childhood lymphoma patients. A total of 18 latent variables were identified, summarizing the pathogenic syndrome elements, including Qi deficiency, blood deficiency, Yin deficiency, Qi stagnation, blood stasis, dampness, excessive heat, and dyspepsia. Additionally, this study highlighted the involvement of five organs — the spleen, lungs, heart, liver, and kidneys — in pathogenesis.

Finally, through comprehensive cluster analysis, we identified 6 types of syndrome types for CRF in lymphoma childhood, including blood deficiency and stasis syndrome, vigorous fire due to Yin deficiency syndrome, deficiency of both the lung and spleen syndrome, liver constraint and Qi stagnation syndrome, damp-heat condensation syndrome, and heart-spleen heat accumulation syndrome. These syndrome types differ from the 6 major TCM syndrome types in the adult CRF guideline, which are deficiency of the kidney Yang, stagnation of the liver Qi, deficiency of the spleen and stomach, cold dampness disturbing the spleen, deficiency of the lung Qi, and deficiency of the spleen Qi. A cross-sectional study that included 200 adult patients with CRF classified them into six categories of Chinese medicine syndrome: deficiency of kidney Yang, stagnation of liver Qi, Yin deficiency of the spleen and stomach, cold dampness disturbing the spleen, deficiency of lung Qi, and deficiency of spleen Qi.³¹ This classification was adopted from the Clinical Practice Guidelines for Cancer-Related Fatigue in China (2021 edition).¹⁴ In contrast, our findings show that CRF in children has characteristics different from those in adults.

4.2. High-frequency signs and symptoms analysis

The top 10 high-frequency symptoms and signs in children with CRF were representative and reflective of various factors in TCM. These factors include the following: Qi and blood deficiencies (e.g., weakness, pale lips and nails), lung and spleen issues (e.g., spiritlessness, poor appetite), emotional and mental disturbances (e.g., upset and irritability), internal heat (e.g., deep-colored urine, constipation, craving for cold), stagnation of blood (e.g., cancer-related pain), and internal dampness (e.g., thick and greasy moss).These high-frequency symptoms and signs accounted for more than half of the cases, indicating a significant overlap in the distribution of syndrome elements for CRF in patients with childhood lymphoma. Interestingly, the most frequent symptom observed was yellow urine, rather than fatigue-related symptoms such as general weakness and mental fatigue. However, fatigue-related symptoms were still among the high-frequency findings, suggesting that CRF in patients with childhood lymphoma encompasses multidimensional fatigue. This observation may also be related to the child's level of cognition and ability to express their symptoms.

4.3. Syndrome element analysis

The distribution of syndrome elements in children with CRF was similar to that reported in previous studies in adults, showing the same deficiency elements of Qi deficiency, blood deficiency and Yin deficiency, the solid elements of Qi stagnation, blood stasis and dampness-heat, and the five visceral factors. However, Yang deficiency and cold-dampness syndrome are rarely observed in children.¹⁴

Additionally, some differences exist in the clinical manifestations of the same syndrome elements between children and adults. For example, in cases of blood deficiency, adults with CRF typically experience symptoms such as dizziness, palpitation, insomnia, and numbness, which are uncommon in children with CRF. In pediatric patients, we observed pale lips, pale nails, and a pale tongue, among other signs, highlighting the pediatric diagnostic emphasis on visual examination.

Moreover, in cases of Yin deficiency, adults typically present with symptoms such as dry mouth, dry throat, vertigo, tinnitus, blurred vision, soreness of loins, and tidal fever. In contrast, children with CRF were more prone to night sweats, heat in the palms and soles, and emaciation, reflecting their unique physiology, where “Yang is often in excess, while Yin is often insufficient”, making them more susceptible to abnormal sweating. Furthermore, according to TCM theory, chemotherapy is considered a “heat toxin” therapy that depletes Yin Qi (vital essence). This poses a particular risk to children whose Yin system is still developing, making them more vulnerable to damage to their immature Yin.

4.4. Syndrome analysis

4.4.1. Blood deficiency and stasis syndrome had the widest distribution

In children, the spleen is often insufficient, leading to a failure in transporting Qi and blood. If the disease severely affects the kidneys, the transformation functions of essence and blood become abnormal, causing blood to stagnate. This further disrupts transformation processes, creating a vicious cycle that ultimately leads to deficiency syndromes over time.

The latent structure model identified Qi deficiency, blood deficiency, and blood stasis as common syndrome elements in childhood lymphoma CRF. Among these syndromes, blood deficiency and stasis syndrome had the widest distribution, affecting 54.2% of the population. This finding aligns with the findings of Liu’s study³² on TCM syndrome distribution in children with solid tumors after chemotherapy. In addition to the identification of Qi and blood deficiency syndrome in CRF in adults,^14,33 children are more likely to consider the complex role of pathological factors that take into account Yin essence deficiency and blood stasis and obstruction.

4.4.2. Heart-spleen heat accumulation is a unique syndrome

Unlike in adults, heart-spleen heat accumulation syndrome was identified in 39.2% of children with CRF, indicating that it is a distinct syndrome type specific to pediatric CRF.¹⁴ Owing to children's special excess of Yang -biased physique, the condition tends to heat easily. After chemotherapy, the spleen can further weaken, impairing the ability to digest and transform food and water, leading to food accumulation. On the one hand, it affects the transformation of Qi and blood, which ultimately results in Qi and blood deficiency. Additionally, it impacts other organ functions, with excessive heart fire manifesting as mouth and tongue sores. Latent structure model analysis revealed that food accumulation and heat are common syndrome elements of CRF in children, representing unique pathological factors in children with this condition.

4.4.3. Children with tender Yin and tender Yang bodies were prone to fire hyperactivity because of Yin deficiency.

The study revealed that children with CRF exhibited signs of hyperactivity of fire because of Yin deficiency, such as night sweating, dysuria, and feverishness in the palms and soles. The main symptoms were dry eyes and amnesia, indicating a deficiency of liver blood and kidney essence. Compared with adult CRF,^14,33 children with CRF are more prone to injury to the tender Yin body, particularly the liver and kidney Yin, which is accompanied by a tendency to transform heat.

Chemotherapeutic drugs act as "heat toxins", interacting with children's physiological characteristics of "immature Yin and Yang" to form a unique pathological progression: heat toxin damage Yin; Yin deficient fails to restrain Yang; deficient fire arises internally, further exacerbating heat and Yin depletion. Clinically, this pathogenesis manifests as vigorous fire due to Yin deficiency syndrome (47.7% of the population), with high-frequency symptoms such as constipation (57.52%) and craving for cold drinks (58.82%), prominently reflecting the "hyperactive Yang and depleted Yin" pathological essence. The treatment for this syndrome may involve the use of animal-derived medicines to nourish liver blood and kidney essence, thereby replenishing Yin, subduing fire, and restoring genuine Yin.

4.4.4. Deficiency of both lung and spleen in line with the characteristics of pediatric constitution

In children, the lungs and spleen are often deficient, increasing susceptibility to disturbances caused by pathogenic Qi. Lung Qi deficiency can lead to shortness of breath, whereas spleen Qi deficiency may cause a heavy body and drowsiness. Deficiencies in both the lungs and spleen, especially Qi and Yin (46.4% of the population), are more common in the pediatric CRF population than in adults. Unlike the single syndrome of lung Qi deficiency or spleen Qi deficiency in adult CRF,¹⁴ the combined impact of lung and spleen deficiencies and Yin and fluid deficiencies should be taken into account when diagnosing CRF in children.

4.4.5. Liver depression and Qi stagnation syndrome indicates emotional problems in the child

Psychological factors are well-established risk factors for tumors,³⁴ and mood disturbances are frequently observed in children with CRF.³⁵ Unlike adult tumor patients who are prone to depression, high-frequency symptoms such as CRF in children with lymphoma are irritable.³⁶ The latent structure model analysis revealed that the pathogenic syndrome elements in children with CRF included Qi depression and heat syndrome, with the liver being a common pathogenic site. Specifically, liver Qi stagnation and fire syndrome are prevalent in childhood lymphoma patients with CRF, affecting 42.4% of cases, and heat signs are more pronounced in these patients than in adults with liver Qi stagnation.¹⁴ This suggests that children's emotional problems during illness and treatment (including treatment away from home, delayed schooling, etc.) should be taken seriously.

4.4.6. Damp-heat condensation syndrome is more common in children

In children, dysfunction of the lungs, spleen, and kidneys can disrupt water metabolism, leading to dampness and the generation of heat from Qi stagnation. This combination of dampness and heat depletes both Qi and Yin, resulting in damp-heat condensation syndrome. The high frequency of symptoms of deep-colored urine and greasy fur suggest that dampness syndrome and heat syndrome are common syndrome elements in childhood lymphoma patients with CRF. Latent structure model analysis revealed that damp-heat condensation syndrome was found in 41.8% of childhood lymphoma with CRF, which was more prevalent than in adults.¹⁴

4.5. Dynamic evolution of TCM syndrome in children’s CRF

The study utilized Sankey diagrams to visually analyze the dynamic evolution patterns of TCM syndromes in pediatric CRF. The results revealed that in the early stages of lymphoma (stages Ⅰ and Ⅱ), the distribution of syndromes was relatively balanced, reflecting the complex pathophysiological characteristics of a mixture of deficiency and excess. However, in the late stages of the disease (stages Ⅲ and Ⅳ), the manifestations of organ deficiency were particularly prominent, with blood deficiency and stasis syndrome accounting for 23.24% and 24.20%, respectively. The syndrome evolution fully corroborates the TCM theory that “prolonged illness often leads to deficiency.” This finding also suggests that for children with advanced lymphoma, CRF treatment should prioritize “tonifying the body and nourishing the root,” aiming to improve the fundamental pathological mechanism of Qi deficiency by regulating and tonifying Qi and blood.

From the perspective of treatment duration, during the early stages of lymphoma treatment (within 3 months), the various syndromes exhibited a relatively balanced distribution pattern, suggesting that CRF in children leads to dysfunction of multiple organ systems and multidimensional pathophysiological changes from the onset of the disease. As treatment duration increases, the distribution of syndromes significantly changes. Between 4 and 6 months, the percentage of patients with blood deficiency and stasis syndrome increased to 21.45%, indicating that chemotherapy drugs induced pathological alterations in the Yin blood layer. Between 7 and 12 months, vigorous fire due to Yin deficiency syndrome has the widest distribution (24.19%), reflecting the cumulative effects of prolonged treatment leading to “heat toxicity damaging Yin.” After more than 12 months of treatment, the main syndromes were deficiency of both the lung and spleen syndrome (22.73%) and liver constraint and Qi stagnation syndrome (22.73%), presenting the late-stage characteristics of “root deficiency and branch excess.” Among these, the long-term impairment of spleen and stomach function is particularly prominent, leading to “postnatal malnutrition”, dysregulation of visceral Qi movement, and ultimately malnourishment of both the body and spirit. This dynamic evolutionary syndrome suggests that the 4-6 month period is a critical window for intervening in blood deficiency and stasis obstruction. After 6 months, emphasis should be placed on nourishing Yin and reducing fire, whereas long-term treatment should focus on strengthening the spleen and tonifying Qi, as well as regulating Qi circulation.

4.6. Application prospects

In conclusion, we summarized the Chinese medical diagnosis of CRF in childhood lymphoma into the above 6 types of syndromes and explored their identification rules. Compared with the current clinical practice guidelines for Chinese medicine for CRF, we found that the distribution of CRF in children with lymphoma is different from that in adults. Compared with adults, CRF in children is also a mixed syndrome of deficiency and solidity, but the tendency and degree of deficiency of the Qi of each viscera are different.

Second, there are specific pathological factors in the pathology of childhood lymphoma CRF, such as the food stagnation and solid heat. On the basis of TCM theory and expert clinical experience, we recommend that treatment for pediatric CRF prioritizes the use of tonic herbs supplemented with herbs to aid in digestion, resolve stagnation, regulate Qi, clear heat, and normalize bowel movements. Given the dynamic nature of the disease and standardized oncological treatment, we advise combining lymphoma clinical staging and chemotherapy duration with TCM syndrome differentiation to optimize personalized therapeutic strategies.

To date, international database searches have revealed a scarce number of TCM studies for pediatric tumors. Research on CRF in children remains in the primary exploration stage. We provide syndrome-based scientific support for TCM research on pediatric CRF. By analyzing the data of TCM syndrome patterns in children with CRF using a latent structural model, we provide a relatively objective scientific research method for Chinese medicine syndrome research. Our findings may contribute to the development of a systematic and standardized TCM diagnostic and treatment model for children with CRF.

4.7. Limitations

The survey was conducted with data from children's hospitals in large cities in northern China. To minimize geographic or climatic variations in the classification of pediatric CRF patients, future research should consider multicenter studies across different regions. Additionally, the limited sample size of our study may not fully represent the broader landscape of pediatric tumor-related CRF. To address this limitation, it is essential to expand our sample size in future studies to better validate the consistency of the study model with actual clinical syndrome.

5. ACKNOWLEDGMENTS

We are grateful to the medical staff of the Lymphoma Group at Beijing Children's Hospital and the TCM Oncology Group for their data support and assistance.

6. SUPPORTING INFORMATION

Supporting data to this article can be found online at http://www.journaltcm.com.

Funding Statement

Supported by Preliminary Study on Tiered Healthcare Based on a Pediatric Collaborative Development Model (No. XTGL201914); Establishment and Application of a Standardized Diagnosis and Treatment System for Pediatric Lymphoma (No. XTZD20180204)