$$\rightleftharpoonup{xx}$$
$$\longleftharp{xx}$$,
$$\longrightharp{xx}$$,
Clinical study section
General information
The results indicated that the observation group consisted of 30 patients, comprising 16 men and 14 women, averaging 59.10 ± 6.56 years of age; the control group also consisted of 30 patients, with 17 men and 13 women, averaging 58.97 ± 6.22 years of age. No statistically significant differences were observed between the two groups in terms of gender, age or duration of illness (p > 0.05), as shown in Table 1.
Comparison of white blood cell, neutrophil, haemoglobin, and platelet levels post-treatment
Univariate repeated measures ANOVA was employed to explore the impact of different treatment methods on WBC, NEU, Hb, and PLT levels within 20 days. The Shapiro-Wilk test indicated that the data in each group followed an approximately normal distribution (p > 0.05). Mauchly's test of sphericity confirmed that the variance-covariance matrices of each group were equal (p > 0.05). Data were presented as x̄ ± s, as shown in Table 2. The results are summarised below.
The interaction effects of time × treatment on Hb and PLT levels were significant (FHb interaction = 13.572, FPLT interaction = 11.985, both p < 0.0001), indicating that the effects of different treatment methods on Hb and PLT levels at four time points varied. Additionally, WBC, NEU, Hb, and PLT levels in both groups changed over time (FWBC time = 46.612, FNEU time =3.398, FHb time =8.764, FPLT time = 30.168, all p < 0.0001). Finally, different treatment methods had varying impacts on WBC, NEU, Hb, and PLT levels (FWBC treatment = 12.416, FHb treatment = 92.43, FPLT treatment =57.672, all p < 0.0001). Further comparison of WBC, NEU, Hb, and PLT levels at 4, 8, 12, and 20 days post-treatment revealed that the observation group had higher WBC and Hb levels at 12 and 20 days, higher NEU levels at 12 days, and higher PLT levels at 8 and 12 days than the control group (p < 0.05).
Comparison of post-treatment data
The findings revealed that the onset of grade III-IV myelosuppression in the observation group occurred later than in the control group (5.63 ± 1.10 vs. 4.10 ± 1.24 days), with a shorter duration (7.07 ± 1.72 vs. 9.97 ± 1.16 days) and a faster recovery time (12.17 ± 0.20 vs. 15.17±1.12 days), all with p < 0.05 (Table 3).
Comparison of quality-of-life brief version scores between groups
Similarly, using one-way repeated measures ANOVA, the impact of different treatment modalities on patients' WHOQOL-BREF scores within 2 weeks was explored. The Shapiro-Wilk test confirmed that the data for each group approximated a normal distribution (p > 0.05); Mauchly's test of sphericity indicated homogeneity of variance-covariance matrices across groups (p > 0.05). Data were expressed as x ± s, as shown in Supplementary Table 1. The results are summarised below.
The interaction between time and treatment on WHOQOL-BREF scores was significant (FQOL-BREF interaction = 137.262, p < 0.0001), indicating that the magnitude of the effect of different treatments on WHOQOL-BREF scores varied across three time points. Moreover, WHOQOL-BREF scores in both groups changed over time (FQOL-BREF time = 32.848, p < 0.0001). Finally, the impact of different treatment modalities on WHOQOL-BREF scores varied (FQOL-BREF treatment =91.908, p < 0.0001). Further comparison of WHOQOL-BREF scores at weeks 1 and 2 showed that the observation group had higher scores than the control group at both time points (p < 0.05).
Comparison of liver and kidney function indicators post-treatment
Post-treatment comparison of ALT, AST, CREA, Na+, and K+ levels between the two groups showed no statistically significant differences (all p > 0.05), indicating no differential impact of treatment modalities on liver and kidney function indicators between the groups, as shown in Supplementary Table 2.
Comparison of adverse reactions post-treatment
The observation group experienced four cases of liver function impairment, two cases of kidney function impairment, and five cases of electrolyte imbalance, with an adverse reaction rate of 36.7%. The control group experienced six cases of liver function impairment, one case of kidney function impairment, seven cases of electrolyte imbalance, and one case of electrocardiograph abnormalities, with an adverse reaction rate of 50.0%. The difference in adverse reaction rates between the two groups was not statistically significant (χ2 = 1.086, p = 0.297), as shown in Supplementary Table 3.
Network pharmacology section
Acquisition and screening of compound components
A total of 400 compound components of PSD were retrieved from the TCMSP and ETCM databases. Following a screening process using criteria of oral bioavailability ≥30% and drug likeness ≥0.18, key effective components were identified. The Swiss Target Prediction platform was then utilized to predict the targets of the potential chemical components of each herb, selecting 'Homo sapiens' as the species. Using the median probability value of 0.105 as a threshold, the targets for the potential active ingredients were screened. The final quantities of potential active ingredients obtained for each herb were as follows: 10 for peanut skin, 15 for Astragalus mongholicus, 5 for Hairyvein Agrimony, 12 for Spatholobus suberectus, 6 for Codonopsis pilosula, 8 for Donkey-hide gelatin, 6 for Atractylodes macrocephala, 4 for Poria, 6 for prepared liquorice root, 6 for Angelica sinensis, 11 for Forsythia suspensa, 47 for Malaytea Scurfpea Fruit, 11 for Ligustrum, 10 for Eclipta, and 9 for Jujube.
Prediction of compound targets and disease-related targets
After consolidation and deduplication of the potential active ingredients, a total of 113 compound-related targets were obtained from the TCMSP and Swiss databases. In the GeneCards and OMIM databases, 7,649 disease-related targets were retrieved, and 2,946 disease targets were filtered using a 'score ≥4.0' criterion. Based on the obtained targets, to study the interactions, BPs, and pathways of the targets, an intersection of 113 compound targets and 2,946 myelosuppression-related targets was identified, 62 intersecting targets, as illustrated in Figure 1A.
Construction of active ingredient-target network
Utilising the aforementioned analysis, active ingredients and potential targets were imported into Cytoscape 3.8.2 software to construct an active ingredient-target network, as shown in Figure 1B. The network comprised 131 nodes and 248 edges. Arrows denote the active ingredients of PSD, ellipses represent potential targets, and the lines between nodes represent their corresponding relationships. Each active ingredient can act on multiple potential targets, reflecting the multi-component, multi-target effects of PSD.
Gene ontology analysis
Gene ontology analysis primarily includes BPs, CCs, and MFs, with BPs being the most critical. To investigate the BPs of PSD in treating myelosuppression, 62 potential targets were imported into the DAVID database for GO analysis. A total of 503 GO terms were obtained, including 267 BP terms, 45 CC terms, and 84 MF terms. Using −Log10p as the screening standard, the top 10 BP, CC, and MF GO terms were selected and analysed, as depicted in Figure 2A. The length of the bar represents the number of genes enriched in the GO term, and the colour represents the −Log10p value. The top five predicted BP terms were 'positive regulation of ERK1 and ERK2 cascade', 'positive regulation of transcription from RNA polymerase II promoter', 'positive regulation of nitric oxide biosynthetic process', 'negative regulation of apoptotic process' and 'positive regulation of cell proliferation'. The top five CCs were 'plasma membrane', 'perinuclear region of cytoplasm', 'cytoplasm', 'cell surface' and 'extracellular region'. The top five MFs were 'enzyme binding', 'protein binding', 'RNA polymerase II transcription factor activity, ligand-activated sequence-specific DNA binding', 'nitric-oxide synthase regulator activity' and 'glycoprotein binding'.
Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis
The 62 potential targets were imported into the DAVID database for KEGG pathway enrichment analysis related to the treatment of myelosuppression with PSD. A total of 111 significantly enriched KEGG pathways were obtained. Using the p-value as the criterion, the top 20 signalling pathways were selected and analysed, as illustrated in Figure 2B. The horizontal axis represents the ratio of enriched genes to all genes in the pathway, the colour of the circles represents the −Log10p value, and the size of the circles represents the number of enriched genes. Pathways related to myelosuppression were mainly enriched in the phosphoinositide 3-kinase-protein kinase B (PI3K-Akt) signalling pathway, hypoxia-inducible factor (HIF)-1 signalling pathway, vascular endothelial growth factor (VEGF) signalling pathway, and oestrogen pathway.
Protein-protein interaction network construction
To predict inter-target relationships and identify core targets, 62 potential targets were imported into the STRING database to construct a PPI network, which was then visualised as shown in Figure 3A. This PPI network comprises 58 nodes (with 4 target proteins not participating) and 374 interaction lines. Nodes correspond to the proteins associated with potential targets, with the size of a node indicating the degree value of that target protein - the larger the node is, the greater its degree value within the network. Lines between nodes signify potential interaction relationships between target proteins. The network underwent cluster analysis using the MCODE plugin in Cytoscape 3.8.2, resulting in the identification of a core subnetwork with the highest score, as depicted in Figure 3B. This subnetwork, scoring 15.29, involves 18 proteins deemed to have significant roles within the PPI. Further analysis using the Cytohubba plugin led to the selection of hub proteins, with the interaction diagram presented in Figure 3C. Darker-coloured proteins indicate stronger interactions, whereas lighter-coloured ones suggest weaker interactions. The intersection of proteins with higher scores from six algorithms identified tumour necrosis factor (TNF), interleukin (IL)-6, VEGF-A, steroid receptor coactivator (SRC), Harvey rat sarcoma viral oncogene homolog (HRAS), and signal transducer and activator of transcription 3 (STAT3) as the hub proteins.
Molecular docking
The IGEMDock v2.1 software package was utilised for molecular docking between the six hub proteins (IL-6, TNF, SRC, VEGF-A, STAT3, HRAS) selected through network pharmacology and 28 active components involved in the network pharmacology construction. Four compounds, Campneoside II, Darendoside B, Leucosceptoside A, and Purpureaside C, demonstrated stable binding with all six hub proteins, as indicated in Supplementary Table 4. This suggests their potential as pharmacologically active components in modified PSD for treating myelosuppression.
DATA AVAILABILITY:
The raw data is provided in Supplementary File 1.

Figure 1: Network pharmacology analysis of modified peanut skin decoction. (A) Venn diagram of component-related targets and disease-related targets. (B) Network diagram of active ingredients and myelosuppression targets (HSYD: Modified peanut skin decoction). Please click here to view a larger version of this figure.

Figure 2: Functional enrichment analysis of modified peanut skin decoction. (A) Gene oncology analysis diagram of modified peanut coat decoction for treating myelosuppression. (B) Kyoto encyclopedia of genes and genomes pathway enrichment analysis diagram of modified peanut skin decoction for treating myelosuppression. Please click here to view a larger version of this figure.

Figure 3: Protein-protein interaction network of key targets. (A) Protein-protein interaction diagram of potential action targets. (B) Core subnetwork with the strongest interactions among potential action targets. (C) Protein-protein interaction diagram of hub proteins selected by six algorithms. Please click here to view a larger version of this figure.
Table 1: Comparison of general data. Please click here to download this Table.
Table 2: Comparison of whiteblood cell, neutrophil, haemoglobin, and platelet levels between the two groups post-treatment. Please click here to download this Table.
Table 3: Comparison of post-treatment data. Please click here to download this Table.
Supplementary Table 1: Comparison of World Health Organization Quality of Life Brief Version scores between groups. Please click here to download this File.
Supplementary Table 2: Comparison of liver and kidney function-related indexes post-treatment. Please click here to download this File.
Supplementary Table 3: Comparison of adverse reactions post-treatment between groups (cases [%]). Please click here to download this File.
Supplementary Table 4: Molecular docking results. Please click here to download this File.