This study included 25 patients with SLE treated with Telitacicept at the Second Xiangya Hospital of Central South University, from December 31, 2022, to May 31, 2024, based on the 1997 American College of Rheumatology (ACR) classification criteria (14). The specified inclusion and exclusion criteria were list in Supplementary Table 1. This study was conducted in accordance with the Declaration of Helsinki (15) and was approved by the Clinical Research Ethics Committee of Second Xiangya Hospital of Central South University (Ethics Approval Number: LYF2022151). All SLE patients signed informed consent forms prior to treatment.

This study references the Phase III clinical trial protocol for Telitacicept in China (NCT06456567). Patients received standard therapy plus regular administration of 80 mg Telitacicept subcutaneous injection once week for 24 weeks. Clinical data collection occurred at Week 0 (baseline) and at the end of Week 24 of treatment, and serum samples were also collected for proteomics and metabolomics sequencing. The standard treatment regimen involves the stable use of any two of the following: glucocorticoids (methylprednisolone or prednisone), antimalarials (hydroxychloroquine), and immunosuppressants (including mycophenolate mofetil, tacrolimus, cyclosporine, azathioprine, cyclophosphamide, methotrexate, leflunomide, etc.).

According to the primary efficacy endpoint in the Phase III clinical trial protocol for Telitacicept, this study adopted SRI-4 as the efficacy assessment standard (16). SRI-4 comprises three core components: the SLEDAI score (17), BILAG score (18), and PGA score (19). SRI-4 response criteria are defined as follows: (1) A decrease in SLEDAI score of ≥4 points; (2) No new organs reaching grade A on the BILAG score, or fewer than 2 new organs reaching grade B; (3) An increase in PGA score of no more than 0.3 points compared to baseline. According to the SRI-4 response criteria, SLE patients fulfilling the criteria are categorized as the responder (R) group, while those failing to meet the criteria are classified as the non-responder (NR) group. For pre-post treatment comparisons, subgroups are defined as follows: R group includes Pre-treatment responder (Pre-R) group and Post-treatment responder (Post-R) group; NR group comprises Pre-treatment non-responder (Pre-NR) group and Post-treatment non-responder (Post-NR) group.

Clinical data of SLE patients were collected from the outpatient system of Second Xiangya Hospital of Central South University. Statistical analysis was performed using SPSS software (IBM SPSS Statistics for Windows, Version 26.0). The pairwise t-test was used for normally distributed variables; otherwise, the Wilcoxon signed-rank test was applied. For all quantitative data, descriptive statistics were presented as mean ± standard deviation (X ± SD) for normally distribution, or as median and interquartile range [M (P25, P75)] for skewed distribution.

The proteomics analysis primarily includes protein extraction, peptide digestion, liquid chromatography-tandem mass spectrometry (LC-MS/MS) data acquisition, and database searching (20). The peptides were analyzed by LC-MS/MS in data independent acquisition (DIA) mode. The DIA data were processed using DIA-NN software with the following parameter settings: enzyme of trypsin, max miss cleavage site of 1, fixed modification of Carbamidomethyl(C), and the dynamic modifications of Oxidation(M) and Acetyl(Protein N-term). Subsequently, the mass spectrometry results underwent bioinformatics analysis, primarily comprising differential expression analysis and functional analysis. The pairwise t-test was used to compare the differences in protein abundance before and after treatment (FC >1.5 or <0.67, and P value <0.05) (21, 22). Benjamini and Hochberg procedure was applied to adjust for multiple hypothesis testing. The functional analysis included subcellular localization analysis, domain analysis, and transcription factor analysis. Finally, key differentially expressed proteins associated with therapeutic efficacy were screened, followed by receiver operating characteristic (ROC) curve analysis, area under the curve (AUC) calculation, and correlation analysis, exploring the association between the expression of these key proteins and clinical indicators, as well as their biological significance.

The metabolomics analysis primarily involves metabolite extraction and chromatography-mass spectrometry analysis. Serum samples were subjected to chromatographic analysis using Vanquish LC ultra-high performance liquid chromatography (UHPLC) system, followed by mass spectrometry using an Orbitrap Exploris™ 480 mass spectrometer. Detection was performed in both positive ion mode (POS) and negative ion mode (NEG) using electrospray ionization. After the raw data were converted to mzXML format by ProteoWizard, the XCMS package was used for peak alignment, retention time correction, and peak area extraction, followed by metabolites identification, data preprocessing, and data quality evaluation. Subsequently, the mass spectrometry results underwent bioinformatics analysis, primarily comprising differential expression analysis and pathway enrichment analysis. The significantly differential metabolites were determined based on the VIP obtained by the orthogonal partial least squares discriminate analysis (OPLS-DA) model and the P value of pairwise t test. Benjamini and Hochberg procedure was applied to adjust for multiple hypothesis testing. VIP >1 and P value <0.05 were considered significantly differential metabolites (23, 24). The pathway enrichment analysis included Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis (25) and metabolite set enrichment analysis (MSEA). KEGG enrichment analysis is used to explore the primary metabolic pathways within the target metabolic set, but it has certain limitations. MSEA can compensate for these shortcomings by identifying metabolites and their pathways that exhibit low abundance changes yet remain significant for disease regulation. Finally, key differentially expressed metabolites suggesting therapeutic efficacy are screened, followed by ROC analysis and correlation analysis to investigate the association between the expression of these key metabolites and clinical indicators, as well as their biological significance.

KEGG pathway analysis and correlation analysis were performed on differentially expressed proteins and metabolites. Pathway analysis, based on KEGG annotation, utilized venn diagrams to illustrate pathways jointly involved in both proteins and metabolites. Correlation analysis involved log₂ normalization of quantitative data from differentially expressed proteins and metabolites and construction of pearson correlation-based hierarchical clustering heatmaps using the pheatmap R package (Version 1.0.12).

After 24 weeks of treatment, 20 patients achieved an SRI-4 response according to the SRI-4 criteria, while 5 patients did not meet the criteria, resulting in an SRI-4 response rate of 80%. Patient demographics, including organ involvement and laboratory test results, were summarized in Table 1. Comparisons between pre- and post-treatment in the R group of SLE patients revealed statistically significant changes in several clinical indicators, including SLEDAI scores, PGA scores, anti-dsDNA antibodies, albumin, globulin, urinary albumin, total urinary protein, urinary albumin/creatinine ratio, total urinary protein/creatinine ratio, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), C3, and IgG. In the comparison of the NR group, only the change in CRP levels showed statistical significance in the clinical indicators. However, the CRP levels increased after treatment, consistent with the outcome of treatment failure. These findings indicate that Telitacicept can effectively improve clinical symptoms and laboratory data in the majority of SLE patients.

With FC > 1.5 or < 0.67 and P value < 0.05, 218 (46 upregulated and 172 downregulated) and 163 (147 upregulated and 16 downregulated) differentially expressed proteins were identified in the Pre-R vs Post-R group and Pre-NR vs Post-NR group, respectively (Supplementary Table 2). To identify key proteins indicative of therapeutic efficacy, 172 proteins lowly expressed in Post-R were intersected with 147 proteins highly expressed in Post-NR, and 46 proteins highly expressed in Post-R were intersected with 16 proteins lowly expressed in Post-NR (Figure 2A). Results indicate that only six common proteins were identified: Q9NQH7 (XPNPEP3), Q13243 (SRSF5), Q13247 (SRSF6), P23381 (WARS1), O75874 (IDH1), and Q8WWA0 (ITLN1). As shown in Figures 2B, C, these six proteins were downregulated in the R group and upregulated in the NR group post-treatment.

An in-depth analysis and investigation into the functions of the six key differentially expressed proteins was performed. As shown in Table 2, subcellular localization analysis revealed that Q9NQH7 (XPNPEP3) is present not only in the nucleus but also in mitochondria. Q13243 (SRSF5) and Q13247 (SRSF6) are primarily localized to the nucleus, while P23381 (WARS1) and O75874 (IDH1) are mainly found in the cytoplasm. Q8WWA0 (ITLN1) is predominantly localized to the extracellular. In domain analysis, Q9NQH7 (XPNPEP3) contains the Aminopeptidase P, N-terminal domain (PF05195) and the Metallopeptidase family M24 (PF00557). Q13243 (SRSF5) and Q13247 (SRSF6) contain the RNA recognition motif (PF00076), P23381 (WARS1) contains the tRNA synthetases class I (PF00579) and WHEP-TRS domains (PF00458), O75874 (IDH1) contains an isocitrate/isomethylmalate dehydrogenas (PF00180), and Q8WWA0 (ITLN1) has no identified domains. None of these six proteins exhibit transcription factor functions.

The ROC analysis indicated that the AUC values for the six proteins—Q9NQH7 (XPNPEP3), Q13243 (SRSF5), Q13247 (SRSF6), P23381 (WARS1), O75874 (IDH1), and Q8WWA0 (ITLN1)—were 0.891, 0.802, 0.800, 0.765, 0.755, and 0.667, respectively, with the first three proteins demonstrating higher diagnostic value (Figure 3A). Due to the absence of internal validation, these ROC curves represent internal, exploratory performance. We conducted a preliminary correlations analysis between these six proteins and clinical indicators associated with treatment efficacy (SLEDAI, PGA, ESR, C3, and IgG) prior to treatment in the R group (Figure 3B). Q9NQH7 (XPNPEP3) was positively correlated with PGA; Q13243 (SRSF5) positively correlated with SLEDAI, PGA, and ESR; Q13247 (SRSF6) positively correlated with PGA; P23381 (WARS1) showed negative correlation with complement C3 and positive correlation with IgG; O75874 (IDH1) showed negative correlation with complement C3; Q8WWA0 (ITLN1) showed positive correlation with SLEDAI and ESR, and negative correlation with complement C3. These findings indicate that these proteins may predict treatment efficacy.

With OPLS-DA VIP > 1, and P value < 0.05, significantly differentially expressed metabolites were screened (Supplementary Table 3). In the POS mode, 35 significantly differentially expressed metabolites were identified in the Pre-R vs Post-R group, and 4 in the Pre-NR vs Post-NR group. In the NEG mode, 17 significantly differentially expressed metabolites were identified in the Pre-R vs Post-R group, and 12 were identified in the Pre-NR vs Post-NR group.

Using between-group comparisons, key differentially expressed metabolites were identified between R and NR groups in both POS and NEG modes. However, no metabolites were found to be common to both modes. Therefore, we shifted our approach to jointly screen for key differentially expressed metabolites by integrating KEGG pathway analysis results with MSEA results. In the KEGG enrichment analysis, 8 pathways were significantly enriched in the Pre-R vs Post-R group, and 8 pathways were significantly enriched in the Pre-NR vs Post-NR group (P value < 0.05 & |DA Score|≥1, Figure 4A). The MSEA results revealed 33 pathways significantly enriched in the Pre-R vs Post-R group and 9 pathways significantly enriched in the Pre-NR vs Post-NR group (P value < 0.05 & Enrichment ratio > 2, Figure 4B). The venn diagrams indicated that five pathways were significantly enriched in both the Pre-R vs Post-R group and the Pre-NR vs Post-NR group (Figure 4C). As shown in Table 3, pyruvate participated in all five pathways in the Pre-R vs Post-R group, and gamma-aminobutyric acid (GABA) similarly participated in all five enriched pathways in the Pre-NR vs Post-NR group. This suggests that these two metabolites may play a crucial role in the therapeutic process of Telitacicept for SLE and could serve as key differentially expressed metabolites.

The box plots exhibited that both pyruvate (P = 0.00039) and GABA (P = 0.01136) levels decreased post-treatment, with the change in pyruvate exhibiting high statistical significance (Figures 5A, B). To evaluate the value of pyruvate and GABA in clinical therapeutic efficacy, ROC analysis was conducted (Figures 5C, D). The AUC value of pyruvate in the Pre-R vs Post-R group was 0.845, and the AUC value of GABA in the Pre-NR vs Post-NR group was 0.920, suggesting potential diagnostic utility that required validation in independent cohorts to confirm generalizability. We conducted a preliminary analysis of the correlation between these two metabolites and pre-treatment clinical efficacy indicators (SLEDAI, PGA, ESR, C3, and IgG). The results revealed no significant correlations, with all P values exceeding 0.05. This may be attributed to sample size limitations, as the NR group contained only five samples. Second, metabolite levels are susceptible to significant variability influenced by individual differences, diet, medication, and other factors, potentially obscuring their relationship with clinical indicators. NR-associated candidate biomarkers identified through exploratory analysis require validation in larger cohorts to confirm their potential diagnostic utility, given the inherent limitations of small sample size.

The KEGG enrichment analysis indicated that 219 and 54 pathways were enriched by differentially expressed proteins and metabolites in the Pre-R vs Post-R group, respectively. In the Pre-NR vs Post-NR group, 265 and 35 pathways were enriched by differentially expressed proteins and metabolites, respectively. The venn diagrams revealed 25 pathways jointly involved by differentially expressed proteins and metabolites in the Pre-R vs Post-R group, and 26 pathways shared by differentially expressed proteins and metabolites in the Pre-NR vs Post-NR group (Figure 6A). We screened for co-occurring metabolic pathways among 6 key proteins (XPNPEP3, SRSF5, SRSF6, WARS1, IDH1, and ITLN1) and 2 key metabolites (pyruvate and GABA). In the Pre-R vs Post-R group, protein IDH1 and metabolite pyruvate were found to share two metabolic pathways: central carbon metabolism in cancer (hsa05230) and the citric acid cycle (hsa00020). In the Pre-NR vs Post-NR group, no shared metabolic pathways were identified. In addition, the correlation analysis between 6 key proteins and 2 key metabolites were performed, indicating that all six proteins exhibited positive correlations with pyruvate and negative correlations with GABA (Figure 6B). Among these, XPNPEP3 and GABA (r = −0.82) and ITLN1 and GABA (r = −0.63) demonstrated relatively high negative correlation coefficients. However, their statistical significance was not directly confirmed in this study and requires further validation with increased sample size.