This study included all RCTs targeting DN patients without restrictions on the patient age, disease duration, ethnicity, or geographic location.

Patients in the treatment group received JLD in addition to the standard treatment regimen used for the control group. The pharmaceutical product known as JLD was manufactured by Shijiazhuang Yiling Pharmaceutical Co., Ltd. The national drug approval number is Z20050845. The specification of JLD is 9 g per sachet (finished product weight), and each sachet corresponds to a total crude drug dosage of 45 g (detailed crude drug composition of the 17 herbal ingredients is provided in Supplementary Material S1, in accordance with the Chinese Pharmacopoeia (2020 Edition, Volume I) and official technical data from the manufacturer). The administration method was as follows: one sachet was dissolved in hot water for oral administration, three times daily (tid), with an 8-week course of treatment. The dosage of the finished product is calculated at 0.45 g·kg^-1·d^-1, and that of the crude drug is 2.25 g·kg^-1·d^-1. These values are based on a commonly used average body weight of 60 kg for adult patients in clinical studies. The total dose for the 8-week course is 1512 g for the finished product and 7560 g for the crude drug.

The treatment regimen for the control group in this study was designed to provide comprehensive management, which incorporated symptom control, diabetes education, dietary modifications, exercise interventions, and blood glucose monitoring. Oral hypoglycemic therapy focused on selecting drugs with minimal nephrotoxicity, including metformin, SGLT-2 inhibitors, thiazolidinediones, and α-glucosidase inhibitors. Angiotensin-converting enzyme inhibitors (ACEIs) and angiotensin receptor blockers (ARBs) were used for antihypertensive therapy. Lipid-lowering therapy involved the use of fibrates.

There were 3 distinct outcome measures in this study: primary, secondary, and safety outcome measures.

Primary outcome measures: The primary outcomes consisted of the clinical efficacy rate, serum creatinine (Scr), blood urea nitrogen (BUN), and 24-hour urine protein quantification (24h-UTP). These measures were used to assess clinical improvements in the renal function of patients with DN.

Secondary outcome measures: The secondary outcomes included fasting blood glucose (FBG), 2-hour postprandial glucose (2hPG), glycated hemoglobin (HbA1c), total cholesterol (TC), triglycerides (TG), vascular endothelial growth factor (VEGF), insulin-like growth factor-1 (IGF-1), interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α), and high-sensitivity C-reactive protein (hs-CRP). These measures were used to comprehensively evaluate the therapeutic effects of JLD on blood glucose control, lipid profiles, inflammatory markers, and relevant growth factors.

Safety outcome measures: The safety profile of treatment regimens was evaluated by documenting all adverse events occurring during RCTs, including hypoglycemic reactions, gastrointestinal discomfort, and other drug-related adverse reactions.

All included studies in this study were RCTs focused on patients with DN.

Animal studies, case reports, meta-analyses, reviews, conference proceedings, and letters were excluded.

Patients receiving kidney dialysis and those with ambiguous or unclear diagnostic criteria were excluded from this study.

Other forms of TCM such as acupuncture, cupping, and acupressure patch therapy were excluded from this study.

Studies with incomplete or missing results were excluded from this study.

The clinical efficacy rate in this study included 8 RCTs with 886 patients (treatment group = 451 patients; control group = 435 patients). These 8 studies showed low heterogeneity (I² = 27%, P > 0.1). Therefore, a fixed-effects model was used to pool the effect sizes. The treatment group showed significantly superior clinical efficacy compared to the control group [RR = 1.30 (95% CI: 1.21, 1.39), P < 0.001]. The forest plot for clinical efficacy rate is shown in Figure 3.

The SCr analysis in this study included 11 RCTs with 1,179 patients (treatment group = 598 patients; control group = 581 patients). Since substantial heterogeneity was observed between the 11 studies (I² = 80%, P < 0.1), a random-effects model was used to pool the effect sizes. The SCr levels were significantly lower in the treatment group compared to the control group [SMD = -2.01 (95% CI: -2.33, -1.69), P < 0.001]. The forest plot for SCr is shown in Figure 4.

The subgroup analysis by intervention strategy showed that JLD monotherapy [SMD = -1.79 (95% CI: -2.04, -1.54)], JLD combined with Tongxinluo [SMD = -2.06 (95% CI: -2.66, -1.46)], and JLD combined with other conventional interventions [SMD = -2.29 (95% CI: -3.50, -1.09)] all significantly reduced SCr levels. Although numerically superior efficacy was observed in the combination therapy subgroups compared to monotherapy, the difference between subgroups was not statistically significant (Chi²=1.22, df=2, P = 0.54, I²=0%), which may be due to insufficient statistical power caused by the small sample size in the “JLD combined with other conventional interventions” subgroup. Stratified analysis by intervention duration revealed significantly superior outcomes in the >8 weeks subgroup [SMD = -2.24 (95% CI: -2.71, -1.77)] compared to the ≤8 weeks subgroup [SMD = -1.65 (95% CI: -1.93, -1.37)], suggesting that extending the JLD treatment cycle can sustainably reduce serum creatinine levels. The differences between subgroups were statistically significant (Chi²=4.53, df=1, P = 0.03, I²=77.9%). For stratification by disease duration, the treatment group demonstrated superior efficacy in the ≤10 years disease duration subgroup [SMD = -1.95 (95% CI: -2.22, -1.69)] compared to the >10 years disease duration subgroup [SMD = -1.44 (95% CI: -1.67, -1.22)], suggesting that JLD exhibits greater restorative potential for early renal impairment. Statistically significant differences were observed between the subgroups (Chi²=8.31, df=1, P = 0.004, I²=88%). Additionally, stratified by the control group’s background therapy, the JLD-containing experimental group significantly reduced SCr in both subgroups: the RAAS Blocker subgroup [SMD = -1.97 (95% CI: -2.46, -1.49), P < 0.001] and the Non-RAAS Blocker subgroup [SMD = -2.06 (95% CI: -2.52, -1.61), P < 0.001]. Notably, there was no statistically significant difference in SCr-lowering efficacy between these two subgroups (Chi²=0.07, df=1, P = 0.79, I²=0%), suggesting that the SCr-lowering efficacy of JLD is consistent regardless of whether the control group uses RAAS blockers or other background therapies. The forest plot for this subgroup analysis of SCr is presented in Supplementary Material S5.

The BUN analysis in this study included 11 RCTs with 1,119 patients (treatment group = 568 patients; control group = 551 patients). Since substantial heterogeneity was observed among the 11 studies (I² = 80%, P < 0.1), a random-effects model was used to pool the effect sizes. The treatment group showed significantly lower BUN levels compared to the control group [SMD = -0.79 (95% CI: -1.07, -0.52), P < 0.001]. The forest plot is shown in Figure 5.

The subgroup analysis by intervention strategy showed that JLD monotherapy [SMD = -0.45 (95% CI: -0.66, -0.24)], JLD combined with Tongxinluo [SMD = -0.53 (95% CI: -1.04, -0.01)], and JLD combined with other conventional interventions [SMD = -1.12 (95% CI: -1.47, -0.76)] all significantly reduced BUN levels. JLD combination therapy exhibited superior efficacy over monotherapy, and the difference between subgroups was statistically significant (Chi²=10.15, df=2, P = 0.006, I²=80.3%), indicating a synergistic effect of JLD combined with other interventions on BUN reduction. Stratified analysis by intervention duration revealed numerically better BUN-lowering outcomes in the ≤8 weeks subgroup [SMD = -0.85 (95% CI: -1.34, -0.37)] than in the >8 weeks subgroup [SMD = -0.76 (95% CI: -1.13, -0.40)]. This suggests JLD exerts a rapid regulatory effect on urea metabolism in the early phase, which subsequently stabilizes with prolonged treatment. Notably, the difference between subgroups was not statistically significant (Chi²=0.09, df=1, P = 0.77, I²=0%), possibly due to the small sample size in the stratified subgroups leading to insufficient statistical power. For stratification by disease duration, patients with disease duration >10 years [SMD = -1.02 (95% CI: -1.48, -0.57)] showed numerically superior therapeutic efficacy compared with those with disease duration ≤10 years [SMD = -0.82 (95% CI: -1.42, -0.22)], suggesting JLD may exert stronger regulatory effects on glomerular urea permeability in patients with long-term diabetic nephropathy (DN), thereby achieving more pronounced BUN reduction. However, no statistically significant difference was observed across subgroups (Chi²=0.29, df=1, P = 0.59, I²=0%). Additionally, stratified by the control group’s background therapy, the JLD-containing experimental group significantly reduced BUN in both subgroups: the RAAS Blocker subgroup [SMD = -0.77 (95% CI: -1.14, -0.39), P < 0.01] and the Non-RAAS Blocker subgroup [SMD = -0.83 (95% CI: -1.28, -0.38), P = 0.0003]. Notably, there was no statistically significant difference in BUN-lowering efficacy between these two subgroups (Chi²=0.05, df=1, P = 0.82, I²=0%), suggesting that the BUN-lowering efficacy of JLD is consistent regardless of whether the control group uses RAAS blockers or other background therapies.The forest plot for subgroup analyses of BUN is shown in Supplementary Material S6.

The 24h-UTP analysis in this study included 9 RCTs with 969 patients (treatment group = 493 patients; control group = 476 patients). Since substantial heterogeneity was observed between the 9 studies (I² = 89%, P < 0.1), a random-effects model was used to pool the effect sizes. The 24h-UTP levels were significantly lower in the treatment group compared to the control group [SMD = -1.44 (95% CI: -1.88, -1.00), P < 0.001]. The forest plot for 24h-UTP is shown in Figure 6.

The subgroup analysis further indicated that JLD monotherapy [SMD = -0.91 (95% CI: -1.13, -0.69)], JLD combined with Tongxinluo [SMD = -2.42 (95% CI: -3.88, -0.97)], and JLD combined with other conventional interventions [SMD = -1.07 (95% CI: -1.37, -0.76)] all significantly reduced 24h-UTP levels. Although JLD combination therapy was numerically superior to monotherapy, no statistically significant difference was observed across subgroups (Chi²=4.48, df=2, P = 0.11, I²=55.3%), which may be attributed to moderate heterogeneity among subgroups and insufficient statistical power. Stratified analysis by intervention duration showed that patients with treatment duration >8 weeks [SMD = -1.85 (95% CI: -2.67, -1.03)] had significantly better 24h-UTP-lowering outcomes than those with duration ≤8 weeks [SMD = -1.00 (95% CI: -1.21, -0.80)], suggesting that extending the JLD treatment cycle can sustainably reduce 24h-UTP levels. The difference between subgroups was marginally statistically significant (Chi²=3.84, df=1, P = 0.05, I²=73.9%). For stratification by disease duration, patients with disease duration >10 years [SMD = -2.12 (95% CI: -3.64, -0.61)] showed numerically superior efficacy compared with those with disease duration ≤10 years [SMD = -1.15 (95% CI: -1.52, -0.77)]. However, no statistically significant difference was observed between subgroups (Chi²=1.51, df=1, P = 0.22, I²=33.8%), implying that JLD may exert stronger regulatory effects on glomerular protein permeability in patients with long-term DN, thereby achieving more pronounced 24h-UTP reduction. Additionally, stratified by the control group’s background therapy, the JLD-containing experimental group significantly reduced 24h-UTP in both subgroups: the RAAS Blocker subgroup [SMD = -1.85 (95% CI: -2.67, -1.03), P < 0.001] and the Non-RAAS Blocker subgroup [SMD = -1.00 (95% CI: -1.20, -0.80), P < 0.001]. Notably, there was a marginally statistically significant difference in 24h-UTP-lowering efficacy between these two subgroups (Chi²=4.22, df=1, P = 0.05, I²=74.2%), suggesting that the 24h-UTP-lowering efficacy of JLD may have a slight difference depending on whether the control group uses RAAS blockers or other background therapies. The forest plot for subgroup analyses of 24h-UTP is shown in Supplementary Material S7.

Secondary outcomes were categorized into four domains (glucose metabolism, lipid metabolism, inflammatory factors, and growth factors) for analysis. Given the moderate to high heterogeneity observed across studies, random-effects models were consistently applied to pool effect sizes to account for inter-study variability. Detailed results are presented below:

For glucose metabolism assessment using FBG, 2hPG, and HbA1c, combined JLD therapy significantly reduced FBG levels [SMD = -0.63 (95% CI: -1.01, -0.24), I² = 83%], 2hPG levels [SMD = -0.71 (95% CI: -1.20, -0.23), I² = 89%], and HbA1c levels [SMD = -0.95 (95% CI: -1.55, -0.35), I² = 92%] in DN patients. In terms of lipid-lowering effects, combined JLD therapy significantly decreased TC levels [SMD = -0.91 (95% CI: -1.75, -0.08), I² = 92%], but no statistically significant reduction in TG was observed [SMD = -3.07 (95% CI: -6.06, 0.08), I² = 99%], as the 95% CI crossed 0. Regarding anti-inflammatory activity measured by TNF-α, hs-CRP, and IL-6, combined JLD therapy led to a marked reduction in TNF-α [SMD = -1.75 (95% CI: -2.37, -1.13), I² = 88%], hs-CRP [SMD = -2.48 (95% CI: -2.81, -2.15), I² = 54%], and IL-6 [SMD = -1.77 (95% CI: -2.46, -1.09), I² = 81%] levels. Additionally, analysis of growth factors (VEGF and IGF-1) showed that combined JLD therapy significantly downregulated VEGF [SMD = -1.50 (95% CI: -2.53, -0.48), I² = 95%] and IGF-1 [SMD = -0.59 (95% CI: -0.97, -0.21), I² = 74%] expression levels.

Overall, combined JLD therapy exerted significant beneficial effects on glucose metabolism regulation, inflammation suppression, and pro-fibrotic growth factor downregulation in DN patients, whereas its lipid-lowering efficacy was only evident for TC reduction.

Safety data were collected from 13 RCTs included in this meta-analysis, involving 1,333 participants (675 in the JLD combined therapy group and 658 in the control group). Among these studies, 10 explicitly reported adverse events (AEs), with a total of 31 mild AEs documented in Table 3. The overall incidence of AEs in the JLD group ranged from 3.64% to 10.00%, predominantly characterized by mild gastrointestinal reactions (62.5%, n=19/31), including abdominal distension, gastric discomfort, nausea, and mild diarrhea. Other infrequent AEs included hypoglycemia (12.5%, n=4/31), rash (10.0%, n=3/31), dizziness/headache (7.5%, n=2/31), and orthostatic hypotension (7.5%, n=2/31). Notably, no severe AEs (e.g., hepatotoxicity, nephrotoxicity, myelosuppression) were reported across all studies. Meta-analysis of AE rates showed no statistically significant difference between the JLD group and the control group [RR = 0.71 (95% CI: 0.42, 1.20), I² = 0%), indicating that JLD adjunctive therapy does not increase additional safety risks compared to conventional treatment alone. Most AEs were self-limiting or resolved spontaneously without specific interventions, further supporting the mild and reversible nature of JLD-related adverse reactions in clinical settings.

Inflammatory infiltration is a key driver of DN progression. Consistent with the “multiple components targeting multiple pathways” feature of JLD, the anti-inflammatory effect of the compound is achieved through coordinated regulation of multiple active ingredients on the NF-κB-related signaling cascade: ginsenoside Rg2 in Panax ginseng reduces excessive reactive oxygen species production, thereby inhibiting NF-κB/NLRP3 signaling pathway activation and blocking the inflammatory cascade and pyroptosis (29); brachangobinan A, a lignan from Perilla frutescens, directly inhibits NF-κB pathway activation and reduces pro-inflammatory factors (31); Pueraria root extract PTY-2r normalizes inflammatory factor and inducible nitric oxide synthase expression in renal tissue by downregulating PKC-α and NF-κB pathways (32). Furthermore, Polygonatum polysaccharide modulates inflammatory factor transcription by promoting IκB-α degradation and NF-κB p65 nuclear translocation, further reinforcing the compound’s anti-inflammatory network (33). This multi-component targeted regulation of inflammation lays the foundation for subsequent attenuation of DN progression.

Renal fibrosis represents the core pathological alteration in end-stage DN. JLD targets key fibrosis-related pathways such as the TGF-β/Smad through multiple active components: cycloartane-type triterpenoid glycosides and aminoguanidine in Cornus officinalis reduce serum TGF-β1 protein and glomerular TGF-β1 mRNA expression, decreasing abnormal deposition of fibronectin and laminin in renal tissue (30); Sophora flavescens inhibits the activation of TGF-β/Smad pathway to reduce extracellular matrix-associated protein accumulation (34); saponins from Semen Litchi lower TGF-β1 and fibronectin levels while suppressing proliferation of human glomerular mesangial cells (35); 2,3,5,4’-tetrahydroxystilbene-2-O-β-D-glucoside in Polygonum multiflorum extract reduces VEGF expression in renal tissue, alleviates oxidative stress damage in renal vascular endothelium and proteinuria, mitigates glomerulosclerosis and interstitial fibrosis, and delays the progression of renal insufficiency (36). Collectively, these components form a multi-component anti-fibrotic regulatory network to attenuate the progression of DN to end-stage renal disease.

Oxidative stress-induced podocyte injury is a major contributor to DN proteinuria. Rehmannia glutinosa water extract alleviates renal pathological damage in DN model rats by enhancing antioxidant capacity and reducing inflammatory factor levels (37); meanwhile, Poria cocos and its outer layer components enhance antioxidant defense systems by regulating the Keap1/Nrf2 pathway (38); specifically, Tanshinone IIA targets the LAVL1-ACSL4 axis to mitigate high-glucose-induced damage and ferroptosis in MPC5 podocytes, collectively safeguarding podocyte integrity and renal function (39). Protection of podocytes and attenuation of oxidative stress are critical for reducing proteinuria and preserving renal function in DN.

Metabolic disorders serve as the initiating factor for DN. Polysaccharides from Coptis chinensis improve pancreatic and hepatic morphological abnormalities, elevate serum insulin levels, reduce FBG and glycated serum protein, inhibit AGE accumulation, and concurrently enhance renal function (40); meanwhile, Timosaponin B-II from Anemarrhena asphodeloides exerts dual hypoglycemic and anti-inflammatory effects by regulating the TXNIP, mTOR, and NF-κB pathways, thereby ameliorating alloxan-induced DN (41); furthermore, extracts from Lycium chinense indirectly alleviate renal metabolic burden by reducing hepatic lipid accumulation and white adipose tissue mass, improving adipocyte hypertrophy (42). Collectively, these components act in a coordinated manner to correct metabolic disorders and block AGE-mediated renal damage, reinforcing the multi-target therapeutic advantage of JLD.

JLD components block renal tissue epithelial-mesenchymal transition (EMT) and cellular injury by regulating key signaling pathways: Icariin from Epimedium brevicornu inhibits the PI3K/AKT pathway, blocking high-glucose-induced EMT in HK-2 human renal tubular epithelial cells (43); specifically, extracts from Ophiopogon japonicus mitigate renal tissue damage in DN model rats by modulating the EGFR/PI3K/AKT pathway (44); additionally, Atractylodes macrocephala directly improves renal function in DN patients through its anti-inflammatory and antioxidant properties, which in turn indirectly suppress renal epithelial cell EMT progression (45). Collectively, these components form a multi-component regulatory network targeting diverse signaling pathways, further reinforcing JLD’s multi-target therapeutic efficacy against DN progression.