Intra-articular platelet-rich plasma demonstrates superior clinical and serum biomarker outcomes compared with corticosteroids and NSAIDs in late-stage knee osteoarthritis: a randomised controlled trial

Abstract

Objective

To assess the efficacy of two autologous platelet-rich plasma injections (PRP) in treating late-stage knee osteoarthritis and their effects on joint pain, stiffness and function as well as serum biomarkers reflecting inflammation, and cartilage turnover.

Methods

Ninety patients (aged 40–80 years) with symptomatic KL 3–4 KOA on arthroplasty waiting lists were randomized to PRP (two injections one week apart) and CS (a single injection of betamethasone), or NSAID (aceclofenac tablets) as the control groups. Primary outcomes included VAS pain scores, WOMAC scores, opioid use, and serum biomarkers (COMP, MMP-3, CGRP, CCL2, VEGF, IL-6, IL-18, TNF-α, CX3CL1, sTREM2, sRAGE, TGF-β1, BDNF) were measured at baseline, 3 months, and 6 months.

Results

Eighty-two patients completed the 6-month follow-up. Baseline characteristics (VAS, WOMAC) were comparable among the groups. PRP treatment resulted in significant and sustained reductions in VAS pain scores at 3 and 6 months compared with those at baseline (p < 0.001) and in both control groups (3 months: p < 0.001 vs. CS and NSAID; 6 months: p = 0.004 vs. CS; p = 0.002 vs. NSAID). WOMAC total and subscale scores (pain, stiffness, function) improved significantly only in the PRP group at both follow-up points (all p ≤ 0.03), with significant intergroup differences favoring PRP. Opioid consumption was significantly lower in the PRP group than in the CS and NSAID groups at 3 months (p = 0.002 and p = 0.025, respectively) and than in the CS group at 6 months (p = 0.006). At 3 months, PRP significantly reduced levels of proinflammatory and cartilage degradation biomarkers (COMP, MMP-3, IL-6, IL-18, TNF-α) and CGRP compared with controls, while increasing levels of anti-inflammatory markers (sTREM2, sRAGE, TGF-β1). Several effects were maintained at 6 months, including continued reductions in COMP, IL-6, and IL-18 levels compared with the CS group and in IL-6 and TNF-α levels compared with the NSAIDs group. No significant changes were observed in MCP-1, VEGF, CX3CL1, BDNF, or β-NGF. Adverse events were mild and transient in the PRP group.

Conclusion

Two PRP injections demonstrated greater 6-month clinical and biomarker improvements compared to CS and NSAIDs in late-stage KOA, supporting PRP as a safe bridge therapy prior to arthroplasty.

Introduction

Knee osteoarthritis (KOA) is a common degenerative joint disorder characterized by progressive cartilage degradation, subchondral bone remodeling, and chronic low-grade inflammation. Disease severity is radiographically assessed using the Kellgren–Lawrence (KL) grading scale ranging from 0 to 4. Advanced KOA (KL grades 3–4) remains a therapeutic challenge, as standard treatments such as non-steroidal anti-inflammatory drugs (NSAIDs), corticosteroids (CS), and hyaluronic acid provide only temporary symptomatic relief without modifying disease progression [1–3].

In this context, biological therapies have gained attention as potential disease-modifying approaches. Platelet-rich plasma (PRP), an autologous product derived from centrifuged whole blood, has emerged as a promising orthobiologic, providing supraphysiological concentrations of growth factors and cytokines that modulate synovial inflammation, reduce chondrocyte apoptosis, and promote extracellular matrix synthesis [4]. The role of PRP injections remains under debate and currently the American Academy of Orthopaedic Surgeons recommendation for the use of PRP in KOA is limited but not excluded from clinical practice [5]. Corticosteroid (CS) injections are widely used due to their rapid anti-inflammatory and analgesic effects; however, these benefits are typically short-lived and do not address underlying tissue degeneration. In contrast, PRP exerts sustained biological effects on the joint microenvironment and are therefore commonly administered as a series of injections rather than a single dose. Although meta-analyses have confirmed the superiority of PRP over CS and hyaluronic acid in the terms of pain and function in early to moderate OA (up to 12 months), evidence in advanced disease remains limited, particularly regarding serum biomarkers reflecting cartilage turnover and inflammation [4, 6]. Previous randomised controlled trials on KL grades 1 to 3 KOA have demonstrated that PRP reduces the levels of proinflammatory markers (IL-6, TNF-α, MCP-1) while increasing the levels of anabolic factors (BMP-2, sTREM2), which is correlated with improvements in the WOMAC and VAS scores [7]. However, the optimal dosing strategy, including the clinical benefit of repeated PRP administration over single-injection approaches and its comparative efficacy against active treatments such as CS and NSAIDs in pre-arthroplasty populations, remains to be fully elucidated.

To comprehensively capture the multifactorial pathophysiology of advanced KOA and the pleiotropic effects of PRP, the selected biomarker panel includes molecules reflecting key biological domains: cartilage degradation (COMP, MMP-3) [8, 9], inflammation and immune activation (IL-6, IL-18, TNF-α, CCL2, CX3CL1, sTREM2, sRAGE), angiogenesis (VEGF) [10–13], neurogenic inflammation and pain modulation (CGRP, BDNF) [14], and tissue remodeling/anabolic signaling (TGF-β1) [15]. This targeted selection enables the simultaneous assessment of structural joint turnover, inflammatory cascades, neuro-immune interactions, and regenerative responses, providing a mechanistic framework to interpret clinical outcomes in response to PRP therapy.

This study addresses these gaps by evaluating dual PRP injections compared with intra-articular CS and oral NSAID controls in KL grade 3–4 KOA patients and assessing clinical outcomes alongside a targeted biomarker panel.

The purpose of this study was to assess the clinical efficacy of two autologous platelet-rich plasma injections for the treatment of late-stage knee osteoarthritis and their effects on serum biomarkers reflecting pain, inflammation, and cartilage turnover associated with osteoarthritis.

Methods

The study was designed as a single-center, prospective, randomised, active-controlled trial with three groups receiving different treatments. Between September 2024 and June 2025, 90 patients with knee osteoarthritis, Kellgren–Lawrence classification grades 3 and 4, already on the waiting list for knee replacement, aged between 40 and 80 years, were enrolled. Osteoarthritis was diagnosed according to the American College of Rheumatology criteria [16]. All radiographs were taken under weight-bearing conditions.

The exclusion criteria included a body mass index greater than 40 kg/m², intra-articular injection of CS or hyaluronic acid in the previous six months, infectious diseases, systemic diseases (severe cardiovascular disease, diabetes mellitus, rheumatoid arthritis, hematological disorders, and malignancy), immunosuppression, anticoagulant or antiaggregant therapy, and known allergies to the study medications.

Patients who met the eligibility criteria and provided consent to participate in the clinical trial were randomly allocated, using online statistical computing software (www.graphpad.com/quickcalcs), to one of three intervention groups: the study group – the PRP group (two injections of 3 mL of autologous PRP over a one- week period); the first control group – the CS group (one injection of 6 mg betamethasone sodium phosphate and 6 mg betamethasone dipropionate [Schering-Plough] in 1 mL with 2 mL of trimecaine hydrochloride 1% [Zentiva]); or the second control group – the NSAID group (aceclofenac 2 × 100 mg orally). The CONSORT diagram illustrating patient enrollment is shown in Fig. 1.

Fig. 1.

Consolidated standards of reporting trial flow diagram describing the grouping and flow of patients in the study

Sample size

To determine an adequate sample size for the study, we performed a power analysis using one-way analysis of variance (ANOVA). A sample size of 27 per group would achieve 90% power to detect a difference of 1.5 points in the VAS score, with a standard deviation (SD) of 1.5, at the 5% significance level in a two-sided hypothesis test. In anticipation of a 10% dropout rate, the study was designed to enroll 30 participants in each group.

Study approval

The study protocol was approved by the local ethics committee (registration number: 2022/EK/10078). The study was conducted according to the principles of the Declaration of Helsinki.

Outcome measures

The baseline characteristics for each group included the mean age, male/female ratio, body mass index (BMI), and mean serum biomarker levels at the beginning of the study (baseline).

Clinical examinations were conducted before treatment and at the 3- and 6th-month follow-ups. Changes in pain were assessed using the visual analogue scale (VAS), with a numerical rating ranging from 0 to 10. For all patients, pain, stiffness, and function were evaluated using the Western Ontario and McMaster Universities Arthritis Index (WOMAC) score [17]. The type and severity of all adverse events that occurred during the study, as well as their relationship to treatment, were recorded.

Preparation and administration of PRP

The protocol for the preparation of PRP was based on a previous study evaluating the efficacy and safety of PRP treatment in patients with osteoarthritis [18]. A blood sample (27 mL of venous blood) was drawn in three 10 mL citrate anticoagulant treated vacutainer tubes (S-Monovette, Sarstedt, Germany). First, whole blood was centrifuged for 15 min at 3200 rpm at room temperature (RT). In the second step, the buffy coat and plasma layer were centrifuged at 1500 rpm for 15 min at RT. The plasma layer was collected, and a third centrifugation at 3200 rpm for 10 min was performed to obtain two plasma fractions: the upper part consisting of platelet-poor plasma (PPP) and the lower part consisting of PRP. The PPP was first discarded to avoid mixing with the PRP. The tubes were shaken for 30 s to suspend the platelets. The buffy coat layer containing blood platelets was then carefully aspirated into a syringe in a volume of 3 mL of PRP and used for intra-articular injection within 30 min. All open procedures were performed in a high-efficiency, particle-filtered, laminar flow cabinet class II. Platelet and leucocyte levels in the PRP product were not analyzed prior to injection, as these values had already been examined in a previous study [18]. The PRP products were not activated before injection, as platelets are known to be activated upon contact with collagen tissue.

Intervention procedures

Under aseptic conditions, two PRP products were injected intra-articularly without local anesthetic through a superolateral port at one-week intervals. A CS injection was administered once through a superolateral port. Rest, as needed for pain, and cryotherapy were recommended during the first 24 h after injection. Participants were asked to discontinue oral or topical pain medications (NSAIDs or analgesics), if they were taking any, for at least two weeks before the initiation of treatment and until completion of the trial.

Opioid consumption

In cases of insufficient analgesia, patients in all groups were allowed to use oral opioids (50 mg or 100 mg tramadol), as rescue medication. The maximum daily dose of opioids was determined according to the recommendations of the ADC (Approved Drug Catalogue) for tramadol. The total mean opioid consumption was recorded and standardized using morphine milligram equivalents (MMEs) for all time points [19].

Blood sampling

Approximately 5 ml of peripheral venous blood was collected from each patient by sterile venipuncture before treatment and at months 3 and 6, prior to breakfast and no later than 10:00 a.m. to prevent diurnal variation [20]. The blood was allowed to clot naturally for 30 min and then centrifuged at 4000 rpm for 10 min at RT. The serum was separated, and aliquots were stored at − 80 °C until further analysis.

Multiplex assay for quantification of cytokine/chemokine and matrix metalloproteinase

The concentrations of 10 biomarkers ( IL-6, IL-18, MCP-1, sTERM2, BDNF, β-NGF, VEGF, TNF-α, sRAGE, CX3CL2) were quantified in duplicate for each sample/patient using the LEGENDplex™ Human Neuroinflammation Panel 1 (13-plex) (BioLegend San Diego, CA, USA), a bead-based multiplex assay employing fluorescence-encoded beads, and measured by flow cytometry (BD FACSCalibur). COMP and MMP-3 were analysed using MILLIPLEX^® Assays (Merck Life Science, Darmstadt, Germany) according to the manufacturer’s protocol and the MAGPIX Luminex platform. xPONENT software version 4.2 for MAGPIX (Luminex Corporation, Austin, TX, USA) and Bio-Plex Manager 6.1 (Bio-Rad Laboratories, Hercules, CA, USA) were used for data analysis. Once standard curves were generated, the concentrations for each sample were interpolated using a 5-parameter curve fitting equation and expressed in pg/mL.

Measurement of CGRP and MMP-13 by ELISA

The concentrations of the target proteins in the serum of the patients studied were determined using the MMP-13 and CGRP ELISA kits (Abcam, Cambridge, UK) according to the manufacturer’s protocols and recommendations. These kits used the double antibody sandwich ELISA technique. The optical density value of the target biomarker in the sample was determined at 450 nm using a microplate reader (TriStar LB941, Berthold Technologies, Bad Wildbad, Germany). The standard curve was generated using a five-parameter logistic curve-fitting method, and the protein concentrations (pg/mL) of the test samples were calculated according to the manufacturer’s instructions.

Statistical analysis

Descriptive statistics were used to calculate means and standard deviations. Differences between groups for continuous variables were tested using one-way analysis of variance (ANOVA) for normally distributed data and the Kruskal–Wallis test for non-normally distributed data. Pearson’s chi-square test was used to analyse categorical variables. Pairwise comparisons were conducted using Student’s t-test for normally distributed data and the Mann–Whitney U test for non-normally distributed data. Effect sizes were quantified using Cohen’s d with 95% confidence intervals (95% CI). Given the exploratory nature of the biomarker data and the biological interrelatedness of markers, we did not apply multiple testing correction; results are interpreted in the context of effect sizes and clinical relevance. Statistical significance was defined as p < 0.05. SigmaPlot version 15 (Systat Software, Inc., San Jose, CA, USA) was used for the statistical analyses.

Results

Ninety patients (30 in each group) were recruited. Two patients in the CS group and four patients in the NSAID group discontinued participation after three months of follow-up because of insufficient pain control and undergone earlier total knee replacement. Two patients in the NSAID group experienced worsening of ischemic heart disease and were excluded from the study. Three patients in the PRP group and two patients in the CS group experienced transient increases in knee pain and swelling that resolved within one day. No other adverse events were recorded in the PRP or CS groups. In total, 82 patients completed six months of follow-up.

Per-protocol analysis was used as primary due to drop-outs reflecting treatment failure. No statistically significant differences were found between the groups in terms of mean age, sex, BMI, or baseline blood biomarker concentrations (Tables 1 and 4).

Table 1.

Basic demographics of the three groups of patients (n = 82)

Demographics	PRP (n = 30)	CS (n = 28)	NSAID (n = 24)	p-value (ANOVA)
Mean Age	66.4 (8.0)	67.0 (8.4)	67.8 (7.3)	0.87
% female	57%	57%	54%	0.97
BMI	31.2 (4.3)	30.6 (4.7)	31.2 (4.3)	0.73

n: Number; BMI: Body mass index; The values are expressed as the means, with standard deviations in parentheses

Table 4.

Intergroup comparisons of the serum concentrations of biomarkers

Biomarkers	PRP	CS	NSAID	Pairwise comparisons: PRP vs. CS			Pairwise comparisons: PRP vs. NSAID
				p-value	Cohen’s d	95% CI	p-value	Cohen’s d	95% CI
COMP
Baseline	9216.4 (3111.9)	9973.9 (2987.0)	9171.1 (3170.1)	0.35	−0.2482	[−0.7652, 0.2688]	0.97	0.0144	[−0.5223, 0.5512]
3 months	8352.4 (3681.8)	10984.9 (4271.1)	9050.8 (4411.1)	0.015	−0.662	[−1.1909, −0.133]	0.79	−0.1737	[−0.7115, 0.3641]
6 months	8710.8 (2581.4)	10244.2 (2879.9)	9395.1 (5603.1)	< 0.001	−0.5618	[−1.0868, −0.0367]	0.53	−0.1631	[−0.7007, 0.3746]
MMP-3
Baseline	21603.7 (8255.4)	22094,9 (8927.9)	21230.1 (11599,3)	0.96	−0.0572	[−0.5723, 0.4579]	0.6	0.0378	[−0.499, 0.5747]
3 months	18237.1 (6166.5)	24538,4 (11583.1)	22425.2 (12607,9)	0.024	−0.686	[−1.2159, −0.156]	0.3	−0.4378	[−0.9809, 0.1053]
6 months	21093.3 (6912.1)	24915,9 (15842.5)	21505.3 (8878,6)	0.97	−0.315	[−0.8332, 0.2032]	0.7	−0.0525	[−0.5894, 0.4843]
CGRP
Baseline	3237.7 (1389.4)	3135.9 (1941.0)	3278.9 (1509.4)	0.47	0.0607	[−0.4545, 0.5758]	0.93	−0.0285	[−0.5653, 0.5083]
3 months	2312.2 (1070.5)	3391.9 (1924.1)	3160.3 (1491.2)	0.032	−0.7001	[−1.2307, −0.1696]	0.024	−0.6658	[−1.217, −0.1145]
6 months	3505.3 (2973.1)	3728.6 (2265.0)	2916.5 (855.6)	0.36	−0.0841	[−0.5993, 0.4312]	0.81	0.255	[−0.2839, 0.7939]
MCP-1
Baseline	141.5 (49.9)	146.1 (42.6)	147.0 (56.1)	0.35	−0.0965	[−0.6119, 0.4188]	0.70	−0.1045	[−0.6416, 0.4327]
3 months	171.1 (68.5)	167.4 (56.3)	154.3 (57.3)	0.98	0.0915	[−0.4238, 0.6068]	0.29	0.2945	[−0.2452, 0.8341]
6 months	155.4 (66.7)	155.7 (61.9)	144.2 (44.2)	0.36	−0.0032	[−0.5183, 0.5118]	0.34	0.1949	[−0.3432, 0.7329]
VEGF
Baseline	138.3 (56.7)	138.4 (56.5)	135.8 (91.1)	0.84	−0.0016	[−0.5166, 0.5135]	0.17	0.0343	[−0.5026, 0.5711]
3 months	136.6 (47.8)	131.9 (64.9)	131.0 (67.1)	0.42	0.082	[−0.4332, 0.5973]	0.34	0.0985	[−0.4386, 0.6356]
6 months	139.6 (60.3)	130.3 (50.7)	130.0 (70.1)	0.65	0.1659	[−0.35, 0.6818]	0.52	0.1145	[−0.4227, 0.6517]
IL-6
Baseline	7.4 (8.6)	7.1 (10.1)	8.7 (9.6)	0.15	0.0378	[−0.4772, 0.5529]	0.71	−0.1466	[−0.682, 0.3929]
3 months	7.6 (9.6)	7.9 (17.4)	8.4 (9.1)	0.035	−0.278	[−0.7963, 0.2387]	0.8	−0.0847	[−0.6217, 0.4523]
6 months	6.6 (5.6)	6.8 (12.7)	8.9 (4.9)	0.014	−0.0377	[−0.5527, 0.4774]	0.024	−0.4298	[−0.9726, 0.1131]
IL-18
Baseline	117.8	122.5	121.7	0.91	−0.1029	[−0.6183, 0.4124]	0.63	−0,0667	[−0.6036, 0.4703]
3 months	87.0	103.7	122.8	0.037	−0.5624	[−1.0875, −0.0373]	0.003	−0.8382	[−1.3978, −0.2787]
6 months	95.9	122.3	115.9	0.013	−0.693	[−1.2232, −0.1627]	0.12	−0.4309	[−0.9738, 0.112]
TNF-α
Baseline	5.9 (5.9)	5.8 (5.6)	5.9 (3.8)	0.69	0.0995	[−0.4159, 0.6148]	0.93	0.1049	[−0.4322, 0.6421]
3 months	3.8 (6.7)	5.8 (5.3)	5.7 (3.2)	0.017	−0.3094	[−0.8275, 0.2087]	0.02	−0.3248	[−0.8651, 0.2154]
6 months	3.9 (5.2)	5.4 (5.1)	5.7 (3.9)	0.17	−0.255	[−0.7722, 0.2621]	0.028	−0.3617	[−0.9028, 0.1794]
CX3CL1
Baseline	140.9 (115.0)	148.4 (111.4)	164.4 (141.2)	0.99	−0.066	[−0.7117, 0.5797]	0.91	−0.1853	[−0.7648, 0.3941]
3 months	143.7 (137.1)	133.7 (112.0)	168.4 (138.9)	0.89	0.0786	[−0.6272, 0.7843]	0.51	−0.1786	[−0.8078, 0.4505]
6 months	135.2 (129.3)	155.1 (115.6)	161.0 (128.5)	0.49	−0.1612	[−0.8062, 0.4838]	0.51	−0.2002	[−0.8073, 0.4068]
sTREM2
Baseline	6479.9 (3012.7)	6461.3 (2362.9)	6598.1 (1930.9)	0.98	0.0069	[−0.5082, 0.5219]	0.5	−0.0456	[−0.5824, 0.4912]
3 months	7953.8 (3418.3)	6065.6 (2013.2)	6534.2 (1960.9)	0.019	0.6674	[0.1382, 1.1965]	0.27	0.4592	[−0.0496, 1.0401]
6 months	6919.9 (2791.6)	6647.8 (2877.8)	6316.2 (2277.2)	0.77	0.1569	[−0.3807, 0.6945]	0.4	0.0961	[−0.4193, 0.6114]
sRAGE
Baseline	918.8 (489.3)	925.7 (895.0)	904.8 (597.4)	0.08	−0.0076	[−0.5226, 0.5074]	0.79	0.0259	[−0.5109, 0.5627]
3 months	1247.6 (537.6)	909.8 (433.4)	921.2 (771.6)	0.017	0.6891	[0.159, 1.2192]	0.009	0.5009	[−0.0441, 1.046]
6 months	1218.0 (953.2)	897.1 (762.9)	920.8 (522.1)	0.13	0.3709	[−0.1528, 0.8947]	0.33	0.3752	[−0.1703, 0.9208]
TGF-β1
Baseline	93.3 (56.0)	94.1 (65.5)	94.9 (88.1)	0.98	−0.0136	[−0.5286, 0.5014]	0.52	−0.0228	[−0.5596, 0.5139]
3 months	131.1 (74.1)	97.3 (39.5)	93.4 (49.2)	0.048	0.7321	[0.2001, 1.264]	0.006	0.7425	[0.1878, 1.2972]
6 months	125.9 (83.4)	99.0 (51.1)	95.8 (67.8)	0.14	0.3865	[−0.1333, 0.9063]	0.45	0.3957	[−0.1462, 0.9376]
BDNF
Baseline	7879.2 (2637.5)	7643.9 (3132.3)	7859.3 (2544.9)	0.76	0.0815	[−0.4338, 0.5967]	0.29	0.0076	[−0.5291, 0.5444]
3 months	7690.4 (2233.5)	7853.1 (2397.5)	7524.9 (1740.4)	0.79	−0.0703	[−0.5855, 0.4449]	0.77	0.0815	[−0.4555, 0.6185]
6 months	7827.5 (2149.6)	7509.7 (2421.6)	7719.3 (2919.8)	0.6	0.1391	[−0.3766, 0.6547]	0.88	0.0429	[−0.4939, 0.5798]
β-NGF
Baseline	2.0 (1.3)	1.8 (1.2)	2.0 (1.1)	0.33	0.1583	[−0.3576, 0.6741]	0.68	−0.0636	[−0.6005, 0.4733]
3 months	1.9 (1.4)	1.8 (1.2)	2.1 (1.5)	0.27	0.1643	[−0.3516, 0.6802]	0.93	−0.06	[−0.5969, 0.4769]
6 months	2.1 (1.1)	1.8 (1.0)	2.1 (1.5)	0.16	0.2853	[−0.2323, 0.803]	0.56	−0.0053	[−0.542, 0.5315]

The values are expressed as the means, with standard deviations in parentheses. Cohen’s d effect sizes with 95% confidence intervals (CI) for group comparisons.

Clinical outcomes

Analysis of differences among the three cohorts via one-way analysis of variance (ANOVA) revealed no statistically significant differences in the mean baseline VAS pain scores or WOMAC scores. However, the VAS pain scores and WOMAC total scores at the follow-up differed significantly across all time points. After two PRP injections, there was a statistically significant decrease in the VAS score at month 3 (p < 0.001), and the decrease continued during the six-month follow-up (p < 0.001) compared with that at baseline. VAS scores decreased significantly only in the PRP group, with significant differences between groups at the 3-month period (PRP vs. CS: p < 0.001; PRP vs. NSAID: p < 0.001) and the 6-month time points (PRP vs. CS: p = 0.004; PRP vs. NSAID: p = 0.002; Table 2). Clinical improvement was observed in the total score of the WOMAC and the subscores of pain, stiffness, and function only in the PRP group. At the third-month visit, significant improvements in pain (p < 0.001), stiffness (p = 0.02), function (p = 0.001), and total score (p < 0.001) were detected, and the improvement in WOMAC scores was sustained at the sixth-month visit (pain: p < 0.001, stiffness: p = 0.03), function: p = 0.003), and total score: p < 0.001) compared with the baseline. All the WOMAC scores were significantly different between the PRP and control groups at 3- and 6-months of follow-up (Table 3).

Table 2.

Intragroup and intergroup comparisons of visual analogue scale pain severity

VAS	PRP	CS	NSAID	Pairwise comparisons (p-value):
				PRP vs. CS	PRP vs. NSAID
Baseline	6.2 (1.2)	6.3 (1.1)	6.3 (1.5)	0.73	0.75
3 months	3.3 (2.3)	6.0 (1.5)	5.7 (1.5)	< 0.001	< 0.001
6 months	3.7 (2.4)	5.8 (1.8)	5.8 (1.6)	0.004	0.002
Pairwise comparisons (p-value): Baseline vs. 3 months
	< 0.001	0.28	0.12
Pairwise comparisons (p-value): Baseline vs. 6 months
	< 0.001	0.2	0.22

The values are expressed as the means, with standard deviations in parentheses.

Table 3.

WOMAC scores changes from baseline (intragroup and intergroup comparisons)

WOMAC	PRP	CS	NSAID	Pairwise comparisons (p-value):
				PRP vs. CS	PRP vs. NSAID
Baseline
Pain	9.8 (4.5)	10.1 (3.0)	11.3 (3.7)	0.39	0.18
Stiffness	3.4 (2.1)	3.3 (1.4)	4.3 (2.8)	0.6	0.36
Function	34.2 (14.5)	32.7 (11.8)	35.9 (12.3)	0.77	0.64
TOTAL	49.0 (19.4)	46.0 (15.1)	48.8 (15.9)	0.74	0.94
3 months
Pain	6.0 (4.9)	9.4 (3.5)	12.7 (11.1)	< 0.001	< 0.001
Stiffness	2.4 (2.2)	3.4 (2.1)	3.3 (3.3)	0.02	0.03
Function	21.9 (14.1)	31.2 (11.1)	24.6 (16.3)	0.005	0.003
TOTAL	28.1 (18.2)	44.0 (15.2)	44.2 (19.8)	< 0.001	0.003
6 months
Pain	6.3 (3.5)	8.9 (3.7)	13.1 (4.2)	0.008	< 0.001
Stiffness	2.5 (1.3)	3.8 (1.9)	4.0 (2.6)	0.007	0.009
Function	23.7 (11.4)	33.0 (13.6)	42.7 (19.8)	0.007	< 0.001
TOTAL	32.5 (15.6)	45.6 (18.8)	57.9 (20.9)	0.005	< 0.001
Pairwise comparisons (p-value): Baseline vs. 3 months
Pain	< 0.001	0.26	0.59
Stiffness	0.02	0.89	0.12
Function	0.001	0.63	0.57
TOTAL	< 0.001	0.57	0.38
Pairwise comparisons (p-value): Baseline vs. 6 months
Pain	< 0.001	0.24	0.1
Stiffness	0.03	0.3	0.57
Function	0.003	0.92	0.11
TOTAL	< 0.001	0.93	0.06

The values are expressed as the means, with standard deviations in parentheses. WOMAC: Western Ontario and McMaster Universities Osteoarthritis Index.

Opioid consumption

The average opioid consumption, expressed in MMEs, was the lowest in the PRP group (Fig. 2). A significant difference was observed at the 3-month follow-up compared with the CS (p = 0.002) and NSAID (p = 0.025) groups and at the 6-month follow-up compared with the CS group (p = 0.006).

Fig. 2.

Total rescue opiate consumption (MMEs – morphine milligrams equivalents). The error bars indicate the standard deviations; differences were analyzed using Student’s t-test (* p ≤ 0.05, **: p ≤ 0.01, ***: p ≤ 0.001)

Laboratory outcomes

At the 3-month follow-up in the PRP group, there was a significant decrease in the concentrations of the pro-inflammatory biomarkers COMP (p = 0.015), MMP-3 (p = 0.024), IL-6 (p = 0.035), IL-18 (p = 0.037), and TNF-α (p = 0.017), as well as a significant increase in the levels of the anti-inflammatory biomarkers sTERM2 (p = 0.019), sRAGE (p = 0.017) and TGF-β1 (p = 0.048) compared with those in the CS group. There were significant differences in the levels of the biomarkers IL-18 (p = 0.003), TNF-α (p = 0.02), sRAGE (p = 0.009), and TGF-β1 (p = 0.006) between the PRP and NSAID groups. Three months after PRP administration, the serum CGRP levels were significantly lower than those in the CS (p = 0.032) and NSAID (p = 0.024) groups, reflecting reduced peripheral sensitization and pain signaling in OA patients. There were no significant differences in the concentrations of other biomarkers (MCP-1, VEGF, CX3CL1, BDNF, β-NGF) between the study groups.

In addition, in the PRP group, there was a statistically significant decrease in the concentrations of the pro-inflammatory biomarkers COMP (p < 0.001), IL-6 (p = 0.014), and IL-18 (p = 0.013), as well as a significant increase in the anti-inflammatory biomarkers sTERM2 (p = 0.026), TGF-β1 (p = 0.03), and sRAGE (p = 0.054) compared with those in the CS group at the 6-month follow-up. Compared with those in the NSAID group, significant changes in the IL-6 (p = 0.024) and TNF-α (p = 0.028) levels were observed (Table 4).

Discussion

This randomized controlled trial provides robust evidence that two weekly intra-articular injections of nonactivated autologous PRP offer superior clinical benefits and favorable biomarker modulation in patients with late-stage KOA (KL grades 3–4) compared with a single intra-articular CS injection or oral NSAID therapy. At the 3 and 6 months, PRP significantly reduced the VAS pain score (compared with baseline and controls), improved all the WOMAC domains, and decreased opioid consumption, with outcomes being sustained without plateauing, in contrast with the transient or absent improvements of the control groups, which led to a 13% dropout rate.

Meta-analyses have indicated that PRP is more effective than CS in reducing pain and improving function over medium- and long-term follow-up [21, 22]. PRP may not provide immediate pain relief as steroids do; however, it offers longer-lasting symptom relief than CS injections do, which lose effectiveness after 4–6 weeks. PRP outperformed CS in terms of pain and functional outcomes 3 and 6 months after injection. Most published works on the effectiveness of PRP in treating OA are case series studies, with an average patient age under 60 years and patients with early-stage OA. Patients with KL grades 3 and 4 KOA are less common. Only a few studies have compared the efficacy of PRP in different stages of KOA, showing short-term and poor results in KL grade 3 and 4 KOA patients, but better outcomes in mild to moderate KOA patients [23–25]. The findings of this study may be considered alongside the RESTORE trial by Bennell et al. (2021), which reported no significant benefit of PRP over placebo in mild-to-moderate knee osteoarthritis [26]. These discrepancies likely reflect differences in patient population and study design, as our cohort included patients with advanced disease and active comparators. Notably, beyond clinical improvement, we demonstrate significant modulation of key inflammatory and cartilage degradation biomarkers, including reductions in COMP, MMP-3, IL-6, IL-18, and TNF-α, alongside increases in anti-inflammatory mediators such as sTREM2, sRAGE, and TGF-β1. These findings suggest that PRP exerts measurable biological effects within the joint microenvironment that may not be captured by structural imaging endpoints. Collectively, this supports a context-dependent role of PRP, particularly as a biologically active intervention in advanced knee osteoarthritis. In contrast, the results of Ismaiel et al. [27], which are consistent with our study, revealed significant improvements in the VAS and WOMAC scores for KL grade 3 and 4 KOA patients after 3 and 6 months, compared with those of the CS group.

CS injections provide quick, short-term relief but may require frequent administration due to decreasing efficacy. This may be related to systemic and local side effects, such as hyperglycemia, adrenal suppression, cartilage damage with OA progression or septic arthritis [28]. NSAIDs are effective for treating OA pain, but carry significant gastrointestinal, cardiovascular, renal, hematologic and other systemic risks that increase with dose, duration, age and comorbidities [29]. PRP, as a minimally invasive treatment, poses a low risk of immune rejection or disease transmission because it uses the patient’s own blood. Common local reactions include pain and swelling at the injection site, which are generally mild and temporary [6]. We observed a transient increase in knee pain and swelling in three patients after PRP injection. In this study, we recorded a 13% dropout rate due to side effects or insufficient pain reduction in the CS and NSAID groups, in contrast to the 0% dropout rate in the PRP group.

There is no consensus on the optimal dosing of PRP in KOA treatment. Most studies recommend at least two doses for a positive effect lasting three to six months, while three doses are superior for maintaining the effect for up to twelve months [30, 31]. Based on the study duration and to avoid significant bias, such as administering only one corticosteroid injection, we decided to evaluate two doses of PRP.

The therapeutic rationale for PRP in OA is based on the high concentration of growth factors and bioactive molecules in platelets, which are released from alpha granules after platelets are activated by collagen exposure or external activators [6, 32]. The mechanism of pain relief by PRP is still being explored. PRP is rich in growth factors, including platelet-derived growth factor (PDGF), transforming growth factor-beta (TGF-β) and vascular endothelial growth factor (VEGF). These growth factors stimulate cellular proliferation, differentiation, and angiogenesis, improving nutrient and oxygen delivery to damaged areas, which is crucial for cartilage regeneration. Additionally, PRP increases the production of cartilage matrix components, such as collagen type II and aggrecan, while decreasing the levels of matrix metalloproteinases (MMPs) that break down cartilage. This helps to maintain cartilage integrity. However, cartilage is structurally free of blood vessels and nerves, and studies have shown a weak correlation between pain and cartilage defects [33, 34]. OA joints show an imbalance between pro-inflammatory and anti-inflammatory factors. Higher levels of inflammation are usually associated with more severe pain and disease progression [35]. Thus, pain relief is likely due mainly to the modulation of inflammation [36]. PRP contains anti-inflammatory cytokines, such as interleukin-1 receptor antagonist (IL-1ra) and soluble tumor necrosis factor receptors, which inhibit pro-inflammatory cytokines, such as interleukin-1 (IL-1) and tumor necrosis factor-alpha (TNF-α). In addition, pain neurotransmitters such as nerve growth factor (NGF) are released into OA joints [37]. CGRP is a neuropeptide released primarily from sensory C-fiber neurons in the OA synovium, that acts through CLR/RAMP1 receptors to induce vasodilation, neurogenic inflammation, and mechanical hypersensitivity. In knee OA, elevated synovial/serum CGRP correlated with pain intensity (VAS, WOMAC pain subscale; r = 0.56, p < 0.01) and osteophyte formation, independent of radiographic severity (KL grade), positioning it as a biomarker of the “pain-dominant” OA phenotype rather than structural degradation [38, 39]. Three months following intra-articular PRP administration, serum CGRP concentrations were significantly reduced in OA patients. This decline reflects PRP-induced modulation of the prostaglandin E2 (PGE2)-EP receptor signaling pathway, which downregulates CGRP expression in synovial fibroblasts and sensory afferents, culminating in sustained analgesia as evidenced by a > 30% reduction in visual analog scale (VAS) pain scores at 3 months [40]. Compared with that in both the CS and NSAID groups, the CGRP levels was lower, highlighting the PRP in targeting the neuropathic pain component of OA. PRP can decrease nociceptor stimulation and reduce sensitization. Several studies have suggested that PRP modulates the inflammatory environment of the OA joint and suppresses catabolic processes [41, 42]. However, the exact mechanism by which PRP regulates inflammation and pain has rarely been studied.

Serum biomarkers can be used to monitor the response to knee OA therapy. In this study, we evaluated serum biomarkers identified by OARSI (Osteoarthritis Research Society International) using the BIPED (Burden of Disease, Investigative, Prognostic, Efficacy of the intervention, and Diagnostic) classification [43–45]. MMP-3, COMP, MCP-1, IL-6, IL-18, VEGF, TNF-α, and CX3CL1 are inflammatory biomarkers that increase with synovial inflammation and cartilage breakdown. In addition to these inflammatory biomarkers, several anti-inflammatory biomarkers, such as sTREM2, TGF-β1, and sRAGE, as well as pain biomarkers with significant roles in osteoarthritis pain (BDNF, β-NGF, CGRP) which have also shown great potential for determining the prognosis and responsiveness to OA treatments, have recently been identified. There is a lack of literature reporting the effects of PRP injection on OA blood serum biomarkers, and the evidence remains inconclusive [7, 46–49]. To date, no studies have evaluated OA biomarkers in a specific group of KL grade 3–4 KOA patients treated with PRP injections.

Three months after two intra-articular PRP injections, serum concentrations of the pro-inflammatory biomarkers MMP-3, IL-6, IL-18, and TNF-α were significantly lower than those in the group receiving CS injections. These reductions persisted at 6 months for IL-6, and IL-18. These findings align with the multifaceted mechanism of PRP, involving the release of growth factors (e.g., PDGF, TGF-β) and anti-inflammatory cytokines that promote chondrocyte proliferation, extracellular matrix synthesis, and the suppression of degradative enzymes, in contrast with the anti-inflammatory action of CS, which is primarily symptomatic via glucocorticoid receptor-mediated NF-κB inhibition, which lacks anabolic effects and may accelerate cartilage loss with repeated use. Consistent with prior studies, the observed decrease in MMP-3 levels after PRP corroborates reports of reduced serum MMP-3 levels at 3–9 months compared with those after CS, where CS failed to alter MMP-3 levels despite short-term pain relief, underscoring the superior ability of PRP to modulate matrix turnover. Similarly, the decrease in IL-6 and IL-18 levels mirrors the systemic dampening effect of PRP on pro-inflammatory cytokines, as evidenced in randomized trials showing sustained IL-6 suppression at 3 months post-PRP compared with that of hyaluronic acid or CS, where CS had only transient effects [50–52]. This study also releaved a significant increase in the anti-inflammatory protective biomarkers STREM2, sRAGE, and TGF-β1 at 3 months following PRP injections compared with those in the CS group. Conversely, elevations in sTREM2, sRAGE, and TGF-β1 highlight the neuroprotective and antiglycation roles of PRP: an increase in sTREM2 indicates myeloid cell-mediated resolution of inflammation, sRAGE upregulation neutralizes AGE-induced catabolism (inversely correlating with OA severity), and the increase in TGF-β1 drives anabolic repair patterns, as demonstrated by studies revealing that PRP increases TGF-β1 levels in synovial fluid at 1–3 months, which are absent in CS cohorts [11, 53].These results support the findings that PRP has an anti-inflammatory effect on KL grade 3–4 KOA, which was not observed in the CS or NSAID groups.

COMP is a non-collagenous protein resulting from cartilage breakdown. Some relevant studies have shown a positive correlation between COMP and the OA stage [54]. Kuculmez et al. [48] demonstrated that PRP injections had no effect on COMP concentrations. In this study, we observed significant reductions in COMP levels after PRP injection at all time points compared with those in the CS injection group. This may indicate that PRP reduces the degree of cartilage degradation for up to 6 months.

However, the study did not demonstrate a significant reduction in the levels of serum pain biomarkers (BDNF, and β-NGF). Because these biomarkers act directly on OA joints, assessing their concentration in synovial fluid may be more appropriate.

Despite growing evidence of the symptomatic benefits of PRP injections for KOA, major clinical guidelines advise caution in routine use. The guidelines of the Osteoarthritis Research Society International classified PRP as an “uncertain” treatment for KOA [55]. The American Academy of Orthopaedic Surgeons guidelines concluded that evidence supporting PRP use remains inconclusive [5]. The ESSKA ORBIT 2024 consensus supports PRP for early to moderate OA [56]. The International Cartilage Regeneration and Joint Preservation Society cautiously supported PRP, focusing on standardized protocols [57]. This caution stems from heterogeneity in PRP formulations, dosing protocols, leukocyte content, outcome measurements, and insufficient data on long-term effects, which limits the generalizability of the results. Future research should prioritise standardising PRP preparation protocols and patient stratification. According to the results of this study and those of previous studies, PRP can be a viable and safe alternative for the management of pain associated with KOA, especially in elderly patients, who may have diseases, who may take medications that contraindicate the use of NSAIDs or CS, or who are waiting for or have not yet decided on knee arthroplasty.

Limitations

This study has several limitations. The major limitation is the differences in route of administration, dosing frequency, and overall treatment intensity, which introduce potential bias and limit direct comparability between groups. The limitation is also that the participants were not blinded to the intervention procedure. Another limitation is the lack of quantitative analysis of the PRP product, although the blood cell content was demonstrated in our previous study [10]. Physical activity and circadian rhythm can also influence biomarker levels. To avoid this limitation, fasting blood samples were collected before 10 am. The NSAID group had a significant dropout rate due to adverse effects or inadequate pain control.

Conclusion

Over six months, two autologous PRP injections demonstrated greater and more sustained clinical improvements and favourable biomarker changes compared with CS and NSAIDs in patients with late-stage KOA, consistent with anti-nociceptive, anti-inflammatory, and cartilage-protective effects. These findings support PRP as a safe bridge to arthroplasty, however, more randomized controlled trials with longer durations and larger populations are needed to confirm these results and to investigate the persistence of the beneficial effects following PRP injections in KOA treatment.

Author contributions

ML: project administration, writing, statistical analysis; OA: data extraction, laboratory analysis; MM: investigation; MS: investigation; JB: investigation; VF: data extraction, investigation; LS: laboratory analysis; JM: laboratory analysis; DH: supervision, validation, writing.All authors agreed with the final version of the manuscript.

Funding

This project has received funding from the Scientific Grant Agency of the Ministry of Education, Science, Research and Sports of the Slovak Republic – VEGA no. 1/0686/24.

Data availability

The data underlying this article are available at reasonable request to the author ML ( [marek.lacko@upjs.sk](mailto: marek.lacko@upjs.sk) ).

Declarations

Ethics approval and consent to participate

The present study was approved by the ethics committee of the University hospital of Louis Pasteur in Kosice (ID: 2022/EK/10078). All patients signed informed consent to participate to the present study.

Competing interests

The authors declare no competing interests.