Assessment of adverse events and safety associated with intra‐articular platelet‐rich plasma injections compared to other injectates for knee osteoarthritis: A systematic review and meta‐analysis

Abstract

Objective

To evaluate adverse events (AEs) associated with intra‐articular platelet‐rich plasma (IA‐PRP) injections for knee osteoarthritis (KOA) and compare their rates with those of intra‐articular corticosteroids (IA‐CS), hyaluronic acid (IA‐HA), and normal saline (IA‐NS).

Literature Survey

A systematic search of PubMed, Embase, and Cochrane Library databases identified randomized controlled trials (RCTs) published after 2015. Studies were included if they reported AEs related to IA‐PRP injections for KOA and provided comparator data for IA‐CS, IA‐HA, or IA‐NS.

Methodology

Data on AEs were extracted and categorized into mild knee pain and swelling, severe knee pain requiring withdrawal, knee stiffness, other musculoskeletal (MSK) events, non‐MSK events, and severe AEs. A random‐effects meta‐analysis was performed to calculate odds ratios (OR) with 95% confidence intervals (CI) for AE rates in IA‐PRP versus comparator groups. Subgroup analysis was conducted to evaluate the effects of leukocyte concentration on AE rates.

Synthesis

Thirty‐two RCTs published articles, involving 1268 IA‐PRP‐treated knees, met inclusion criteria. AEs were reported in 18.7% of IA‐PRP cases, with mild knee pain and swelling being the most common (10.6%). Compared to IA‐HA, IA‐PRP had significantly higher rates of mild knee pain and swelling (p < .001). Subgroup analysis revealed that this difference was significant only for leukocyte‐rich IA‐PRP (p < .05), whereas leukocyte‐poor IA‐PRP showed no significant difference compared to IA‐HA (p = .51). Knee stiffness was more frequent in IA‐PRP versus IA‐NS (p = .031). There were no significant differences in other AE categories, and no severe AEs were reported across any groups.

Conclusions

IA‐PRP injections are associated with mild, transient AEs such as knee pain and swelling, particularly with leukocyte‐rich PRP formulations. These symptoms typically resolve without intervention. Leukocyte‐poor IA‐PRP showed a similar safety profile to IA‐HA. No severe AEs were observed, supporting the overall safety of IA‐PRP for KOA. Clinicians should counsel patients on the likelihood of mild postprocedure symptoms, especially with high leukocyte formulations, while considering IA‐PRP as a treatment option.

INTRODUCTION

The knee joint is the most commonly affected site by osteoarthritis, with the Framingham study reporting a 9.5% prevalence of symptomatic knee osteoarthritis (KOA) in the general population. ¹ , ² The pathogenesis of KOA is multifactorial and complex, involving repetitive mechanical stress alongside genetic and immune factors contributing to disease progression. ² Nonoperative treatments typically include weight reduction, activity modification, physical therapy, nonsteroidal anti‐inflammatory drugs (NSAIDs), and intra‐articular (IA) injections of hyaluronic acid (HA) or corticosteroids (CS). ³ , ⁴ , ⁵ However, these standard nonoperative therapies are limited in effectiveness, often providing only short‐term symptomatic relief, with many patients eventually requiring surgical interventions such as arthroplasty. ⁶ , ⁷ , ⁸ For those who have not responded to standard treatments, are not candidates for surgery, or prefer to avoid surgery, orthobiologic therapies, including platelet‐rich plasma (PRP), may offer a promising option to address the treatment gap.

PRP is believed to modulate inflammation through the release of growth factors from platelets, which recruit reparative cells and promote analgesic effects. ⁹ , ¹⁰ In vitro studies have demonstrated variable effects of different PRP preparations on differentiation, migration, and proliferation of chondrocytes and synoviocytes, with some evidence suggesting a dose‐dependent relationship with platelet concentration. ¹¹ , ¹² , ¹³ In vivo research has also shown promising results, highlighting PRP's potential for delivering analgesic effects. ¹⁴ Recent meta‐analyses indicate that PRP injections are more effective in reducing pain and improving function compared to IA‐HA or IA‐CS injections for KOA. ¹⁵ , ¹⁶ , ¹⁷ PRP has the highest‐quality evidence for the treatment of symptomatic KOA along with lateral epicondylosis. ¹⁸ , ¹⁹

Many clinicians consider PRP to be safe due to its autologous nature. ⁹ However, concerns remain regarding the variability in adverse events (AEs) resulting from the heterogeneity of PRP products. ²⁰ Additionally, IA injections carry inherent procedural risks. Although previous meta‐analyses on IA‐PRP for KOA have evaluated safety to some extent, most have not thoroughly examined AEs or included a sufficient number of studies, raising concerns about the statistical power of their conclusions. ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ The purpose of this study was to assess the AEs associated with IA‐PRP injections for KOA treatment and to compare their incidence rates with those of commonly used IA injectates, including CS, HA, and normal saline (NS).

METHODS

Search strategy

This systematic review adhered to the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses 2020. ³¹ Randomized controlled trials (RCTs) investigating KOA treated with IA‐PRP injections were included. The exclusion criteria were as follows:

1. Studies that did not report AE or lacked sufficient information to extract data on AEs (eg, number of AE not reported).

2. Studies combining orthobiologic products (eg, PRP + bone marrow aspirate concentrate) unless IA‐PRP was used alone in one study arm.

3. Studies involving extra‐articular injections (eg, interosseous injections).

4. Studies focusing on the surgical application of PRP or its use in surgical procedures.

5. Cadaveric studies.

6. Animal studies.

7. Reviews, technical notes, commentaries, abstract‐only articles, position statements, consensus documents, letters to the editor, clinical trial protocols, and retracted studies.

8. Non‐English language articles.

9. Studies examining inflammatory or hemophilic osteoarthritis.

10. Studies reporting only non‐clinical outcomes (eg, inflammatory marker levels or magnetic resonance imaging findings).

11. Studies published before 2015.

12. When multiple publications on the same patient cohort with varying follow‐up periods were identified, only the most recent publication was included, provided the study methods had not changed since the initial data collection.

A comprehensive literature search was conducted on December 14, 2022, using Elsevier Embase, Ovid Cochrane Central Register of Controlled Trials, Web of Science Core Collection, and Ovid MEDLINE. The search encompassed epub ahead of print, in‐process, in‐data‐review, and other nonindexed citations, as well as daily and weekly updates. Keywords and subject headings included terms such as platelet, thrombocyte, PRP, OA, arthritis, arthrosis, arthritides, polyarthritides, gonarthrosis, gonitis, and gonarthritis (Appendix 1).

The initial search yielded 3665 records, which were imported into a citation management program (EndNote X9, Clarivate Analytics, Philadelphia, PA, USA) for duplicate removal. After removing duplicates, 2195 unique records were uploaded to a data extraction tool (Rayyan, Doha, Qatar) for title and abstract screening. From this process, 195 studies met the inclusion criteria and advanced to full‐text screening.

Two team members (H.N. and J.K.) independently reviewed the full‐text records, with disputes resolved by a third reviewer (W.S.). Following the full‐text review, 32 studies met the inclusion and exclusion criteria and were included in the final review. The review and selection process are summarized in Figure 1, and the characteristics of the included studies are detailed in Table 1.

FIGURE 1.

PRISMA (Preferred Reporting Items for Systematic Reviews and Meta‐Analyses) flow diagram for the identification and selection of studies included in this study. PRP, platelet‐rich plasma; PRGF, plasma rich in growth factors.

TABLE 1.

Details of study characteristics and reports on adverse events for the included studies.

Study ID	Comparator a	# Pts (PRP)	# Pts (comparator)	# Knees (PRP)	# Knees (comparator)	AE (PRP)	AE (comparator)
Filardo et al. 2015	HA	94	89	94	89	0	2 severe knee pain
Montanez‐Heredia et al. 2016	HA	27	26	27	26	10 mild knee pain and swelling	4 mild knee pain and swelling
Papalia et al. 2016	HA	23	24	23	24	0	0
Paterson et al. 2016	HA	10	9	10	9	2 mild knee pain and swelling	0
Smith 2016	NS	15	15	15	15	0	0
Jubert et al. 2017	CS	34	30	34	30	0	0
Buendia‐Lopez et al. 2018	HA	33	32	33	32	0	2 severe knee pain
Güvendi et al. 2018	CS	33	17	33	17	1 other MSK event (erythema)	0
Lisi et al. 2018	HA	28	22	29	25	0	0
Louis et al. 2018	HA	17	17	17	17	3 mild knee pain and swelling 1 other MSK event (knee sprain)	2 mild knee pain and swelling 1 other MSK event (knee sprain) 1 non‐MSK event (amygdalatomy)
Su et al. 2018	HA	25	30	25	30	2 mild knee pain and swelling 3 other MSK events (2 LBP, 1 other knee pain) 3 non‐MSK events (1 URI, 1 bronchitis, 1 HA)	2 mild knee pain and swelling 2 non‐MSK events (1 cold, 1 acute MI)
Wu et al. 2018	NS	20	20	20	20	0	0
Yu et al. 2018	HA	104	88	104	88	2 other MSK events (2 rash) 26 non‐MSK events (4 HTN, 3 diarrhea, 5 proteinuria, 1 vomiting, 3 fatigue, 4 constipation, 3 hypertriglyceridemia, 3 peripheral edema)	2 other MSK events (2 rash) 23 non‐MSK events (5 HTN, 2 diarrhea, 4 proteinuria, 2 vomiting, 2 fatigue, 3 constipation, 3 hypertriglyceridemia, 2 peripheral edema)
Huang et al. 2019 (HA)	HA	40	40	40	40	5 mild knee pain and swelling	2 mild knee pain and swelling
Huang et al. 2019 (CS)	CS	40	40	40	40	5 mild knee pain and swelling	3 mild knee pain and swelling
Tavassoli et al. 2019	HA	56	27	56	27	7 mild knee pain and swelling	0
Elik et al. 2020	NS	30	27	30	27	5 mild knee pain and swelling	3 mild knee pain and swelling
Elksnins‐Finogejevs et al. 2020	CS	19	17	19	17	15 mild knee pain and swelling	0
Kesiktas et al. 2020	HA	18	18	18	18	0	0
Yaradilmis et al. 2020	HA	60	30	60	30	15 mild knee pain and swelling	2 mild knee pain and swelling
Bennell et al. 2021	NS	144	144	144	144	28 mild knee pain and swelling 5 knee stiffness 44 other MSK Events (31 lower limb and 13 upper limb MSK symptoms) 13 non‐MSK events (unspecified)	21 mild knee pain and swelling 41 other MSK events (23 lower limb and 18 upper limb musculoskeletal symptoms) 16 non‐MSK events (unspecified)
Dorio et al. 2021	NS	20	21	20	21	12 mild knee pain and swelling 1 knee stiffness	7 mild knee pain and swelling
Dulic et al. 2021	HA	34	30	34	30	4 mild knee pain and swelling	2 mild knee pain and swelling
Li et al. 2021	HA	42	44	42	47	14 mild knee pain and swelling	5 mild knee pain and swelling
Park et al. 2021	HA	55	55	55	55	3 mild knee pain and swelling 1 other MSK event (1 musculoskeletal pain) 2 non‐MSK events (1 nasopharyngitis, 1 constipation)	5 other MSK events (3 musculoskeletal pain, 2 backache) 3 non‐MSK events (2 nasopharyngitis, 1 HA)
Shoma et al. 2021	HA	65	68	65	68	0	0
Tucker et al. 2021	NS	11	6	11	6	1 mild knee pain and swelling	0
Xu et al. 2021	HA	30	20	40	34	9 mild knee pain and swelling	1 mild knee pain and swelling
Aakash et al. 2022	CS	15	15	15	15	0	0
Nunes‐Tamashiro et al. 2022 (CS)	CS	34	33	34	33	0	0
Nunes‐Tamashiro et al. 2022 (NS)	NS	34	33	34	33	0	0
Szwedowski et al. 2022 (HA)	HA	25	24	25	24	0	0
Szwedowski et al. 2022 (CS)	CS	25	24	25	24	0	0
Wang et al. 2022‐1	HA	54	56	54	56	0	4 mild knee pain and swelling
Wang et al. 2022‐2	HA	42	43	42	43	0	0

Note: Note that there are two entries for Huang et al. (2019), Nunes‐Tamashiro et al. (2022), and Wang et al. (2022), as these studies included two comparators (eg, Huang et al. (2019) compared PRP to both HA and CS, thus there is one entry for each comparator). Abbreviations: AE, adverse events; CS, corticosteroid; HA, hyaluronic acid; HTN, hypertension; ID, identification; LBP, low back pain; MSK, musculoskeletal; NS, normal saline; Pts, patients; PRP, platelet‐rich plasma; URI, upper respiratory infection.

^a Defined as the control group or treatment arm to which the intra‐articular PRP arm was compared in the randomized controlled study.

Data extraction

The following data were extracted from each study:

• Authors and publication year: The names of the authors and the year of publication.

• Comparator: The control or comparator arm against which the IA‐PRP arm was evaluated.

• Patient and knee counts: The number of patients included in the final analysis and the number of knees treated. If the number of knees treated was not specified, it was assumed to match the number of patients.

• Demographics: The age and gender distribution of the study participants.

• Follow‐up duration: The length of the follow‐up period.

• AE: Categorized into the following groups. Examples of AE in each category are shown in Table 2.

  • Overall AEs: All reported adverse events.
  • Mild transient knee pain and swelling: Knee pain and swelling that resolved without intervention.
  • Knee stiffness: Reported episodes of knee stiffness.
  • Severe AEs: Severe knee pain necessitating withdrawal from the study.
  • Other musculoskeletal (MSK) events including backache, rash and unspecified MSK pain
  • Non‐MSK medical events: Medical AE unrelated to the MSK system including gastrointestinal, respiratory, and cardiac complications

TABLE 2.

Descriptions of each category of adverse events and frequency of adverse events.

	Overall AE	Mild transient knee pain and swelling	Severe knee pain	Knee stiffness	Other MSK events	Non‐MSK medical events	Severe AE
Description	Sum of all AE	Mild knee pain and swelling that resolved on its own	Severe knee pain that required withdrawal from the study		Backache Rash Unspecified MSK pain	Gastrointestinal complications Fatigue Hypertension Upper respiratory infection Nasopharyngitis/bronchitis Peripheral edema Headache Amygdalotomy Myocardiac ischemia Unspecified non‐MSK events	Infection Hematoma Thrombosis Hemarthrosis
N (%)	PRP: 237 (18.7%)	PRP: 135 (10.6%)	PRP: 0 (0%)	PRP: 6 (0.5%)	PRP: 52 (4.1%)	PRP: 44 (3.5%)	PRP: 0 (0%)
	HA: 65 (8.0%)	HA: 24 (3.0%)	HA: 4 (0.5%)	HA: 0 (0%)	HA: 8 (1.0%)	HA: 29 (3.6%)	HA: 0 (0%)
	CS: 3 (1.7%)	CS: 3 (1.7%)	CS: 0 (0%)	CS: 0 (0%)	CS: 0 (0%)	CS: 0 (0%)	CS: 0 (0%)
	NS: 88 (33.1%)	NS: 31 (11.7%)	NS: 0 (0%)	NS: 0 (0%)	NS: 41 (15.4%)	NS: 16 (6.0%)	NS: 0 (0%)

Abbreviations: AE, adverse events; CS, corticosteroid; HA, hyaluronic acid; MSK, musculoskeletal; NS, normal saline; PRP, platelet‐rich plasma.

Statistics

Data analysis was performed using SPSS (IBM, Version 28, IL), with statistical significance defined as p ≤ .05. A random‐effects meta‐analysis was conducted to compare complication rates among the IA‐PRP, IA‐CS, IA‐HA, and IA‐NS groups, with odds ratios (OR) reported along with 95% confidence intervals (Cis).

A subgroup analysis was also conducted to evaluate the effect of leukocyte concentration on AEs. This analysis focused specifically on mild transient knee pain and swelling, as this was the only AE with sufficient data for evaluation. The two subgroups—leukocyte rich (LR) and leukocyte poor (LP)—were defined based on PRP preparations with a leukocyte concentrations of ≥1% and <1%, respectively.

RESULTS

A total of 32 studies were included in the review (Figure 1, flow diagram): 21 studies compared PRP to HA, 7 compared PRP to CS, and 7 compared PRP to NS. Notably, three studies included comparisons of PRP to more than one comparator (PRP vs. HA and CS, ³² , ³³ PRP vs. HA and NS ³⁴ ). The review included 2491 patients (2522 knees, 38% male): 1257 patients (1268 knees) in the PRP group, 792 patients (812 knees) in the HA group, 176 patients (176 knees) in the CS group, and 266 patients (266 knees) in the NS group. The average age of participants was 58.5 years, and the median follow‐up time was 6 months. Table 2 summarizes the study results regarding frequency of AEs, and Table 3 provides the results of the meta‐analysis comparing AEs across the injectates. Detailed information regarding how AEs were extracted is available in Appendix 2.

TABLE 3.

Summary of the results of the meta‐analysis comparing adverse events of intra‐articular platelet‐rich plasma to other intra‐articular injections.

AEs	# Studies	Log OR	Lower 95% CI	Upper 95% CI	p value
Overall AE
PRP vs. HA	15	0.510	0.032	0.987	.037***
PRP vs. CS	3	1.508	−0.289	3.306	.1
PRP vs. NS	4	0.467	0.036	0.898	.034***
Mild knee pain and swelling
PRP vs. HA	12	0.981	0.554	1.407	<.001***
PRP vs. CS	2	1.793	−0.594	4.179	.141
PRP vs. NS	4	0.484	−0.022	0.990	.061
Severe knee pain
PRP vs. HA	2	−1.405	−3.022	0.213	.089
PRP vs. CS	0	‐	‐	‐	‐
PRP vs. NS	0	‐	‐	‐	‐
Knee stiffness
PRP vs. HA	0	‐	‐	‐	‐
PRP vs. CS	0	‐	‐	‐	‐
PRP vs. NS	2	1.541	0.142	2.940	.031***
Other MSK events
PRP vs. HA	4	−0.021	−1.456	1.413	.977
PRP vs. CS	1	‐	‐	‐	‐
PRP vs. NS	1	‐	‐	‐	‐
Non‐MSK medical events
PRP vs. HA	4	−0.068	−0.635	0.499	.814
PRP vs. CS	0	‐	‐	‐	‐
PRP vs. NS	1	‐	‐	‐	‐

Note: If the number of studies was either 0 or 1, a meta‐analysis could not be performed, and the results are indicated as “‐”.

Abbreviations: AE, adverse events; CS, corticosteroid; CI, confidence interval; HA, hyaluronic acid; MSK, musculoskeletal; NS, normal saline; OR, odds ratio; PRP, platelet‐rich plasma.

^***Indicates statistical significance.

Overall AEs were reported in 18.7%, 8.0%, 1.7% and 33.1% of patients in the PRP, HA, CS, and NS groups, respectively. Statistically significant differences in overall AEs were observed, with HA (p = .037) and NS (p = .034) showing significantly fewer AEs compared to PRP. To clarify, the overall AE rates represent the pooled data across all studies for each treatment group. The statistical analysis, however, specifically focused on direct comparisons between NS and PRP in studies where both treatments were investigated. This comparison revealed that NS injections were associated with statistically fewer AEs when directly compared to PRP, despite a higher overall pooled rate of AEs for NS across all studies.

Mild knee pain and swelling were reported in 10.6%, 3.0%, 1.7%, and 11.7% of patients in the PRP, HA, CS, and NS groups, respectively. A statistically significant difference was observed between HA and PRP, with fewer AEs reported in the HA group (p < .001). Although the CS group had a slightly lower incidence than HA, the difference was not statistically significant. The duration of mild knee pain and swelling was reported to last less than a week and typically resolved spontaneously or with the use of anti‐inflammatory medication, activity modification, and cryotherapy. ³² , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ Figure 2 shows the results of the subgroup analysis on the effect of leukocyte dosing on mild transient knee pain and swelling. For LR IA‐PRP, the effect on mild knee pain and swelling remained statistically significant when compared to IA‐HA (OR = −1.39, [−2.07, −0.72], p < .05). For LP IA‐PRP, the difference was no longer statistically significant when compared to IA‐HA (OR = −0.44, [−1.74, 0.86], p = .51).

FIGURE 2.

Subgroup analysis comparing mild transient knee pain and swelling between leukocyte rich (LR) intra‐articular (IA)‐PRP and leukocyte poor (LP) IA‐PRP when compared to IA‐Hyaluronic acid. LR IA‐PRP showed a statistically significant increase in mild knee pain and swelling (odds ratio [OR] = −1.39, 95% CI [−2.07, −0.72], p < .05). In contrast, LP IA‐PRP did not demonstrate a significant difference (OR = −0.44, 95% CI [−1.74, 0.86], p = .51).

Severe knee pain was reported in 0%, 0.5%, 0%, and 0% of patients in the PRP, HA, CS, and NS groups, respectively. This outcome was examined in two studies comparing PRP and HA, with no statistically significant difference observed (p = .089). The four cases involved included two instances of severe knee pain and swelling within 2 weeks of HA injection that required NSAID administration, ⁴² and two additional cases of swelling and severe pain of unspecified timeframe following HA injection. ⁴³ Various HA products were used, including Durolane, ⁴² , ⁴⁴ Hyalgan, ⁴⁵ , ⁴⁶ Artz, ³⁸ , ⁴⁷ Synvisc One, ³⁹ Hyalabrix, ⁴³ Adant, ⁴⁸ Sinovial HL, ⁴⁹ Freda, ⁵⁰ Ostenil, ⁴¹ Cartinorm, ³⁶ LBSA0103, ⁵¹ SOFAST, ⁵² Biovisc, ³³ Hyajoint Plus ⁴⁰ and unspecified products. ³² , ⁵³ , ⁵⁴ Most of these products are derived from bacterial fermentation, except for Synvisc One, Hyalgan, and Artz, which are avian derived. Most also have molecular weights >500 kilodaltons, except for Adant. The four reported cases of severe knee pain involved the use of Hyalubrix and Durolane, both of which are high‐molecular‐weight, fermentation‐derived products.

Knee stiffness was reported in 0.5%, 0%, 0%, and 0% of patients in the PRP, HA, CS, and NS groups, respectively. This outcome was examined in two studies comparing PRP and NS, with a statistically significant difference observed, favoring NS (p = .031). ³⁵ , ⁴² Although CS and HA groups also reported 0% incidence, the statistical comparison was limited to PRP versus NS. A total of six cases of knee stiffness were reported.

Other MSK events were reported in 4.1%, 1.0%, 0%, and 15.4% of patients in the PRP, HA, CS, and NS groups, respectively. No statistically significant difference was observed between PRP and HA (p = .977). These events included backache (0.2% in both PRP and HA groups, 0% in CS and NS groups), rash (0.2% in both PRP and HA groups, 0% in CS and NS groups), and other nonspecified MSK events (3.7% in PRP, 0.5% in HA, 2.3% in CS, and 15.4% in NS). A substantial portion of the nonspecified MSK events in the PRP and NS groups was reported in Bennell et al. (2021) as “upper and lower limb musculoskeletal problems”. ⁵⁵

Non‐MSK medical events were analyzed in four studies comparing PRP and HA, and one study comparing PRP and NS. Non‐MSK medical events were reported in 3.5%, 3.6%, 0% and 6.0% of patients in the PRP, HA, CS, and NS groups, respectively. No statistically significant difference was observed between PRP and HA (p = .814). These events included gastrointestinal issues (diarrhea, constipation, vomiting: 0.7% in PRP, 0.9% in HA, and 0% in NS), proteinuria (0.4% in PRP, 0.5% in HA, and 0% in NS), fatigue (0.3% in PRP, 0.2% in HA, and 0% in NS), hypertension (0.3% in PRP, 0.6% in HA, and 0% in NS), upper respiratory infection (0.1% in PRP and HA, and 0% in NS), nasopharyngitis/bronchitis (0.2% in PRP and HA, and 0% in NS), peripheral edema (0.2% in PRP and HA, and 0% in NS), headache (0.1% in PRP and HA, and 0% in NS), amygdalotomy (0.1% in HA, and 0% in PRP and NS), myocardial ischemia (0.1% in HA, and 0% in PRP and NS), and other unspecified non‐MSK events (1% in PRP, 0% in HA, and 6% in NS). It is unclear whether these non‐MSK medical events are related to the IA procedures. Detailed information regarding the timing of these events, such as amygdalotomy or myocardial ischemia, was not available in the studies. ⁴⁴ , ⁵⁰

No severe AEs, including infection, hematoma, thrombosis, or hemarthrosis, were reported in any of the studies.

DISCUSSION

This meta‐analysis found that the overall complication rate for IA‐PRP injections was approximately 18.7%, with the majority of AEs being mild knee pain and swelling. No serious complications were reported. Severe knee pain requiring study withdrawal occurred in only 0.5% of cases in the HA group. The overall complication rate for IA‐PRP was significantly higher compared to IA‐HA and IA‐NS. When examining individual AEs, the complication rate was significantly higher for mild knee pain and swelling, as well as knee stiffness, compared to the control groups. For mild knee pain and swelling, this significant difference was observed only in the LR IA‐PRP group but not in the LP group when compared to IA‐HA.

This study is the first meta‐analysis specifically focused on the AEs associated with IA‐PRP injections. Previous meta‐analyses on this topic have been limited by a small number of studies and by grouping AEs in ways that made it unclear which specific effects were being investigated, raising concerns about the power and clarity of their conclusions. ²¹ , ²² , ²⁵ , ²⁶ , ²⁷ , ⁵⁶ Our study aimed to clarify and define the AEs under investigation. Prior research on this subject has yielded mixed results, with some studies showing no significant differences in AE rates between PRP, HA, and NS. ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ Similar to our findings, a recent meta‐analysis also reported that LR PRP had a significantly higher rate of procedure‐related pain and swelling compared to HA, whereas there was no significant difference between LP PRP and HA. ¹⁶

Postinjection pain and swelling are common with PRP injections due to the growth factors that mediate local inflammation, and neutrophils in PRP may release matrix metalloproteinases and reactive oxygen species, causing inflammatory reactions at existing tissues. ⁹ This mechanism likely explains the higher rates of mild knee pain and swelling observed with LR‐PRP, but not with LP‐PRP, compared to HA. However, these symptoms generally resolve within 2 weeks and are usually self‐limiting. Importantly, no severe knee pain or other serious AEs leading to study withdrawal were reported in the included studies.

The absence of reported non‐MSK AEs in the IA‐CS group in this study should be interpreted in the context of the available evidence and methodological considerations. Large‐scale meta‐analyses and systematic reviews of IA‐CS injections for musculoskeletal conditions, including knee osteoarthritis, consistently report a low incidence of non‐MSK AEs, with most complications being local and self‐limited, such as postinjection pain or swelling. ⁵⁷ , ⁵⁸ , ⁵⁹ Systemic effects—such as transient hyperglycemia, facial flushing, or adrenal suppression—are recognized but occur infrequently and are rarely captured in the context of RCTs with limited sample sizes and short‐term follow‐up. ⁵⁷

In summary, our findings suggest that whereas IA‐PRP may have higher rates of mild, transient knee pain and swelling compared to HA, the procedure is generally safe with no reports of severe adverse outcomes. The safety profile of IA‐PRP appears comparable to that of HA, CS, and NS.

There are several limitations to this study. First, some studies did not clearly specify when AEs occurred, and symptoms such as injection site pain and swelling are expected but self‐resolving, making it debatable whether they should be classified as AEs. Additionally, some studies grouped AEs under broad categories, such as “nonmusculoskeletal medical conditions,” without further clarification, which limited the ability to distinguish between different types of AEs. Some of the non‐MSK events, such as amygdalotomy, were unlikely to be related to the IA‐PRP injection but were still included in the analysis. Moreover, heterogeneity in PRP preparation and administration, including differences in dose, leukocyte content, activation status, and follow‐up duration, as well as variability in post injection rehabilitation protocols and pain medication use, may have influenced AE rates. Unfortunately, only 6 of 32 studies provided sufficient data to calculate platelet dose, preventing a meaningful dose‐dependent analysis. Future research should aim to clarify the impact of PRP activation methods, cellular composition, postprocedural care protocols, and pain management on the occurrence and severity of AEs. Lastly, it is important to acknowledge that this systematic review exclusively included RCTs. Although RCTs are the gold standard for assessing efficacy, they are often conducted in highly controlled environments with selected patient populations and defined follow‐up periods, which can potentially lead to an underestimation of the true incidence and spectrum of AEs observed in real‐world clinical practice.

CONCLUSIONS

This meta‐analysis suggests that IA‐PRP injections for KOA are generally safe, with an overall complication rate of 18.7%, mostly consisting of mild and transient knee pain and swelling that resolve spontaneously. There was no severe AEs, and no cases required study withdrawal. LR IA‐PRP showed significantly higher rates of mild knee pain and swelling compared to HA, but no significant differences were observed with LP PRP. Overall, IA‐PRP has a safety profile similar to other commonly used IA injections, such as HA and CS. However, variability in PRP preparations, AE reporting, and study design limit definitive conclusions, and future research should focus on standardizing protocols and investigating dose‐dependent AEs.

DISCLOSURE

The authors have nothing to report.

Supporting information

Data S1PRISMA_2020_checklist.

APPENDIX 1. SEARCH STRATEGY FOR OVID MEDLINE(R) AND EPUB AHEAD OF PRINT, IN‐PROCESS, IN‐DATA‐REVIEW & OTHER NON‐INDEXED CITATIONS, DAILY AND VERSIONS <1946 TO DECEMBER 13, 2022>

1. exp Platelet‐Rich Plasma/6345

2. PRP.mp. 18235

3. exp Blood Platelets/81641

4. platelet*.mp. 309737

5. thrombocyte*.mp. 7203

6. or/1‐5 322686

7. exp Osteoarthritis, Knee/25939

8. exp Knee/15802

9. knee*.mp. 200647

10. 8 or 9 200647

11. osteo‐arthriti*.mp. 444

12. osteoarthriti*.mp. 108549

13. OA.mp. 44124

14. exp Arthritis/290014

15. arthriti*.mp. 250016

16. arthrosis.mp. 5767

17. arthroses.mp. 601

18. arthritides.mp. 1510

19. polyarthritides.mp. 26

20. gonarthrosis.mp. 1143

21. gonitis.mp. 124

22. gonarthritis.mp. 177

23. or/11‐22 400095

24. 10 and 23 61363

25. 7 or 24 61363

26. 6 and 25 1008

27. limit 26 to yr=“2015 ‐Current” 799

APPENDIX 2. DETAILS ON STUDY SPECIFICS EXTRACTED FOR ANALYSIS

Filardo 2015: reported that there were more post‐injection swelling and pain with respect to HA, but did not provide specific number of cases of these adverse events. The article did report that there were two cases of severe pain and swelling after the first HA injection, and these two patient had to be withdrawn from further injections. These two cases were added to the “severe pain” for our analysis.

Montanez‐Heredia 2016: reported that there were 9 cases of pain related to infiltration and 1 case of transitory swelling after the injection that self‐resolved. These cases were combined as 10 cases of “mild pain and swelling” for our analysis.

Jubert 2017: Figure 1 reports that 35 participants were allocated to receive PRP, and 30 patients also allocated to receive PRP. This is likely an error. For the analysis, used 34 participants as receiving PRP and 30 as receiving CS.

Buendia‐Lopez 2018: 2 participants developed severe pain and swelling immediately after the IA‐HA. Both required the use of NSAID for a week and were withdrawn from the study. These two cases were added to the “severe pain” for our analysis.

Lisi 2018: used 6 months data as it was used to analyze the primary outcomes and it was unclear if adverse events were recorded at later follow‐up times.

Guvendi 2018: 19 participants were allocated to single PRP group and 14 participants were allocated to three PRP group. For the analysis, these participants were combined as 33 participants who underwent PRP injections. There was one report of erythema, which was classified as “other musculoskeletal event” for our analysis.

Louis 2018: There were 1 case of post‐traumatic knee sprain and 1 case of amygdalotomy in the HA group. There was 1 case of post‐traumatic knee sprain in the PRP group. None of these were thought to be related to the study treatment. Nevertheless, they included in our analysis. Amygdalotomy was classified as “non‐musculoskeletal medical event” and post‐traumatic knee sprain was classified as “other musculoskeletal event”

Su 2018: IA‐PRP had 1 case of knee pain & swelling and 1 case of knee swelling that were thought to be related to the procedures by the author. These were combined as 3 cases of mild pain and swelling. IA‐CS group had 1 case of knee swelling and 1 case of knee pain. These were combined as 2 cases of mild pain and swelling. The other AE in the study were categorized as following: low back pain and other knee pain as “other musculoskeletal event”. Upper respiratory infection, bronchitis, cold, headache and acute MI as “non‐musculoskeletal medical event”. There was one report of traffic accident, but this was considered not appropriate to be considered as AE and was excluded form analysis. The group A in the study underwent both intraosseous and intra‐articular PRP injection and was not included in our analysis.

Yu 2018: Table 3 shows that the total number of AE was 30 for the HA group. However, there are only 25 cases when each AE is added together. For our analysis, we used 25 cases for the number of overall AE. Each AE was classified as following: hypertension, diarrhea, proteinuria, vomiting, fatigue, constipation, hypertriglyceridemia and edema peripheral as “non‐musculoskeletal medical event”, and rash as “other musculoskeletal event”.

Huang 2019: the placebo group used NS, which was not included in the analysis as it did not report adverse events in this arm of the study.

Elksnins‐Finogejevs 2020: there were 15 cases of mild synovitis that resolved spontaneously. They were classified as “mild pain & swelling” for our analysis.

Bennell 2022: The supplementary document shows detail of adverse events observed immediately following the injection, at 2 months and 12 months. 12 months data was used for the analysis. Other lower limb musculoskeletal symptoms and upper body musculoskeletal symptoms were classified as “other musculoskeletal event”, and medical condition (non‐musculoskeletal) was classified as “non‐musculoskeletal medical event” for our analysis.

Dorio 2021: Table 3 lists “application site pain”, “index knee swelling” and “application site pain and index knee swelling” separately. For analysis, these three were combined as mild pain and swelling.

Li 2021: Table 2 shows number of injections that resulted in AE, rather than number of patients that experienced patients. Each patient underwent three intra‐articular injections of either PRP or HA. For our analysis, the number of patients that experienced AE was estimated by multiplying number of patients and frequency of the AE. The leukocyte count was 32.3 × 10¹², which was reported to be 6.1 times greater than the level in the peripheral blood. Given that normal leukocyte count is 4.5–11 × 10⁹, we suspect that the leukocyte count report is mis‐reported and should be 32.3 × 10⁹. For the analysis, 32.3 × 10⁹ was used.

Park 2021: Table 5 lists AE in system organ class and preferred term. Classification based on preferred term was used for our analysis. Injection site pain and injection site swelling was combined and classified as “mild pain & swelling”. Musculoskeletal pain and backache were classified as “other musculoskeletal event”. Nasopharyngitis, constipation and headache were classified as “non‐musculoskeletal medical event”.

Tucker 2021: One patient developed hemarthrosis from the knee that underwent IA‐PRP. The knee was aspirated and the symptom resolved to a minimal level after several days. This event was classified as “mild knee pain & swelling” for our analysis.

Xu 2021: Table 3 lists joint swelling and pain after injection separately. These two AE types were combined as “mild pain & swelling” for our analysis.

Aakash 2022: Table 1 shows the breakdown of the age group, but not enough information is available to determine the mean age in each group. The study stated that there were 30 patients (15 male and 15 female) included in the study, but not enough information was available to determine the distribution of the sex in each group.

Nakagawa HF, Kim J, Rabinowitz J, Sussman WI. Assessment of adverse events and safety associated with intra‐articular platelet‐rich plasma injections compared to other injectates for knee osteoarthritis: A systematic review and meta‐analysis. PM&R. 2026;18(9):1072‐1084. doi: 10.1002/pmrj.70141

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.