Short-term outcomes of two PRP-based treatment protocols for Stoller grade I–II meniscal lesions: serial ultrasound-guided perimeniscal PRP injections versus arthroscopic management with a single intra-operative intra-articular PRP injection

Abstract

Objective

To describe and contrast short-term clinical outcomes associated with two routinely used treatment protocols for symptomatic Stoller grade I–II meniscal lesions: (i) serial ultrasound-guided perimeniscal PRP injections and (ii) arthroscopic management with a single intra-operative intra-articular PRP injection.

Methods

We retrospectively reviewed patients treated between September 2023 and December 2024. Protocol A comprised serial ultrasound-guided perimeniscal PRP injections (5 mL per session; three sessions over 6 weeks). Protocol B comprised arthroscopic meniscus management as indicated (debridement/repair) combined with a single intra-operative intra-articular PRP injection (5 mL). Outcomes were VAS, Lysholm, and Japanese Orthopedic Association (JOA) scores at baseline and at 1, 3, and 6 months. Linear mixed-effects models included treatment protocol, time, and their interaction as fixed effects and subject-specific random intercepts, with additional adjustment for age, sex, BMI, and Stoller grade.

Results

Forty patients were included (20 per protocol), and no serious complications occurred. Linear mixed-effects model analysis revealed that both groups demonstrated significant improvement over time. The VAS scores significantly decreased at all follow-ups (e.g., mean reduction of 4.85 points at 6 months; 95% CI: -5.15 to -4.55; P < 0.001), while Lysholm (mean increase of 26.2 points; 95% CI: 25.35 to 27.05) and JOA scores (mean increase of 21.55 points; 95% CI: 20.67 to 22.43) significantly increased (all P < 0.001). Analysis of the group × time interaction suggested different Lysholm trajectories between Protocol A and Protocol B, with higher model-estimated Lysholm scores for Protocol B at 6 months; these between-protocol contrasts are descriptive and hypothesis-generating.

Conclusion

In this retrospective cohort of symptomatic Stoller grade I–II meniscal lesions, both protocols were associated with substantial short-term improvements in pain and knee function over 6 months. Pain trajectories (VAS) did not differ between protocols, whereas Protocol B showed higher model-estimated functional scores (Lysholm and JOA) at 6 months. These between-protocol patterns are descriptive and hypothesis-generating only, because the protocols differed simultaneously in surgical intervention, PRP dosing schedule, injection route/target, and postoperative management. Therefore, the present design does not permit isolation of the independent effect of PRP itself or attribution of the observed differences to any single treatment component. Future adequately powered prospective trials with standardized PRP characterization and blinded outcome assessment should test whether an arthroscopy-based protocol yields greater functional recovery, while prioritizing prespecified functional and imaging endpoints.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12891-026-10078-7.

Introduction

Meniscal lesions, encompassing both acute traumatic lesions and chronic degenerative conditions, represent a prevalent cause of knee symptoms in clinical practice; worldwide, more than 1.7 million individuals undergo surgical intervention due to meniscal lesions annually [1]. The etiology is multifactorial, involving age, occupation, body weight, lifestyle, and sports-related injuries. In younger populations, acute trauma is a common cause, whereas in middle-aged and older individuals, chronic degenerative wear predominates. The intricate structure of the meniscus, coupled with its unique physiological environment, poses challenges to recovery post-injury. Meniscal lesions can lead to further damage to adjacent structures, resulting in knee pain, functional impairment of the knee joint, and potentially progressing to secondary knee osteoarthritis. Research indicates that the risk of developing secondary knee osteoarthritis following meniscal lesions increases threefold [2]. In recent years, the heightened awareness of sports has led to a significant rise in the incidence of meniscal lesions [1].

Currently, a variety of treatment options are available for meniscal injuries in clinical practice, each possessing distinct advantages and disadvantages. The primary treatment modalities include surgical or conservative approaches. The ESSKA consensus emphasizes that preservation of the meniscus should be the first-line treatment whenever possible, while surgical intervention is reserved for patients with mechanical symptoms or those who fail non-operative therapy [3]. Specifically, for lesions identified as Stoller grades I–II on MRI, conservative management is recommended, whereas full-thickness lesions (grade III) typically warrant surgical consideration. This staged approach is further supported by the AAOS guidelines, which advocate non-operative treatment as the initial strategy for non-full-thickness meniscal lesions (corresponding to Stoller I–II) [4]. Arthroscopic surgery is a minimally invasive procedure typically performed in a hospital setting. It involves making small incisions to insert an arthroscope and surgical instruments into the knee joint. This allows for direct visualization and treatment of meniscal lesions, such as repair or partial meniscectomy. The efficacy of arthroscopic surgery as an intervention for meniscal lesions is supported by prior clinical evidence [5, 6]. With advancements in sports and regenerative medicine, platelet-rich plasma (PRP) has emerged as a novel therapeutic option for meniscal lesions and has been increasingly utilized [7, 8].

However, the current evidence supporting the use of platelet-rich plasma (PRP) for meniscal lesions remains limited in quality [9]. In routine practice, PRP may be delivered using different dosing schedules and injection targets, and selected patients may undergo arthroscopic management with intra-operative PRP. Therefore, this retrospective cohort study reports and contrasts short-term outcomes observed under two routinely applied protocols in our center: serial ultrasound-guided perimeniscal PRP injections and arthroscopic management with a single intra-operative intra-articular PRP injection. Given the non-randomized design and multi-component protocol differences, the analysis is exploratory and hypothesis-generating.

Materials and methods

General information

This was a retrospective, non-randomized cohort study (Level of Evidence: III). From September 2023 to December 2024, 61 patients with symptomatic Stoller grade I–II meniscal lesions who received one of the two routinely applied treatment protocols at the Pain Management Center and the Department of Orthopedic Sports Medicine of the People’s Hospital of Xinjiang Uygur Autonomous Region were retrospectively reviewed. According to the treatment received, 32 patients were initially identified in Protocol A, consisting of serial ultrasound-guided perimeniscal PRP injections (5 mL per session, three sessions over 6 weeks), and 29 patients were initially identified in Protocol B, consisting of arthroscopic meniscus management with a single intra-operative intra-articular PRP injection (5 mL). In Protocol A, 12 patients were excluded because of loss to follow-up or lack of final follow-up imaging data. In Protocol B, 9 patients were excluded for the same reasons. Finally, 40 patients with complete 6-month clinical and imaging follow-up data were included in the final analysis, with 20 patients in each protocol. Treatment allocation was not randomized; therefore, selection/indication bias and residual confounding cannot be excluded (see Limitations). The detailed patient selection process is shown in Fig. 1.

Fig. 1.

Schematic for the retrospective study design

Inclusion criteria were as follows: (1) Participants must meet the diagnostic criteria for meniscus lesions as outlined in " 2018 APTA Clinical Practice Guidelines“ [2], irrespective of gender or whether the left or right knee is affected; (2) Participants must present with clinical symptoms correlating to the MRI-identified lesion, evidenced by at least two of the following: joint line tenderness, mechanical locking/catching, a positive McMurray test, or pain on deep knee flexion; (3) MRI examination must confirm the presence of an intrasubstance meniscal signal change (Stoller grade I or II) on the symptomatic side [10]. (In the context of this investigation, Stoller grade I-II signals were deemed to be representative of degenerative meniscal pathology, which encompasses mucoid degeneration or early myxoid transformation); (4) For patients treated under Protocol B (arthroscopic repair with PRP), an additional inclusion criterion was applied: the presence of persistent symptoms or functional limitations after a minimum of 2 months of structured conservative management, which included activity modification, supervised physical therapy, and a trial of oral non-steroidal anti-inflammatory drugs (NSAIDs).

Exclusion criteria included: (1) Patients with a history of knee surgery within the past 6 months, radiographic osteoarthritis (Kellgren–Lawrence grade ≥1), concurrent cruciate ligament injury, or other knee joint injuries were excluded; (2) Patients with systemic conditions that could impair healing or confound outcomes—including uncontrolled diabetes (HbA1c >7.0%), rheumatic diseases, severe cardiovascular or renal dysfunction, active infection, immunosuppression, or hematologic disorders—were excluded. Additionally, patients with body mass index (BMI) >30 kg/m² or significant lower limb malalignment (e.g., varus/valgus >5° on standing radiographs) were excluded to minimize biomechanical confounding factors; (3) Patients who had received other pharmacological treatments, such as anticoagulant therapy within the past week, nonsteroidal anti-inflammatory drugs, or intra-articular injections of corticosteroids or hyaluronic acid within the past two weeks, were also excluded; (4) Additionally, patients with cognitive impairments, mental disorders, malignancies, or those unwilling to cooperate with clinical observations were not included.

All patients underwent standardized baseline knee radiographic evaluation as part of routine assessment (weight-bearing anteroposterior and lateral views). Radiographic osteoarthritis severity was graded using the Kellgren–Lawrence (KL) system based on these baseline radiographs. Only patients with KL grade 0 were eligible, and those with KL grade ≥ 1 were excluded to reduce OA-related confounding on pain and functional outcomes.

Treatment methods

Preparation of Platelet-Rich Plasma (PRP)

In this study, PRP was prepared using a double centrifugation technique. The Shanghai Lu Xiangyi blood cell separator and a disposable PRP preparation tube containing sodium citrate as an anticoagulant were employed to produce fresh autologous platelet-rich plasma(P-PRP). A volume of 15 ml of autologous venous blood was collected from the basilic vein of the left upper limb by nursing staff. The double centrifugation method was utilized to isolate the platelets. The initial centrifugation was conducted at a radius of 15.5 cm and a speed of 1200 revolutions per minute (rpm) for a duration of 10 min. Post-centrifugation, the blood was stratified into three distinct layers: the upper serum layer, the lower red blood cell layer, and the intermediate buffy coat, which contains platelets, white blood cells, and other components. The red blood cells located beneath the buffy coat were carefully aspirated using a syringe and subsequently discarded. The remaining portion of the sample was subjected to a second centrifugation process using a centrifuge with a radius of 15.5 cm, operating at a speed of 2800 revolutions per minute for a duration of 10 min. Post-centrifugation, the sample stratified into two distinct layers: the upper layer, identified as platelet-poor plasma (L-PPP), and the lower layer, designated as platelet-rich plasma (P-PRP). The PPP was carefully aspirated using a syringe, leaving behind 5 ml of PRP for injection purposes.

PRP characterization and classification

The cellular composition of the final PRP product was quantitatively analyzed using an automated hematology analyzer (Sysmex XN-series). Based on the analysis of prepared samples, the mean platelet concentration in the injected PRP was approximately 1200 × 10⁹/L, representing a mean 4- to 5-fold increase over the baseline peripheral blood platelet count (approximately 250 × 10⁹/L). The leukocyte concentration was substantially reduced, confirming a leukocyte-poor profile. According to the widely accepted PAW (Platelets, Activation, White blood cells) classification system [11], the prepared product, characterized by a high platelet concentration, low leukocyte concentration, and the absence of exogenous activation, is classified as Pure Platelet-Rich Plasma (P-PRP).

Injection protocol for the Protocol A

Patients were positioned supine with the affected knee exposed, externally rotated, and flexed. Following standard disinfection and draping, ultrasound imaging was used to localize the meniscus and assess for joint effusion. If present, effusion was aspirated prior to injection to avoid dilution of the PRP concentrate. Under continuous sonographic visualization, the needle tip was advanced to the perimeniscal region, and 5 mL of PRP was injected accurately into the target site. Following the injection, the needle was withdrawn, and the puncture site was covered with a sterile dressing. Patients were advised to refrain from weight-bearing activities, squatting, and running for 48 h post-injection and to avoid vigorous knee joint movements for a period of two weeks. PRP injections were administered biweekly over a period of 6 weeks, resulting in a total of three sessions.

Arthroscopic meniscus repair in the Protocol B

Patients were positioned supine. All procedures were performed under epidural and subarachnoid block anesthesia. Arthroscopy was performed using the Smith & Nephew (USA) visualization system. Meniscal lesions were assessed arthroscopically, and tear characteristics (location and morphology) were documented based on intraoperative findings. A standardized intraoperative decision approach was applied: meniscal repair was preferred for peripheral, reducible lesions with adequate tissue quality, whereas limited debridement/partial meniscectomy was performed for non-repairable lesions (e.g., complex degeneration or insufficient tissue quality), including cases in which the anterior horn was not amenable to repair. Unstable tissue was debrided using a mechanical shaver while preserving stable meniscal tissue. When repair was indicated, a standardized all-inside technique was used, with 2–3 vertical sutures placed according to lesion size and stability. After wound closure and prior to bandaging, 5 mL of leukocyte-poor PRP (P-PRP) was injected intra-articularly under arthroscopic visualization. Residual irrigation fluid was aspirated before PRP injection to minimize dilution. A compressive dressing was then applied.

Post-procedure rehabilitation and activity guidance

A standardized, protocol-based rehabilitation framework was recommended to all patients, with procedure-specific precautions applied where appropriate. For Protocol A, patients were advised to refrain from weight-bearing activities, squatting, and running for 48 h after each injection and to avoid vigorous knee joint movements for two weeks; thereafter, progressive range-of-motion exercises, quadriceps strengthening, and a gradual return to activities were encouraged as tolerated. For Protocol B, postoperative rehabilitation followed our institutional pathway and was tailored to the arthroscopic procedure: after debridement/partial meniscectomy, early range-of-motion and weight-bearing as tolerated were encouraged; after meniscal repair, protected weight-bearing and early motion precautions were recommended during the initial phase before progressive strengthening and return to activity. In both protocols, the core elements included early range-of-motion exercises, progressive quadriceps strengthening, and gradual activity resumption. Compliance was encouraged through verbal instructions and a provided handout; however, rehabilitation exposure and adherence were not formally monitored.

Efficacy evaluation

The efficacy of MT treatment was assessed utilizing the VAS, Lysholm, and Japanese Orthopedic Association (JOA) criteria at baseline and at 1, 3, and 6 months post-treatment. The VAS, Lysholm, and JOA scores were collected at each time point by the same clinical research personnel. Due to the retrospective nature of the study and the clinical context of treatment delivery, this assessor was not blinded to the patient’s treatment protocol. We acknowledge this as a potential source of bias in the Limitations section.

MRI examinations were performed before treatment and at the final follow-up (6 months). All MRI scans were reviewed independently by two experienced musculoskeletal radiologists from our institution. Because of the retrospective workflow and the availability of treatment information in the medical record, the radiologists were not formally blinded to protocol allocation; however, they were not involved in clinical outcome scoring. Assessment was based on direct comparison of pre- and post-treatment images with a focus on changes in meniscal signal intensity, continuity, and morphology at the lesion site. Findings were described qualitatively using consistent terminology (e.g., “signal reduction,” “less conspicuous intrameniscal high signal,” “improved contour,” “persistent high signal”). Any discrepancies between readers were resolved through consensus discussion. Formal inter-reader agreement statistics (e.g., κ) were not calculated, as MRI findings were collected for supportive, illustrative purposes and were not treated as outcome-defining endpoints or analyzed as comparative outcomes; clinical scores (VAS, Lysholm, and JOA) remained the primary outcomes of this study.

Statistical methods

Data analysis was conducted using R project statistical computing (version 4.3.2). Linear mixed-effects models (LMMs) were used to analyze repeated measures data. Dependent variables were VAS, Lysholm, and JOA scores measured at baseline (0 months) and at 1, 3, and 6 months. Fixed effects included time (categorical; baseline as the reference), protocol (Protocol A as the reference), and the protocol × time interaction term to describe differential trajectories between protocols. A subject-specific random intercept was included to account for within-subject correlation. For each outcome, Model 1 was unadjusted and Model 2 additionally adjusted for age, sex, BMI, and Stoller grade. Residual diagnostics were used to assess model assumptions. Post-hoc time-point contrasts were reported without formal multiplicity adjustment and interpreted cautiously given the exploratory retrospective design.

Patient characteristics

A total of 40 patients were included in the final analysis and classified according to the treatment received: Protocol A (n = 20) and Protocol B (n = 20). The baseline demographic and clinical characteristics of both protocols are detailed in Table 1. No significant between-protocol differences were observed in age, sex distribution, Stoller grade (I vs. II), or body mass index (BMI) (all P > 0.05).

Table 1.

Comparison of baseline between the two protocols

Protocol	Number of Cases	Age (years, x ± s)	Gender (Male/Female)	Stoller Grading (I/II)	BMI (kg/m², x ± s)
Protocol A	20	42 ± 6	13/7	12/8	26.7 ± 1.3
Protocol B	20	45 ± 7	15/5	11/9	27.2 ± 0.9
Test Values		t = -1.450	χ² = 0.476	χ² = 0.102	t = -1.388
P Values		>0.05	>0.05	>0.05	>0.05

Continuous variables are presented as mean ± SD and categorical variables as n. Baseline between-protocol comparisons used Student’s t test (t) for continuous variables and Pearson’s χ² test (χ²) for categorical variables. All P values are two-sided; for brevity, P values > 0.05 are shown as > 0.05

Results

All participants successfully completed the 6-month follow-up period, during which assessments of pain and knee joint function were performed for all patients. For transparency, the observed (raw) VAS, Lysholm, and JOA values at baseline and each follow-up time point are summarized as mean ± SD with the number of available observations (n) in Supplementary Table 1. During the course of treatment, 4 patients in total experienced mild post-treatment joint swelling and restricted mobility, which resolved within 48 h with the application of local ice packs, activity limitations, or oral analgesics. No other patients encountered significant complications. Within Protocol B (n = 20), meniscal repair was performed in 16 patients, whereas partial meniscectomy was performed in 4 patients. Because these procedures are biologically and clinically distinct and involve different rehabilitation constraints, Protocol B should be interpreted as a heterogeneous composite arthroscopic strategy rather than a uniform surgical intervention. The partial meniscectomy subgroup was very small and markedly imbalanced relative to the repair subgroup; therefore, neither formal subgroup analysis nor subgroup-level descriptive outcome reporting was considered reliable. Accordingly, the observed Protocol B outcome pattern should be interpreted as reflecting a mixed surgical cohort predominantly composed of repair cases, and whether similar short-term outcome patterns apply to partial meniscectomy cases remains uncertain.

Qualitative MRI comparisons at 6 months were used as supportive/illustrative information and demonstrated descriptive changes in representative cases (e.g., reduced intrameniscal high-signal intensity and/or a more regular meniscal contour), as shown in Figs. 2 and 3. No validated MRI scoring system was applied, and MRI findings were not used as outcome-defining endpoints.

Fig. 2.

Representative imaging findings for Protocol A: (A) weight-bearing anteroposterior radiograph; (B) baseline MRI (coronal) showing an intrameniscal high-signal line (arrow, Stoller grade I); (C) 6-month MRI showing less conspicuous intrameniscal high signal and a more regular-appearing meniscal contour (arrow); (D) PRP preparation schematic; (E) ultrasound localization (triangle); (F) ultrasound-guided perimeniscal PRP injection

Fig. 3.

Representative imaging findings for Protocol B (arthroscopic management with intra-operative intra-articular PRP): (A) baseline MRI (sagittal) showing intrameniscal high signal (arrow, Stoller grade II); (B) 6-month MRI showing less conspicuous intrameniscal high signal and a more regular-appearing meniscal contour (arrow); (C–E) arthroscopic views (lesion identification, suture placement, final repair)

The results of the two models were highly consistent, yielding consistent model estimates across specifications; results should be interpreted as exploratory given the retrospective design and limited sample size. Importantly, all between-protocol contrasts (including the group × time interaction terms) are presented to describe observed outcome trajectories. Because the protocols differed simultaneously in surgical intervention and PRP administration (dose and injection route/target), these contrasts should not be interpreted as evidence of comparative effectiveness or superiority of any single treatment component. Detailed findings are as follows:

VAS scores decreased over time in both protocols (time effect: P < 0.001). Relative to baseline, the model-estimated change in VAS was − 3.7 at 1 month (95% CI − 4.002 to − 3.398), − 4.5 at 3 months (95% CI − 4.802 to − 4.198), and − 4.85 at 6 months (95% CI − 5.152 to − 4.548). No evidence of differential change in VAS over time was observed between Protocol A and Protocol B (Table 2).

Table 2.

Comparison of VAS scores between the two protocols

Term	Model 1		Model 2
	β (95% CI)	P value	β (95% CI)	P value
Time (0 month as reference)
1 month	-3.7 (-4.002, -3.398)	< 0.001	-3.7 (-4.002, -3.398)	< 0.001
3 months	-4.5 (-4.802, -4.198)	< 0.001	-4.5 (-4.802, -4.198)	< 0.001
6 months	-4.85 (-5.152, -4.548)	< 0.001	-4.85 (-5.152, -4.548)	< 0.001
Protocol B (Protocol A as reference)	0.2 (-0.182, 0.582)	0.302	0.237 (-0.15, 0.624)	0.227
1 month: Protocol B	-0.05 (-0.477, 0.377)	0.817	-0.05 (-0.477, 0.377)	0.817
3 months: Protocol B	0.15 (-0.277, 0.577)	0.488	0.15 (-0.277, 0.577)	0.488
6 months: Protocol B	-0.35 (-0.777, 0.077)	0.107	-0.35 (-0.777, 0.077)	0.107

Model 1: unadjusted; Model 2: adjusted for age, sex, BMI, and Stoller grade. β indicates the fixed-effect estimate; CI, confidence interval; P values are two-sided. Baseline (0 month) and Protocol A are reference categories

Lysholm scores increased over time in both protocols (time effect: all P < 0.001). Relative to baseline, the model-estimated Lysholm score increased by 11.45 points at 1 month (95% CI 10.601 to 12.299), by 19.5 points at 3 months (95% CI 18.651 to 20.349), and by 26.2 points at 6 months (95% CI 25.351 to 27.049). The group-by-time interaction identified different Lysholm trajectories between Protocol A and Protocol B, with higher model-estimated Lysholm scores for Protocol B at 6 months (β = 2.3, 95% CI 1.099 to 3.501, P < 0.001) (Table 3).

Table 3.

Comparison of Lysholm scores between the two protocols

Term	Model 1			Model 2
	β (95% CI)	P value		β (95% CI)	P value
Time (0 month as reference)
1 month	11.45 (10.601, 12.299)		< 0.001	11.45 (10.601, 12.299)	< 0.001
3 months	19.5 (18.651, 20.349)		< 0.001	19.5 (18.651, 20.349)	< 0.001
6 months	26.2 (25.351, 27.049)		< 0.001	26.2 (25.351, 27.049)	< 0.001
Protocol B (Protocol A as reference)	-3.3 (-6.48, -0.12)		0.042	-3.292 (-6.311, -0.274)	0.033
1 month: Protocol B	1.25 (0.049, 2.451)		0.041	1.25 (0.049, 2.451)	0.041
3 months: Protocol B	1.6 (0.399, 2.801)		0.009	1.6 (0.399, 2.801)	0.009
6 months: Protocol B	2.3 (1.099, 3.501)		< 0.001	2.3 (1.099, 3.501)	< 0.001

Model 1: unadjusted; Model 2: adjusted for age, sex, BMI, and Stoller grade. β indicates the fixed-effect estimate; CI, confidence interval; P values are two-sided. Baseline (0 month) and Protocol A are reference categories

JOA scores increased over time in both protocols (time effect: all P < 0.001). Relative to baseline, the model-estimated change in JOA was 11.05 points at 1 month (95% CI 10.17 to 11.93), 20.05 points at 3 months (95% CI 19.17 to 20.93), and 21.55 points at 6 months (95% CI 20.67 to 22.43). At 6 months, the model-estimated JOA score was higher in Protocol B than in Protocol A (β = 3.4, 95% CI 2.156 to 4.644, P < 0.001) (Table 4).

Table 4.

Comparison of JOA scores between the two protocols

Term	Model 1			Model 2
	β (95% CI)	P value		β (95% CI)	P value
Time (0 month as reference)
1 month	11.05 (10.17, 11.93)		< 0.001	11.05 (10.17, 11.93)		< 0.001
3 months	20.05 (19.17, 20.93)		< 0.001	20.05 (19.17, 20.93)		< 0.001
6 months	21.55 (20.67, 22.43)		< 0.001	21.55 (20.67, 22.43)		< 0.001
Protocol B (Protocol A as reference)	-2.95 (-5.982, 0.082)		0.056	-2.881 (-5.781, 0.018)		0.051
1 month: Protocol B	-0.35 (-1.594, 0.894)		0.578	-0.35 (-1.594, 0.894)		0.578
3 months: Protocol B	-0.4 (-1.644, 0.844)		0.525	-0.4 (-1.644, 0.844)		0.525
6 months: Protocol B	3.4 (2.156, 4.644)		< 0.001	3.4 (2.156, 4.644)		< 0.001

Model 1: unadjusted; Model 2: adjusted for age, sex, BMI, and Stoller grade. β indicates the fixed-effect estimate; CI, confidence interval; P values are two-sided. Baseline (0 month) and Protocol A are reference categories

Discussion

This retrospective cohort study described short-term outcomes observed under two routinely used PRP-based protocols for symptomatic Stoller grade I–II meniscal lesions. Both protocols were associated with substantial improvements in pain and knee function during the 6-month follow-up period. VAS trajectories were similar between protocols, whereas model-estimated Lysholm and JOA scores were higher in Protocol B at 6 months. These between-protocol findings should be interpreted cautiously. Because treatment allocation was not randomized and the protocols differed simultaneously in surgical intervention, PRP exposure, injection route/target, and postoperative rehabilitation pathway, the present data do not establish comparative effectiveness or superiority of any single treatment component. More importantly, this design does not allow the independent effect of PRP itself to be isolated from the effects of arthroscopic intervention, injection strategy, or rehabilitation-related differences. Current evidence suggests that PRP may have a role both as an adjunct to arthroscopic meniscal procedures and as injection-based treatment for selected meniscal pathology [7, 8, 12–15]. Adjunct/augmentation PRP after repair has been linked to better knee function/mobility, higher healing rates, and lower reoperation in cohort studies and randomized trials [16–19]. Meta-analyses likewise report reduced repair failure and improved outcomes with PRP augmentation versus repair alone, although effects vary across PRP preparations and indications [20, 21]. PRP monotherapy is also emerging; prospective ultrasound-guided perimeniscal injection studies report pain relief and functional gains with supportive MRI changes [22–24]. Our findings add a pragmatic clinical perspective by describing short-term trajectories under two real-world protocols.

From a practical clinical perspective, these two protocols may be better viewed as options for different patient profiles rather than as directly interchangeable alternatives. Serial ultrasound-guided perimeniscal PRP may be considered for patients with relatively stable degenerative-type intrameniscal lesions who prefer a non-operative strategy. By contrast, arthroscopic management with intra-operative PRP may be considered in patients with persistent symptoms despite structured conservative care or with mechanical complaints that warrant operative evaluation. This clinical framing reflects routine selection in our center and should not be interpreted as a formal treatment algorithm. At the same time, because patients with BMI > 30 kg/m² or marked lower-limb malalignment were excluded, the present findings apply to a selected cohort and may not be generalizable to broader patient populations [25].

The favorable trajectory observed in Protocol A may plausibly relate to accurate ultrasound-guided delivery to the perimeniscal target and, when present, aspiration of joint effusion before injection to reduce dilution [26]. Our PRP preparation parameters were also aligned with prior recommendations: we used leukocyte-poor PRP because of its lower synovitis risk for intra-articular use [27], and selected a target platelet concentration of 4–6 times baseline with a 5-mL injection volume based on pragmatic and biologic considerations reported in the literature [28]. The apparently higher 6-month functional scores in Protocol B may also reflect the ability of arthroscopy to address intra-articular mechanical pathology in selected patients. However, the magnitude of the model-estimated between-protocol difference was small. For Lysholm score, the 6-month interaction estimate was 2.3 points, which is below reported MCID thresholds for Lysholm-based knee outcomes in meniscal populations and related knee conditions; therefore, this statistically significant difference should not be interpreted as clear evidence of clinically meaningful superiority. For the JOA knee score, an established MCID threshold for this specific population is not available, further limiting clinical interpretation. In addition, the observed raw values in the supplementary table are unadjusted cross-sectional summaries at each time point, whereas the linear mixed-effects model estimates the overall longitudinal trajectory while accounting for within-subject correlation and covariates. For the Lysholm score, the unadjusted 6-month mean in Supplementary Table 1 was slightly higher in Protocol A than in Protocol B, whereas the adjusted mixed-effects model estimated a modestly higher 6-month value for Protocol B. This directional discrepancy likely reflects the lower baseline Lysholm score in Protocol B; after the longitudinal model incorporated baseline imbalance, repeated measurements, and prespecified covariates, Protocol B showed a steeper estimated recovery trajectory, leading to a higher adjusted estimate in the model space despite the marginally lower raw 6-month mean. Therefore, small model-estimated trajectory differences may not be visually prominent in the raw 6-month means. Because Protocol B also differed in surgical procedure, PRP route, total PRP exposure, and postoperative rehabilitation pathway, this interpretation remains tentative. Mild post-procedural swelling or pain was self-limited, and no serious complications occurred, supporting the short-term feasibility of both protocols in this selected cohort.

This study has several limitations. First, its retrospective, non-randomized design introduces potential selection and indication bias, and residual confounding cannot be excluded. Second, between-protocol contrasts were confounded by design because the groups differed simultaneously in surgical intervention, PRP dosing schedule/total exposure, and injection route/target. In addition, Protocol B should be interpreted as a heterogeneous composite surgical strategy because it included both meniscal repair and partial meniscectomy, which differ in biological rationale, expected recovery, and rehabilitation constraints. Because only 4 patients underwent partial meniscectomy, procedure-specific descriptive outcomes would be unstable and potentially misleading; therefore, no separate subgroup outcome estimates were reported. Third, rehabilitation exposure and adherence were not formally monitored, which is particularly relevant for function-based outcomes such as Lysholm and JOA scores. Fourth, the cohort was restricted to selected patients with Stoller grade I-II lesions and excluded those with BMI > 30 kg/m² or marked malalignment, which limits generalizability. MRI assessment was descriptive rather than based on a validated scoring system, was not formally blinded, and did not include inter-reader agreement statistics; therefore, MRI findings should be considered supportive only rather than outcome-defining evidence. Fifth, the sample size was small, limiting precision and increasing the risk of both type I and type II errors. Although statistically significant between-protocol differences in Lysholm and JOA scores were observed at 6 months in the model-based analysis, the absolute magnitudes were modest. In particular, the Lysholm difference was likely below reported MCID thresholds for Lysholm-based knee outcomes, and the clinical importance of the JOA difference remains uncertain because no established MCID threshold is available for this specific population. These findings should therefore be interpreted as small descriptive trajectory differences rather than clinically meaningful evidence of protocol superiority. Finally, although the PRP preparation method and classification were reported, full cohort-level quantitative characterization of PRP composition was not systematically available, which may limit reproducibility and mechanistic interpretation. Future prospective randomized studies should standardize PRP reporting, include blinded and validated imaging and clinical endpoints, monitor rehabilitation exposure, and stratify analyses by lesion phenotype and surgical procedure.

Conclusion

In this retrospective cohort of symptomatic Stoller grade I–II meniscal lesions, both serial ultrasound-guided perimeniscal PRP injections (Protocol A) and arthroscopic management with a single intra-operative intra-articular PRP injection (Protocol B) were associated with short-term improvements in pain and knee function. Pain trajectories (VAS) were similar between protocols, while Protocol B showed higher model-estimated functional scores (Lysholm and JOA) at 6 months. Because the study was non-randomized and the protocols differed concurrently in surgical intervention, PRP dose/schedule, injection route/target, and postoperative management, these findings are descriptive and hypothesis-generating rather than evidence of comparative effectiveness. The study design does not permit isolation of the effect of PRP itself or any other individual component of the treatment protocols. Future prospective trials should standardize PRP reporting, prespecify function-focused primary endpoints, include pain, return-to-activity, reintervention and blinded MRI outcomes, stratify by Stoller grade/phenotype, and power sample sizes using effect estimates from this cohort.

Authors’ contributions

PCZ and HPL developed the research questions and scope of the study. XM and FA conducted baseline and post-treatment data screening, and data charting. PCZ drafted the manuscript, and prepared tables, and figures with HPL’s contribution. XM developed the literature search strategies in collaboration with the other authors. PCZ, HPL, XM, and FA contributed to the organization, analysis, and interpretation of the results. All authors read and approved the final manuscript.

Funding information

This study was not funded by any foundation.

Data availability

The data sets generated and analyzed during the current study are not publicly available due to restrictions on ethical approvals involving patient data and anonymity but can be obtained from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This retrospective study was approved by the Ethics Committee of the People’s Hospital of Xinjiang Uygur Autonomous Region and carried out by the ethical standards set out in the Helsinki Declaration. Informed consent was received from all participants. Ethical batch number: YK20251013.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.