Introduction

Knee osteoarthritis (KOA) is a prevalent joint disease involving new bone formation, focal cartilage loss, and bone enlargement. KOA can lead to usage-related pain and/or functional limitation. Therefore, people with KOA are at risk of mobility disability [1]. KOA was consistently ranked among the ten most frequently hospitalized diseases in Korean medicine from 2013 to 2021. Furthermore, the number of patients, number of visitation days, and total amount of medical care costs in Korean medicine hospitalizations have steadily increased during the last three years (2021–2023) [2]. Therefore, KOA incurs significant social and economic costs.

According to the American College of Rheumatology (ACR) guideline for the management of osteoarthritis, non-pharmacologic therapies such as exercise, weight loss, tai chi, and acupuncture and pharmacologic therapies such as acetaminophen, oral non-steroidal anti-inflammatory drugs (NSAIDs) and topical NSAIDs can be considered for the management of KOA [3]. If the symptoms persist and cannot be relieved by non-operative treatment, operative treatments, such as total knee replacement arthroplasty (TKRA), may be considered [4].

Low-level laser (LLL) therapy, a non-invasive treatment, is known to relieve pain and improve function through its anti-inflammatory effect [5,6]. Some systematic reviews and meta-analyses have reported that LLL may be effective in treating KOA [7–9]. Laser acupuncture (LA), a branch of non-invasive LLL therapy, stimulates acupoints through the logic of traditional Chinese medicine by LLL [10]. Several systematic reviews have suggested that LA improves pain and neuropathy [11,12]. Invasive laser acupuncture (ILA) combines acupuncture—a modality conditionally recommended for KOA in the ACR guideline [3]—with low-level laser therapy. It is performed by invasive irradiation and acupuncture stimulation using an optical fiber-coupled laser diode (InGaAlP 650 nm; GaAlAs 830 nm) inserted inside the needle (material, stainless steel; length, 30 mm; inner diameter, 0.15 mm; external diameter, 0.3 mm). A previous study demonstrated that the 650 nm ILA restricted the formation of inflammatory mediators in osteoarthritis [13]. Additionally, ILA for chronic non-specific low back pain can reduce pain and improve functional disability [14].

However, no previous studies have evaluated ILA for KOA. Therefore, this exploratory pilot study aimed to explore wavelength-dependent differences in the safety and efficacy of ILA (650 and 830 nm) compared with a sham laser control in patients with KOA. This pilot study provides a basis for future research on the use of ILA in KOA.

Materials and methods

The study design and reporting followed the Standard Protocol Items of the Recommendations for Interventional Trials (SPIRIT) statements [15] and Consolidated Standards of Reporting Trials (CONSORT) guidelines [16]. The methodology of this study has been reported in a previous study [17].

Study design

We conducted a randomized, parallel-arm, patient- and outcome assessor-blinded, placebo-controlled, single-center prospective pilot clinical trial (Dongshin University Gwangju Korean Medicine Hospital, Republic of Korea). A total of 45 participants who were judged eligible for the study were randomly assigned to the 650 group (650 nm wavelength ILA), 830 group (830 nm wavelength ILA), or control group (Sham ILA) at a ratio of 1:1:1. The participants were treated with ILA on one-side (more affected side) of SP9 (Yinglingquan; Spleen 9), SP10 (Xuehai; Spleen 10), ST34 (Liangqiu; Stomach 34), ST35 (Dubi; Stomach 35), EX-LE2 (Heding; Extra Lower Extremity 2), EX-LE4 (Neixiyan; Extra Lower Extremity 4), and GB34 (Yanglingquan; Gallbladder 34) for 10 minutes per session, twice weekly, for a total of 12 sessions over 6 weeks. Measurements for the outcomes, except the patient’s global assessment (PGA) and dosage of rescue medication, were carried out at baseline (Visit 1), after six treatments (Visit 7), 1 week after intervention completion (Visit 13), and 6 weeks after intervention completion (Visit 14). Other variables were measured at Visits 7, 13, and 14. A summary of this study is presented in Table 1.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Standard protocol items: Recommendations for interventional trials (SPIRIT) statement.

https://doi.org/10.1371/journal.pone.0353654.t001

Results

Participants

During the screening, of the 89 participants evaluated for eligibility, 44 were excluded. As a result, 45 participants were accepted as trial participants and were randomly divided into three groups (control group, 650 group, 830 group) of 15 each. During the treatment phase, 5 participants failed to finish the treatment with three dropouts in the control group, one in the 650 group, and one in the 830 group (Fig 1).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. CONSORT 2010 Flow Diagram.

https://doi.org/10.1371/journal.pone.0353654.g001

Baseline characteristics

The demographic characteristics of the three groups are summarized in Table 2. No significant differences were observed among the three groups, except for body mass index (BMI; p = 0.0072).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Baseline characteristics of participants at baseline.

https://doi.org/10.1371/journal.pone.0353654.t002

Efficacy of the outcomes

To account for the baseline imbalance in BMI, we conducted ANCOVA for all outcome measures, using BMI as a covariate to minimize potential bias and ensure the robustness of the findings.

After completion of the intervention and 6 weeks later, we observed significant differences in changes in the VAS scores both at rest (p = 0.0044, Visit 13 – Visit 1; p = 0.0001, Visit 14 – Visit 1) and during activity (p = 0.0038, Visit 13 – Visit 1; p < 0.0001, Visit 14 – Visit 1) among the three groups (Table 3). The distributions of individual changes in the VAS scores at the primary endpoint (Visit 13) are shown in Fig 2, and the detailed scores at Visits 1, 7, 13, and 14 are provided in S2A Table.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Changes in the VAS scores (Visit 7 vs. Visit 1, Visit 13 vs. Visit 1, Visit 14 vs. Visit 1) in the three groups.

https://doi.org/10.1371/journal.pone.0353654.t003

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Raincloud plots of the change in VAS scores between Visit 13 and Visit 1.

(A) VAS at rest; (B) VAS during activity. For each group (control, 650 nm, and 830; n = 15 per group), individual participant values (left), a box plot showing the median and interquartile range with whiskers (center), and a half-violin density estimate (right) are presented. More negative values indicate greater pain reduction. VAS, visual analog scale.

https://doi.org/10.1371/journal.pone.0353654.g002

Additionally, the WOMAC total score (p = 0.0278, Visit 7 – Visit 1; p = 0.0485, Visit 13 – Visit 1; p = 0.0051, Visit 14 – Visit 1), WOMAC pain subscale (p = 0.0073, Visit 7 – Visit 1; p = 0.0210, Visit 13 – Visit 1; p = 0.0038, Visit 14 – Visit 1), WOMAC function subscale (p = 0.0055, Visit 14 – Visit 1), and EQ-5D-5L (p = 0.0131, Visit 14 – Visit 1) were significantly different among the three groups (Table 4). The detailed scores at Visits 7, 13, and 14 are provided in S2A Table.

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Changes in WOMAC scores and EQ-5D-5L (Visit7 vs. Visit1, Visit13 vs. Visit1, Visit14 vs. Visit 1) in the three groups.

https://doi.org/10.1371/journal.pone.0353654.t004

The PGA (p = 0.0174, Visit 14 – Visit 7) was significantly different and rescue medication dosages were not significantly different among the three groups (Table 5). In addition, responder rate was not significantly different among the three groups (Table 6).

Download:

• PNG
  larger image
• TIFF
  original image

Table 5. Changes in the PGA and dosage of rescue medication (Visit13 vs. Visit7, Visit14 vs. Visit7) in the three groups.

https://doi.org/10.1371/journal.pone.0353654.t005

Download:

• PNG
  larger image
• TIFF
  original image

Table 6. Responder rate in one week after intervention completion.

https://doi.org/10.1371/journal.pone.0353654.t006

Details of PGA and dosage of the rescue medication at Visits 7, 13, and 14 are provided in S2B Table.

A supplementary repeated-measures ANCOVA of efficacy outcomes is presented in S3 Table.

Discussion

We conducted the first RCT of ILA to investigate its safety and efficacy for KOA. This study provides a preliminary foundation for future research on this topic.

Our trial revealed several important findings. First, the 650 and 830 groups showed a significant reduction in VAS score at 1 week after intervention completion and until 6 weeks later. Second, the 650 and 830 groups showed a significant reduction in the WOMAC scores (total and pain and function subscales) 6 weeks after intervention completion. Third, a significant improvement in the EQ-5D-5L score at 6 weeks after intervention completion was observed in both the 650 and 830 groups. Fourth, no significant differences in the vital signs and laboratory test results were observed among the three groups, and no ILA-related AEs were reported.

The reasons behind the reduction in the VAS score and WOMAC total scores are likely attributed to the analgesic and anti-inflammatory effects of photobiomodulation (PBM). PBM exerts several effects on cells by therapeutically exposing cells to low levels of red or near-infrared light. First, in the intracellular compartment, PBM light is thought to modulate cytochrome c oxidase activity in mitochondria, thereby increasing adenosine triphosphate production and modulating intracellular levels of reactive oxygen species. Through this pathway, PBM has been reported to modulate the production of key pro-inflammatory mediators, including prostaglandin E2, interleukin-6, and tumor necrosis factor-alpha. Second, at the cell membrane, PBM may modulate ion channels such as transient receptor potential vanilloid 1, thereby attenuating nociceptor excitability and contributing to analgesia. Third, in the extracellular compartment, PBM may activate latent transforming growth factor-β1 through a redox-dependent activation mechanism, which promotes tissue repair and regeneration. Through these molecular mechanisms acting at multiple cellular compartments, PBM exerts analgesic, anti-inflammatory, and reparative effects, which may all contribute to the symptomatic improvement observed in patients with KOA [30–33].

In a KOA-induced rabbit model, LLL inhibited cartilage degeneration and promoted chondrocyte regeneration. Additionally, radiographic and three-dimensional computed tomography analyses revealed the restoration of bone density, articular surface, and joint surface and contour. These effects increased with prolongation of the treatment period [34]. Therefore, these changes are thought to be the reasons for pain and functional improvement in the ILA.

Effective delivery of energy to deeper structures—such as the synovium, articular cartilage, and muscle, which are pathologic sites in KOA—is important, as more than 90% of the energy from lasers used in musculoskeletal applications is absorbed within the first 10 mm of tissue [35]. Therefore, we adopted three key approaches. First, direct contact, which involves tightly pressing the light source onto the skin, is important for preventing dispersion and reflection [36]. ILA enables effective penetration beyond direct contact by allowing positioning of the acupuncture needle tip underneath the skin. Second, the power density, defined as the laser energy delivered per unit area (W/cm²), affects the depth of energy penetration [37]. High-level lasers are known to cause tissue damage, making it important to improve penetration without increasing the laser level [38]. By maintaining the same power output, the light source was positioned at the acupuncture needle tip to narrow the irradiated area. These methods effectively increased power density. Thus, our power density of 63.69 W/cm², which was higher than that employed in previous LA or LLL therapy trials [39], may have contributed to the observed significant results. Third, light wavelengths ranging from 650 to 900 nm were found to be optimal for penetrating the skin. Therefore, we adopted the wavelengths of 650 and 830 nm based on previous studies [14]. The depth of penetration depends not only on parameters, such as wavelength and power density, but also treatment technique, such as direct contact. These combined strategies may have contributed to the intergroup differences observed in multiple outcomes.

For the VAS outcomes, nominal between-group differences were not observed at Visit 7 but were observed at Visit 13 and were sustained at Visit 14. Although this time course might suggest a gradual response after repeated ILA sessions, the exploratory design and small sample size preclude any firm interpretation of this temporal pattern. The observation should therefore be regarded as a preliminary observation rather than evidence of a specific temporal effect.

Although no significant intergroup differences in the WOMAC function subscale and EQ-5D-5L scores were observed at Visits 7 and 13, substantial differences were observed at Visit 14. These findings may have been influenced by the common treatment in all groups, which included teaching of self-management and self-exercise—as strongly recommended in the ACR guidelines for KOA—during the intervention period [3].

The EQ-5D-5L findings in this trial should be interpreted cautiously. In this K-L grade 2–3 KOA population, baseline EQ-5D-5L scores were relatively high across the three groups (control, 0.74; 650 nm, 0.75; 830 nm, 0.73), leaving limited room for further improvement and raising the possibility of a ceiling effect. A previous study of patients with hip or knee OA established an MCID for improvement of 0.32 in surgically treated patients whose health improved, but among non-surgically managed patients the effect sizes were small and no clear minimal clinically important difference (MCID) exceeding the individual-level minimal detectable change could be established [40]. These findings suggest that EQ-5D-5L MCID estimates vary according to the patient population and treatment context. Accordingly, an established EQ-5D-5L MCID may not be directly applicable to a non-surgical, high-baseline population such as ours, which further limits the interpretation of the modest changes observed. Incorporating a complementary measure such as the EQ-VAS in future studies may help better characterize subtle patient-perceived changes in quality of life.

The responder rates were 26.67% in the control group, 53.33% in the 650 group, and 60% in the 830 group. Although the experimental groups showed numerically higher rates compared to the control group, the difference did not reach statistical significance. This lack of significance may be attributed to the small sample size. Minimal clinically important improvement (MCII), defined as the smallest change in measurement that signifies an important improvement in a patient’s symptom, for KOA has been established as −19.9 mm for pain VAS and −9.1 for the WOMAC function subscale [41]. In our study, the 650 group showed reductions of −20.7 mm in VAS at rest, −21.7 mm in VAS during activity, and −14.9 in the WOMAC function subscale. Similarly, the 830 group showed reductions of −22.6 mm in VAS at rest, −22.9 mm in VAS during activity, and −14.9 in the WOMAC function subscale. These improvements observed in both experimental groups exceeded the established MCII thresholds. Despite the lack of a statistically significant difference in responder rates, exceeding MCII thresholds suggests potential clinical relevance. In addition to MCII, MCID values for KOA have been reported as VAS 13.7 mm, WOMAC total 6.4, WOMAC pain 1.5, and WOMAC function 4.6 [42]. At the primary endpoint (Visit 13), the between group differences (vs. control) exceeded these MCID thresholds for most outcomes: VAS during activity (15.0 mm, 650 group; 16.2 mm, 830 group), WOMAC total (13.1, 650 group; 11.8, 830 group), WOMAC pain (3.4, 650 group; 2.4, 830 group), and WOMAC function (8.4 for both groups). For VAS at rest, the difference in the 830 group (15.4 mm) exceeded the MCID threshold, while the difference in the 650 group (13.5 mm) closely approached but marginally fell below the threshold. Considering the clinically meaningful differences across most outcomes, these findings also suggest potential clinical relevance. Nevertheless, further well-designed studies with larger sample sizes are necessary to substantiate these findings.

Despite these strengths, our study has several limitations. First, we conducted a single-center pilot study owing to the limitations of trial duration, research funding, and opportunities for recruitment. Therefore, the sample size and diversity were insufficient to fully validate the efficacy and safety of ILA in the treatment of KOA. These factors may have introduced bias into the trial results. In the future, multi-center trials with larger sample sizes should be conducted to encompass a more diverse patient population and enhance the external validity and clinical applicability of the results. Second, the follow-up duration in our study was relatively short. Consequently, this study was limited in assessing sustained efficacy, long-term safety, and potential AEs of ILA. To determine the long-term efficacy and safety of ILA, future studies should incorporate extended follow-up periods such as three to six months or longer. Third, owing to the characteristics of ILA, it was not possible to blind the practitioners, as they had to directly operate the laser device. To minimize this risk, we adopted a patient- and outcome assessor-blinded design. Nevertheless, because practitioners were aware of group allocation, their awareness and any associated intentions or expectations while administering the control treatment may have introduced performance bias across the treatment and control groups. The absence of practitioner blinding therefore remains a methodological limitation and may have influenced the trial results. In addition, although we implemented multiple measures to maintain participant and assessor blinding, the effectiveness of blinding was not evaluated using a validated blinding index. Therefore, we could not verify whether the blinding procedures were fully successful. Further research should explore strategies for blinding practitioners when possible, consider the inclusion of independent outcome assessments or more objective outcome measures, and assess blinding effectiveness to further mitigate potential bias. Fourth, various parameters are relevant to LLL (e.g., irradiation time, wavelength, intensity, frequency, wave type, peak power, and treatment interval). However, only the wavelength was changed and modifications to other parameters were limited in this trial. Thus, subsequent studies should involve varied parameter changes to determine the optimal parameters. Fifth, LLL therapy application is influenced by a complex combination of parameters, delivery techniques, and irradiation sites [38]. As this exploratory pilot trial was primarily designed to investigate the wavelength-dependent effects of ILA, we prioritized comparing the 650 and 830 nm groups with a sham laser control. Consequently, our study did not incorporate a non-acupoint sham control, limiting our ability to determine whether the observed improvements were attributable to specific acupoints or to non-specific deep-tissue stimulation. Future studies incorporating sham acupoint designs are warranted to clarify the role of site specificity. Sixth, missing data were managed using the LOCF method, which is considered suboptimal compared to more robust approaches, such as multiple imputation or mixed-effects modeling [43]. Future studies using more advanced statistical methods are needed to better establish the efficacy of ILA on KOA. Seventh, this exploratory pilot trial measured multiple outcomes at several time points, and the individual visit-wise p-values in the main ANCOVA analyses were not adjusted for multiple temporal comparisons, raising the possibility of an inflated type I error. To address this limitation, we additionally performed a supplementary repeated measures ANCOVA to evaluate the group-by-visit interaction across visits. The results supported longitudinal group differences for the primary VAS outcomes, whereas most secondary outcomes were less robust. As this analysis was not prespecified in the protocol and the study had a limited sample size, these supplementary findings should be regarded as exploratory. More broadly, the statistically significant findings of the main analyses should therefore be interpreted as preliminary and hypothesis-generating rather than confirmatory, and future adequately powered confirmatory trials with prespecified longitudinal analyses and multiplicity-control strategies are warranted. Eighth, we did not restrict enrollment to patients with unilateral KOA to reflect real-world clinical practice, while treatment was confined to a single predefined target knee and VAS/WOMAC assessments were conducted for the predefined target knee. However, symptoms in the contralateral knee were not monitored over time. Previous evidence indicates that patient-reported pain and physical function in KOA may be influenced by contralateral knee pain and by unilateral versus bilateral symptom status [44,45]. We therefore could not determine whether the relative severity between the two knees shifted during follow-up, and an influence of the untreated contralateral knee on the patient-reported outcomes cannot be excluded. Prior work also suggests that bilateral involvement may matter clinically: between-limb gait asymmetry has been reported to be more common in bilateral than in unilateral mild-to-moderate KOA [46], and coexisting contralateral KOA has been linked to faster structural progression and a higher subsequent arthroplasty rate in the index knee [47]. Because the symptomatic unilateral/bilateral KOA status was not recorded as a study variable, and given the small per-group sample of this exploratory pilot trial, a laterality-based subgroup or sensitivity analysis was not feasible. Future confirmatory trials should prospectively document unilateral versus bilateral involvement, collect outcomes for both knees, and consider stratified or sensitivity analyses based on laterality.

Conclusions

In this exploratory pilot trial, ILA at 650 and 830 nm showed preliminary signals of short-term pain reduction among patients with KOA. However, because this was a pilot study with a small sample size and exploratory design, adequately powered, confirmatory trials are warranted to validate these preliminary observations.

Supporting information

Acknowledgments

The authors would like to thank the members of Clinical Research Center at Dongshin University Gwangju Korean Medicine Hospital for their valuable support.