The PIKASO trial (Preventing Injured Knees from Osteoarthritis: Severity Outcomes): Rationale and design features for a randomized controlled trial

Abstract

Objective

Given the high burden and increasing prevalence of post-traumatic osteoarthritis (PTOA), identifying clinically beneficial strategies to prevent or delay its onset could improve the quality of life of those at high risk of developing the disease.

Methods

Preventing Injured Knees from OsteoArthritis: Severity Outcomes (PIKASO) is a multicenter blinded, parallel, two-arm randomized controlled trial of 512 individuals aged 18–45 years undergoing anterior cruciate ligament reconstruction (ACLR). This study is designed to evaluate the efficacy of a 12-month intervention of oral metformin vs. placebo in decreasing the rate of structural knee changes and pain. Participants will be asked to take up to 1500 ​mg/day of either metformin or placebo as tolerated. The primary outcomes are Knee injury and Osteoarthritis Outcome Score (KOOS) Pain subscale scores averaged between 12 and 24 months after randomization, and MRI derived cartilage structural change at 24 months. The trial will be deemed successful if at least one of the two primary outcomes reaches the preplanned effect size with sufficient statistical certainty. In this paper, we describe PIKASO elements according to the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) guidelines.

Results

Our multidisciplinary team developed the methods and statistical analysis plan for a placebo controlled, randomized clinical trial to determine whether metformin alters pain and early osteoarthritic changes after ACLR.

Discussion

This manuscript outlines the rationale, study design, and implementation of the PIKASO trial aiming to prevent the onset of PTOA after ACLR.

Trial registration

clinicaltrials.gov NCT06096259.

1. Introduction

The outcomes of trials testing treatments for established knee osteoarthritis (OA) have been frustratingly negative, suggesting that by the time patients present clinically with both symptoms and structural damage, OA may be too advanced for pharmacologic treatments to be effective [1]. This problem has led to efforts to identify individuals with early OA and develop treatments to prevent further progression [2]. Experiments testing new treatments in animal models of OA have been promising, often suggesting therapies that might be tested in humans. One reason for the discordance between human and animal studies is the disease course. Animal models frequently evaluate treatments after the deliberate induction of joint trauma, whereas OA can develop in humans following either years of ‘wear and tear’ or a traumatic injury. The human condition closest to a replication of animal models of joint trauma is post-traumatic osteoarthritis (PTOA).

Metformin, a drug commonly used to treat type II diabetes, has emerged as a potential pharmaceutical treatment that may prevent PTOA. Metformin may protect against early PTOA changes by 1) maintaining chondrocyte metabolism and homeostasis through upregulation of the AMP-activated protein kinase (AMPK) pathway [[3], [4], [5]], 2) decreasing inflammation and neuroinflammation by downregulation of the nuclear factor κB (NF-κB) pathway and the NOD-like receptor protein 3 (NLRP3) inflammasome [[6], [7], [8]], and 3) promoting chondrocyte autophagy and mitophagy, and decreasing cellular senescence [[9], [10], [11]]. Metformin has consistently shown efficacy in animal models of trauma-induced OA in rodents and non-human primates [3]. In animal models of PTOA, metformin use was associated with reduced synovitis, pain behaviors, and cartilage degradation [[3], [4], [5],7,[12], [13], [14]]. Clinically, individuals with symptomatic and radiographic knee OA treated with metformin alone or in combination with meloxicam demonstrated significantly reduced pain and systemic concentrations of pro-inflammatory cytokines compared to placebo [[15], [16], [17]], and the risk of developing OA was lower in patients with diabetes who received metformin vs. patients treated with sulfonylurea [18]. A cohort study of participants in the Osteoarthritis Initiative additionally showed that among diabetic patients, metformin use was associated with a significant reduction in the risk of requiring total knee replacement [19]. These prior studies further highlight the potential of metformin as a preventive therapy.

It is estimated that nearly 44 million people worldwide are living with PTOA of the knee [20,21]. As rates of traumatic knee injuries such as anterior cruciate ligament (ACL) injuries continue to rise, the prevalence of PTOA is also expected to increase [22,23]. While both operative and nonoperative treatments have been used to treat ACL injuries, currently more than 1 in 3 people who undergo ACL reconstruction (ACLR) develop irreversible PTOA within 10 years of surgery [24,25]. Development of PTOA appears to begin within the first year after ACLR. Structural changes likely begin shortly after surgery with sustained, persistent increases in pro-inflammatory cytokines and degradative enzymes [26,27], altered joint biomechanics [[28], [29], [30]], and changes in cartilage composition [31,32] that precede frank radiographic changes such as joint space narrowing and the development of osteophytes [25,33]. Similarly, pain and symptoms consistent with early OA have been observed during the first year after ACLR [34,35], with patient-reported outcome measures largely remaining unchanged between 1, 2, 6, and 10 years after surgery [[36], [37], [38]]. These data point toward the need for an intervention that can alter PTOA risk in the first year post-ACLR. We designed the Preventing Injured Knees from OsteoArthritis: Severity Outcomes (PIKASO) trial to evaluate the efficacy of treatment with metformin in this first postoperative year to reduce the incidence of PTOA. We hypothesize that after ACL tear and subsequent reconstruction, a 12-month course of metformin will delay the onset of PTOA, including slowing progression of degenerative knee structure changes and/or reducing knee pain one to two years after ACLR.

2. Methods

2.1. Rationale for trial design features

Trial design features were decided based on the OPTIKNEE consensus recommendations, preliminary data from patients aged 18–35 years who were scheduled for or had recently undergone primary ACLR (unpublished data; [39]) and extensive discussions among study investigators, including an endocrinologist, rheumatologists, orthopedic surgeons, and physical therapists [39,40]. Additionally, for a drug to be approved as a disease modifying osteoarthritis drug (DMOAD), the FDA requires that candidate drugs improve pain or function while also slowing structural changes [41,42]. Recommendations on design considerations for PTOA clinical trials suggest that both symptoms and structural changes are important to demonstrate therapeutic efficacy [43]. This consideration necessitated a decision to adopt co-primary endpoints or multiple primary endpoints. With a co-primary endpoint design, a drug will not be considered effective without demonstrating significant improvement in both pain AND structure, in this case, cartilage loss [44]. When employing a multiple primary endpoint design, a drug would be considered effective if improvement was observed in either of the primary endpoints (i.e., improvement in pain OR reduction of cartilage loss). Prior work suggests that changes in cartilage morphometry may not translate immediately to changes in symptoms [45]; given the relatively short-term (2-year) duration of the PIKASO trial, we opted for the multiple primary endpoint design (Table 1). We chose 24 months for MRI joint structure assessment and the average of KOOS pain scores at both 12 and 24 months as timepoints to evaluate the primary endpoints. The use of pain scores at 12 and 24 months was chosen to distinguish our target of PTOA-related pain from post-surgical pain (thought to resolve by 6–9 months post-surgery).

Table 1.

Key considerations in decision-making regarding PIKASO primary outcome(s).

Outcome Type	Description	Pros	Cons
Single	Symptoms or structure as single primary outcome	No sample size adjustments needed to maintain type I and II error	Guidance on PTOA clinical trial designs advise to include measures of both symptoms and structure, similar to guidance on DMOAD clinical trial design
Composite	Combination of symptoms and structure into one outcome (e.g., as in the FNIH OA biomarkers consortium which defined progression as a concurrent increase in pain and decrease in joint space width)	No sample size adjustments needed to maintain type I and II error	Composite outcomes can be difficult to interpret; it is unclear whether treatment affects symptoms, structure, or both
			The correlation between structure and symptoms may be low, and the two may not improve simultaneously with treatment, thus the number meeting a composite outcome definition may be low and the total required sample size may be very large
Co-primary	Treatment must show effects on both symptoms and structure	Follows guidance on PTOA and DMOAD clinical trial designs to include measures of both symptoms and structure	Risk of type II error is increased; sample size must be increased to ensure overall power is maintained across both endpoints.
			Changes in structure may not translate immediately to changes in symptoms
Multiple primary	Treatment must show effects on either symptoms or structure	Allows for independent analysis of symptoms and structure	Risk of type I error is increased; α is typically decreased for each outcome and thus sample size must be increased to ensure overall α is maintained
			A treatment that affects structure or symptoms but not both might not be considered disease-modifying

2.2. Trial organization

The PIKASO trial is led by two co-Principal Investigators, an orthopedic sports medicine surgeon (MHJ) and a clinical orthopaedic researcher (CAJ). They are advised by the Arthritis Foundation's Osteoarthritis Clinical Trials Network (OA-CTN) Steering Committee, which includes clinician-scientists, biostatisticians, and laboratory scientists, as well as several core coordinating centers (led by DTF). The Data Coordinating Center (DCC) at Mass General Brigham (led by EL and JC) oversees day-to-day operations of the trial, including monitoring screening and recruitment, randomization, organizing staff training, facilitating adherence monitoring, coordinating central resources, and providing reports to the independent Data and Safety Monitoring Board (DSMB). The Imaging Coordinating Center at Cleveland Clinic (led by CSW and XL) manages and coordinates the imaging at each site to ensure uniformity of MRI protocols and image quality and perform centralized image reading and processing. The Biomechanics Coordinating Center at the University of North Carolina at Chapel Hill (led by BP and CL) provides guidance on postoperative physical therapy recommendations and oversees collection, data transmission, and analysis of the biomechanical study outcomes. The Biospecimen Coordinating Center at Duke University School of Medicine (led by VBK and MO) provides oversight for proper collection and analysis of all biological specimens obtained through the study. Finally, the study is supported by an endocrinologist who has performed metformin-related clinical trials in both diabetic and non-diabetic populations, and through active participation from the study sponsor (the Arthritis Foundation). During project preparation, several study groups, including a study-wide operations committee and a clinical committee of the enrolling study surgeons, met regularly to finalize study details.

The intervention will be delivered at nine academic medical centers in the United States: Cleveland Clinic (Cleveland, OH and Weston, FL), Emory University (Atlanta, GA), Hospital for Special Surgery (New York, NY), Mass General Brigham (Boston, MA), The Ohio State University (Columbus, OH), University of Iowa (Iowa City, IA), University of Kentucky (Lexington, KY), University of Nebraska Medical Center (Omaha, NE), and University of North Carolina at Chapel Hill (Chapel Hill, NC).

2.3. Trial design

PIKASO is a multicenter quadruple blinded (participants, care providers, investigators, and outcome assessors), parallel, two-arm randomized controlled trial. It is designed to evaluate the efficacy of a 12-month metformin intervention at slowing the rate of structural changes in the knee joint as evidenced by cartilage morphometry on MRI or reducing pain levels measured by KOOS Pain score compared to placebo. The multiple primary outcomes (structure/pain) dictate that the trial be considered ‘positive’ if at least one of the outcomes reaches the preplanned effect size with sufficient statistical certainty. This study is registered at Clinicaltrials.gov (NCT06096259) and was approved by the Advarra Institutional Review Board (IRB) for all sites under a single IRB. In this paper, we describe the components of PIKASO according to the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) guidelines [46].

2.4. Study sample: inclusion and exclusion criteria

The PIKASO trial is enrolling individuals aged 18–45 years who have experienced an ACL tear within the past 6 months and are planning to undergo ACLR. Table 2 provides the full inclusion and exclusion criteria for the study, and patients will not be excluded based on the presence, severity, or treatment of meniscus injury. These criteria are designed to include patients at the highest risk of PTOA development, without a past medical history of traumatic ACL injury, other concomitant injuries, or pre-existing OA that could confound trial results. Additionally, the inclusion/exclusion criteria aim to preserve participant safety by excluding individuals with contraindications to metformin as described in the package insert (renal impairment and/or history of diabetic ketoacidosis).

Table 2.

PIKASO study inclusion and exclusion criteria.

Inclusion Criteria	Exclusion Criteria
Age 25–45 years or age 18–24 yearsa with preoperative KOOS pain ≤80 (0–100, 100 best) recorded at least 14 days after the day of ACL injury	Known contraindications to metformin, including current use of metformin or topiramate, type I or II diabetes mellitus or diabetic ketoacidosis, and acute or chronic renal insufficiency
Date of injury within 6 months of screening	Pregnancy and/or lactation, or plans to become pregnant in the next 12 months
Plan to undergo ACL reconstruction within 12 months of injuryc	History of prior ACL tear on the index knee, with or without reconstruction, greater than 6 months prior to screening
Surgery scheduled for at least 14 days from the time of screening to allow time for study drug initiation and dose titration	History of ACL tear or reconstruction on the contralateral knee within the past 12 months
Able to undergo MRI	Presence of a condition or abnormality that in the opinion of the surgeon would compromise the safety of the patient or the quality of the data
English speaking	Plan for allograft at time of surgical consent
Willing and able to provide informed consent	Concomitant PCL/MCL/LCLb injury or avulsion fracture on index knee requiring surgical treatment
	Tibial plateau fracture on index knee visible on radiographs and requiring treatment as determined by treating surgeon
	Applying for or receiving workers' compensation for their knee injury
	Joint space narrowing AND definite osteophyte(s) on weight-bearing radiograph on index knee
	Inflammatory arthritis

^aAge 19–45 at the University of Nebraska Medical Center site, where the state-recognized age of majority is 19 years.

^bPCL ​= ​posterior cruciate ligament, MCL ​= ​medial collateral ligament, LCL ​= ​lateral collateral ligament.

^cParticipants are eligible if they are planning to undergo ACLR within 6 months after screening. Should a potential participant not schedule surgery within 6 months of consent or cancel surgery prior to randomization, they will not be included in the study. However, if a participant cancels surgery after completing baseline testing, randomization, and began the study treatment, then they will be included in the trial.

Participants are recruited from outpatient orthopedic and sports medicine clinics by participating surgeons at the enrolling medical centers. At the clinic visit, the recruiting physician introduces the study to potentially eligible and interested patients, who are then screened to confirm eligibility by study research coordinators. The majority of the inclusion/exclusion criteria are assessed by the research coordinator at this visit except for the surgical and radiographic exclusions (joint space narrowing and definite osteophytes on clinical radiographs; tibial plateau fracture requiring treatment; plan for allograft at time of surgical consent; concomitant posterior, lateral, or medial collateral ligament injury or avulsion fracture to be treated surgically; and presence of any other condition or abnormality that, in the opinion of the treating surgeon, would indicate that the patient would not be a good candidate for the study). These surgical and radiographic criteria are assessed by the recruiting physician, and clinical MRI confirmation of the ACL tear is obtained before patients can be scheduled for a baseline study visit, which must occur at least eight days and no more than six weeks before the patient's scheduled ACLR. In line with our risk-enhanced sample, patients aged 18–24 years undergo an additional screening step involving administration of the KOOS Pain scale at least 2 weeks post-injury to assess baseline pain levels. Patients aged 18–24 years with a KOOS Pain score >80 ​at least two weeks post-injury are excluded from the study. All inclusion and exclusion criteria are fully assessed before a patient provides informed consent for the study.

2.5. Randomization and blinding

After successful completion of the baseline visit, participants are randomized 1:1 to either metformin or placebo. Randomization, stratified by site, is performed locally by blinded study staff employing REDCap data collection software and a site-specific randomization schema that uses permuted blocks with randomly varying sizes of two and four. The PIKASO statistician (JC) developed the randomization schema program that was run by an unblinded DCC staff member to ensure that the statistician remains blinded to treatment assignment. Each site's investigational pharmacy receives the randomization scheme prior to recruitment to know which study arm each REDCap-generated random number corresponds to. Only the DCC data manager, trial coordinator, and the investigational pharmacies have access to the schema, maintaining quadruple blinding. The study uses a placebo tablet virtually indistinguishable in appearance and texture from metformin manufactured at one of two production laboratories (Sharp Clinical Services LLC, Allentown, PA, USA; UI Pharmaceuticals, Iowa City, IA, USA). The study drug bottle label includes a number that allows the dispensing pharmacist to determine the type of medication (metformin or placebo) held within a specific bottle.

2.6. Intervention

Metformin is a medication commonly used in the U.S. Based on existing experience with the drug to date, the study does not pose serious risks beyond those related to administration of a standard dose of metformin [47]. Lactic acidosis is a rare side effect observed in <1 case per 100,000 treated patients [47,48]. The exclusion criteria for PIKASO are intended to exclude those with known risk factors for lactic acidosis [[48], [49], [50]].

Participants assigned to metformin will use extended-release metformin 500 ​mg tablets (supplied by India Limited, Hyderabad, India; packaged and labeled by the manufacturers of record, Sharp Clinical Services LLC, Allentown, PA, USA or UI Pharmaceuticals, Iowa City, IA, USA). Consistent with the dosing used in both animal models of PTOA and OA clinical trials, participants are asked to take 1500 ​mg/day as tolerated [[3], [4], [5],7,[12], [13], [14], [15], [16], [17]]. During the dose titration period, study staff call participants weekly to record the current dosing level and assess for side effects. Participants can remain on a lower dose (500 ​mg or 1000 ​mg daily) for the duration of the treatment period if they cannot tolerate the full study dose.

Both the metformin and placebo groups follow standard titration procedures for extended-release metformin [[51], [52], [53], [54]]. Participants begin by taking one 500 ​mg tablet each day for 5–7 days, increase to two 500 ​mg tablets once per day for 5–7 days (1000 ​mg/day), and then increase to three 500 ​mg tablets once per day as tolerated (1500 ​mg/day). Written instructions are provided to participants, and research personnel check in with participants by phone, email, or SMS text message at designated times during the titration process. Medication adherence will be quantified throughout the medication phase of the trial using a Bluetooth-enabled pill dispenser (Wisepill RT3000 pocket dispenser, Wisepill Technologies, Cape Town, South Africa) to track medication adherence.

Participants undergo standard of care ACLR within six weeks of initiating the study drug and are instructed to hold their study drug on the day of surgery to align with standard pre-operative guidance for remaining NPO (taking nothing by mouth) prior to scheduled surgery. Participants resume their study drug on postoperative day 1. All study participants receive identical post-operative care instructions and are encouraged to complete a full course of physical therapy (PT). Participants in both arms have the option to participate in PT at the facility of their choosing, but are offered criterion-based post-operative PT guidelines from the University of Delaware and are eligible to receive financial compensation for documented PT attendance [55] (Supplementary File). Table 3 depicts the study intervention alongside other study components and assessments.

Table 3.

Schedule of study procedures and assessments.

2.7. Study outcomes

2.7.1. Primary outcomes

The PIKASO study has two primary outcomes: KOOS Pain score (assessed at 12- and 24-months post-randomization and averaged across these timepoints) and the change in modified MRI Osteoarthritis Knee Score (MOAKS) cartilage score at 24 months post-randomization [56]. If KOOS pain is available at either the 12- or 24-month timepoints, but not both, then the available KOOS pain data will be used.

The MRI scans will be obtained at baseline, 12 months, and 24 months post-randomization using 3-T MRI systems (Siemens Healthineers, GE HealthCare). The modified MOAKS assesses cartilage damage in three cartilage thickness domains [surface, high grade cartilage loss (loss that extends to the middle of the cartilage and its extent), full thickness cartilage loss] within each of 16 knee subregions [56]. Structural damage will be quantified as any increase in MOAKS score (including within-grade) in each of 16 subregions at 24 months compared to the baseline. An example of a within-grade change would be high grade cartilage loss that extended to the middle of the cartilage at baseline but progressed to extend further into the cartilage without being full thickness loss. The domain-specific scores may be summed within each subregion to create an ordinal cartilage damage score ranging from 0 for no cartilage loss to 9 for full thickness cartilage loss involving >75 ​% of the subregions surface area. Such score will be used as a secondary outcome.

2.7.2. Additional outcomes

The study was designed by a multidisciplinary group with expertise in areas of OA research involving imaging, biomechanics and functional outcomes (i.e., isokinetic lower extremity muscle strength, 3-dimensional motion capture walking gait analysis, habitual walking speed and hop testing), and biospecimen analyses, and in alignment with the OPTIKNEE 2022 consensus [40], the variety of secondary and exploratory study outcomes is reflective of this breadth of knowledge (Table 4). During the consent process, participants either opt in or out of the following components: contrast MRI (non-contrast is required but contrast is offered at select sites), blood draw (genetic and non-genetic sample types), urine collection, the collection of synovial fluid and tissue during surgery.

Table 4.

PIKASO study measures.

Outcome Type	Outcome Measure	Method of Assessment	Timepoint(s)
Medication adherence	Number of days study medication was taken	Wisepill RT3000 pill dispenser data	3, 6, 9, 12 months
Patient reported outcome measures (PROMs)	KOOS (all subscales)	Questionnaire	BL, 3, 6, 9, 12, 18, 24 months
	MARX activity rating scale	Questionnaire	BL, 12, 24 months
	Work productivity and activity impairment (WPAI)	Questionnaire	BL, 3, 6, 9, 12, 18, 24 months
	EQ-5D-5L	Questionnaire	BL, 3, 6, 9, 12, 18, 24 months
	Tampa scale of kinesiophobia	Questionnaire	BL, 3, 6, 9, 12, 18, 24 months
	MHI-5	Questionnaire	BL, 3, 6, 9, 12, 18, 24 months
	Healthcare utilization/Comorbidities	Questionnaire	BL, 3, 6, 9, 12, 18, 24 months
	Pain catastrophizing scale	Questionnaire	BL
	Demographics	Questionnaire	BL
Imaging	Knee bone and joint space changes	X-ray	BL, 24 months
	Modified MOAKS cartilage, knee structure changes	MRI	BL, 12, 24 months
	Changes in cartilage T1rho, T2	Compositional MRI	BL, 12, 24 months
	Synovitis	Contrast MRI	BL, 12 months
	Synovitis	FS FLAIR MRI	BL, 12, 24 months
Knee function and biomechanics	Knee ROM	Physical exam	12, 24 months
	3D gait biomechanics	3D gait biomechanics	12, 24 months
	Lower extremity strength	Isokinetic knee flexion/extension strength at 60°/sec	12, 24 months
	Physical function	10 ​m/1 ​m habitual walking speed, single leg hop	12, 24 months
Biospecimens	Urine sample	Urine collection	BL, 12, 24 months
	Serum and plasma samples	Blood collection	BL, 12, 24 months
	PAXgene® blood RNA	Blood collection	BL, 12, 24 months
	Buffy coat (DNA)	Blood collection	BL, 12, 24 months
	Synovial fluid	Synovial fluid aspiration	Surgery
	Synovial tissue	Synovial biopsy	Surgery

2.8. Statistical analyses

The outcomes will be analyzed separately. KOOS pain will be analyzed with linear regression, adjusted for age, sex, education (bachelor's degree or higher vs. less than bachelor's degree), baseline KOOS pain, and study site. In a secondary analysis, a repeated measures approach using all longitudinal observations will be conducted using linear mixed-effects models, including the fixed effects of study arm, study site, baseline value of the outcome, timepoint, arm by timepoint interaction, age, sex, education, and a random participant effect. Associations will be reported as adjusted mean differences and 95 ​% confidence intervals. Data may be transformed (e.g., log) if found to be non-Gaussian.

Structural damage will be measured with the MOAKS cartilage score. In each of the 16 subregions, we will categorize progression as either “progression” or “no progression” within that specific subregion. All 16 subregions will be included in a mixed-effects log-binomial regression model to estimate the relative risk of subregional progression in the metformin vs. placebo group. The model will include study arm, study site, age, sex, education, and a random participant effect (to account for 16 subregions per participant).

Continuous secondary outcomes will be analyzed in a similar method as the primary outcomes. Binary secondary outcomes will be analyzed with logistic regression adjusted for study arm, study site, baseline value of the outcome, timepoint, age, sex, education, and arm by timepoint interaction.

The intention-to-treat (ITT) analysis implies that each participant will be analyzed according to the randomization assignment, regardless of treatment received or adherence. The full ITT set will include all participants randomized. The modified ITT (mITT) set will include all participants randomized with a complete dataset (MRI obtained at 24 months and KOOS pain collected at least once at 12 or 24-months post-randomization). The per-protocol (PP) set will include all participants randomized with >80 ​% adherence to the assigned intervention, assessed via Wisepill adherence data. The primary analysis will be on the mITT set, with secondary analyses on the full ITT set and PP set.

2.8.1. Addressing missing data

For analyses of the full ITT set, multiple imputation will be used for missing data. Data will be imputed under the missing at random assumption, whereby missing data are imputed based on observed covariates and outcomes. Continuous outcomes will also be analyzed with linear mixed-effects models, which do not require that participants have complete data [57].

2.9. Sample size and statistical power

Sample size is based on the two primary outcomes of KOOS pain averaged between 12 and 24 months and MOAKS cartilage damage at 24 months. Alpha is set at 0.025 (two-sided) to account for two primary outcomes. The reported Minimal Detectable Change (MDC) for the KOOS pain subscale is 6.1 points in athletes after ACLR [58]. A difference of 6.1 points with a standard deviation of 17.5 is an effect size of 0.35 [[59], [60], [61], [62], [63]]. To detect an effect size of 0.35 with α ​= ​0.025 and power ​= ​90 ​% using baseline adjusted analysis of covariance assuming a correlation between baseline and follow-up of 0.25, a sample size of 192 per group (384 total) is required [64]. Based on MRI data from the MOON study, we expect approximately 2.5 ​% of subregions to demonstrate progression in the placebo arm [65]. A sample size of 192 per group will allow us to detect a relative risk of approximately 1.9 (1.3 ​% progression in the metformin group vs. 2.5 ​% in the placebo group) with 90 ​% power and alpha ​= ​0.025. We feel this is sufficient to be clinically meaningful but also acknowledge no established MCID for progression of structural cartilage damage currently exists. We also accounted for an anticipated dropout rate of 25 ​%, resulting in a final sample size of 256 per group (512 total).

2.10. Adverse Events and participant safety

Adverse events (AEs) and serious adverse events (SAEs) including emergency room visits for any reason, hospital admissions (must be an overnight stay or longer), reoperation on the index knee after primary ACLR for any reason, ACL surgery on the contralateral knee, or death are recorded from the time of randomization through a participant's completion of the study at 24 months post-randomization. AEs will be identified from the study participant directly via telephone calls and questionnaires (3, 6, 9, 12, 18, and 24 months) and through the electronic medical record (EMR) of each site (EMR alerts and a limited medical record review). For the PIKASO study, an adverse event is any untoward medical occurrence associated with the use of an intervention in humans, whether or not considered intervention-related, that warrants medical intervention [66]. AEs are classified by expectedness (unexpected or expected), relatedness to intervention (definitely, probably, possibly, or not related), severity (Grade 1–5, assessed using the CTCAE v5 AE reporting system [67]) and seriousness (non-serious and serious).

Lactic acidosis is the most severe of potential metformin-associated side effects, resulting in hypothermia, hypotension, bradyarrhythmia, or even death [[48], [49], [50]]. Participants are asked to contact study staff and seek immediate medical attention if they develop severe symptoms of lactic acidosis (including muscle pain, respiratory distress, and abdominal pain). The exclusion criteria for PIKASO (Table 2) are intended to account for many of the known risk factors for lactic acidosis [[48], [49], [50]].

Overall study monitoring is conducted by the DCC to ensure participant safety. AEs are monitored on an on-going basis and shared with the sIRB and Data Safety Monitoring Board (DSMB) in a timely manner consistent with the type of event or deviation. The independent external DSMB for the PIKASO trial consists of five members, including an endocrinologist, an orthopedic surgeon, a physical therapist, a biostatistician, and an imaging expert. The DSMB will meet twice annually.

3. Discussion

More than 1 in 3 people develop irreversible PTOA within 10 years after ACLR, and to date, no treatment options have successfully altered PTOA progression [24,25]. As such, the purpose of this placebo-controlled, multicenter randomized clinical trial is to determine if a 12-month course of daily oral metformin results in either reduced knee pain or structural changes on MRI one to two years after ACLR. This study represents a unique collaboration involving clinicians, clinician-scientists, and investigators with expertise in clinical trials, orthopedics and sports medicine, physical therapy, rheumatology and inflammation resolution, imaging, biomarkers, and biomechanics.

Many rehabilitation-related and pharmacological interventions were considered by the OA-CTN, including intra-articular injections and other oral medications. Metformin was selected because of its consistently positive results in animal models of PTOA, positive results in limited OA clinical trials [[3], [4], [5],7,[12], [13], [14], [15], [16], [17],68,69], well-established safety profile [47], low cost, and accessibility. Furthermore, by beginning treatment prior to surgery, metformin may help mitigate inflammation secondary to the surgical insult [26]. While inflammation lessens over time relative to the initial postoperative spike, the first postoperative year has multiple ‘triggers’ that may either prevent inflammation resolution or re-initiate the inflammatory response [26]. These ‘triggers’ include resumption of unassisted walking (first month after surgery), running and jumping (4–6 months), and cutting and sports-specific activity (6–9 months) before resuming full activity (9–12 months) [70]. The current standard of care does not alter this persistent inflammation, and elevated synovial fluid cytokine concentrations have been demonstrated to persist up to five years after ACLR [27]. We hypothesize that sustained delivery of metformin for 12 months may more successfully protect against inflammation throughout the first postoperative year because the long-duration therapy may improve the temporal regulation of cytokine activity better than shorter-duration treatments [71].

4. Conclusion

The PIKASO Trial is a quadruple-blind, placebo-controlled, multicenter randomized clinical trial to determine if a 12-month course of daily metformin results in either reduced knee pain or structural changes on MRI one to two years after ACLR. This novel approach has the potential to alter the development and irreversible progression of PTOA in what is an otherwise young and healthy patient population, and to contribute to a paradigm shift in postoperative care.

Contributions

All authors should have made substantial contributions to the conception and design of the study, drafting the article or revising it critically for important intellectual content, and have provided final approval of the submitted version of the manuscript. The study PIs, Dr. Cale Jacobs (cjacobs@bwh.harvard.edu) and Dr. Morgan Jones (mhjones@bwh.harvard.edu) take responsibility for the integrity of the work as a whole, from inception to finished article.

Role of funding source

Funding was provided by the Arthritis Foundation Osteoarthritis Clinical Trial Network. Arthritis Foundation representatives (Jason Kim, Michelle McLeod) contributed to study design and writing of the manuscript. The funding source will not be involved with data collection, analysis or interpretation, or the decision to submit the manuscript for publication.

Declaration of competing interest

No author has any financial or personal relationships that could potentially influence their work or conclusions. For a comprehensive list of an author's disclosures, please refer to their respective ICMJE form.

Handling Editor: Professor H Madry

Appendix A. Supplementary data

The following is the Supplementary data to this article: