Effects of initial graft tension and patient sex on knee osteoarthritis outcomes after ACL reconstruction: A randomized controlled clinical trial with 10-to-12-year follow-up

Abstract

Background:

The “initial graft tension” applied during ACL graft fixation may promote posttraumatic osteoarthritis (PTOA).

Hypothesis/Purpose:

This study sought to assess the impact of initial graft tension and patient sex on PTOA outcomes at 10 to 12 years post-anterior cruciate ligament reconstruction (ACLR). The hypotheses were that there would be no group- or sex-based differences in outcomes.

Study Design:

Randomized Controlled Clinical trial.

Methods:

Patients were randomized to ACLR with a low or high initial graft tension. Outcomes were evaluated at 10 to 12 years postoperatively and compared with a matched, uninjured control group. Outcomes included clinical (AP knee laxity measurement, International Knee Documentation Committee [IKDC] examination score), functional (1-legged hop for distance), patient-reported (Knee injury and Osteoarthritis Outcome Score [KOOS], Short Form-36 [SF-36], Tegner activity level, patient satisfaction), and PTOA imaging (Osteoarthritis Research Society International [OARSI] radiographic score, Whole-Organ Magnetic Resonance Imaging Score [WORMS]) assessments. Two-way mixed model analyses of variance were used to evaluate differences in outcomes between tension groups and the control group and between females and males.

Results:

Both tension groups scored worse than the control group for the IKDC examination (P≤.021), KOOS (Pain, ADL, Sport, and QOL subscales) (P≤.049), and WORM difference score (P≤.042) outcomes. The low-tension group scored worse than the control group for KOOS-symptoms (P=.016) and the OARSI difference score (P=.015). The index limb had worse scores than the contralateral limb within the high-tension group for AP laxity (P=.030) and hop deficit (P=.011). This result was also observed within both tension groups for the WORMS (P≤.050) and within the low-tension group for the OARSI score (P=.001). Males had higher mean±standard error Tegner [males=5.49 (±1.88) vs. females=4.45 (±1.65)] and worse OARSI difference scores [males=1.89 (±5.38) vs. females=.244 (±.668)] relative to females (P=.007 and .034, respectively). However, no significant differences were detected between tension groups for any of the outcomes measured.

Conclusion:

Overall, ACLR failed to prevent PTOA regardless of initial graft tension. However, males treated with a low initial graft tension may be at greater risk for PTOA. These results do not support the hypothesis of no sex differences in outcomes at 10 to 12 years post-ACLR.

Clinical Relevance:

The results suggest that males may be at risk for worse PTOA post-ACLR, which may be relevant to clinical management post-ACL injury.

INTRODUCTION

Anterior cruciate ligament (ACL) injuries affect approximately 400,000 individuals annually in the United States,³⁶ and place patients at risk for post-traumatic osteoarthritis (PTOA).³ ACL reconstruction (ACLR) is commonly performed to restore knee function and decrease the risk of PTOA after ACL injury.²¹ However, clinical studies demonstrate that PTOA often progresses despite surgical intervention.^14,19,21,33 The reason for this progression remains unclear, although the “initial graft tension” applied at the time of fixation has been implicated as an important contributing factor.¹⁹

Initial graft tension alters native joint contact mechanics and kinematics, which may predispose the knee to PTOA development post-ACLR.¹⁰ Recommendations for graft tensioning protocols have been made, but the optimal tension to restore native knee stability, while at the same time minimizing the risk of PTOA, remains unknown. In addition, female sex is an established risk factor for ACL injury and has also been suggested as a risk factor for worse outcomes post-ACLR, although this remains a topic of debate, with some studies reporting worse post-operative outcomes in female patients,^{1,2,9,26,29,31} and others reporting no sex-based differences.^{7,17,25,38,40,44}. Short- (i.e., 3 year)¹⁸ and mid-term (i.e., 7 year)⁴ analyses from an ongoing randomized controlled trial examining the outcomes of two different initial graft tension conditions have been previously published. Minimal differences in outcomes between the two tension cohorts at 3 and 7 years postoperatively were observed, as well as inferior clinical, functional, patient-reported, and imaging outcomes among the ACL reconstructed group relative to the matched, uninjured control subjects.^4,18 No analysis of patient sex effects was conducted within this cohort. As PTOA is a progressive disease and given the emerging differences between the tension groups at the mid-term time point, we sought to investigate long-term PTOA development and potential sex-based differences in outcomes among this patient cohort.

The primary aim of this longitudinal trial was to evaluate the progression of PTOA following surgical reconstruction of the ACL between low- and high-tension groups 10 to 12 years after surgery. The secondary aim was to determine how patient sex, initial graft tension, and the interaction between these variables may influence long-term clinical (anteroposterior (AP) knee laxity, International Knee Documentation Committee [IKDC] examination score), functional (i.e., 1-legged hop for distance), patient-reported (Knee injury and Osteoarthritis Outcome Score [KOOS], Short Form-36 [SF-36], Tegner activity level, patient satisfaction), and imaging (Osteoarthritis Research Society International [OARSI] radiographic score, Whole-Organ Magnetic Resonance Imaging Score [WORMS]) outcomes related to osteoarthritis. To our knowledge, the investigation of the potential interaction between initial graft tension and patient sex in the context of long-term outcomes is a novel pursuit. The present analysis leverages the ACLR and matched, uninjured control patient cohorts from the ongoing trial evaluating the effects of initial graft tension.^4,18 We hypothesized that there would be no differences in outcomes between the two initial graft tension groups or between female and male patients at 10 to 12 years post-ACLR.

METHODS

Trial Design

The current long-term analysis is an extension of previous short-term (i.e., 3-year)¹⁸ and mid-term (i.e., 7-year)⁴ analyses from an ongoing randomized controlled trial (NCT00434837), and leverages the follow-up data obtained at 10 and 12 years post-ACLR (Fig. 1). Due to the higher rate of loss to follow-up at the 12-year time point, 10-year data were substituted for missing 12-year data when available. The Institutional Review Board of Rhode Island Hospital approved this study and all participants provided written informed consent.

Figure 1.

CONSORT diagram of the study design, patient allocation, and loss to follow-up through 12 years. *Note that of the 108 patients that were originally recruited into the tension groups, 18 were excluded (1 cancelled surgery, 2 opted out of randomization, 2 positive pivot shift examinations in uninjured knee, 2 partial ACL tears not reconstructed, 1 allograft, 1 quadriceps tendon repair, 4 meniscal tears involving more than 1/3 of the meniscal body, and 5 chondral lesions).

The study was designed as a randomized double-blind controlled trial. The randomization assignment, which was performed by the study statistician (GJB), was disclosed to the operating surgeon at the time of intra-operative graft placement. All subsequent evaluations were performed by blinded assessors. Power calculations specific to analyses of long-term outcomes (i.e., 10-to-12-year) were performed a priori based on available 7-year data, and estimated that a sample size of 30 participants per surgical group would have sufficient power (1-β=0.80 using α=.05) to detect a mean difference between groups of 1.4 mm for AP knee laxity; 11.5 points (i.e. less than a mild deficit⁴¹) for KOOS-QOL; and 0.50 points for the IKDC clinical examination score. For the secondary aim of evaluating sex differences, the detectable mean difference between females and males corresponding to 80% power was estimated to be 2% larger than those specified above. This calculation assumed a female to male ratio of 3:2 based on the 7-year data. Post hoc power calculations for other outcomes resulted in estimated power of 80% for detecting mean differences between groups of 2.6 points for the OARSI radiographic difference score; 15.1 points for the WORMS difference score; 9.5% for the 1-legged hop test ratio; and 9.8 points for the SF-36 physical functioning subscale.

Participants and Entry Criteria

All patients who presented with an isolated unilateral ACL injury in the clinics of three surgeons (P.D.F., M.J.H., R.M.S.) over three years (February 2004 to February 2007) were evaluated for eligibility as previously described.^4,18 Briefly, female and male patients between 15 and 50 years of age with a unilateral ACL injury who were candidates for surgical reconstruction with a bone–patellar tendon–bone (B-PT-B) or a 4-stranded semitendinosus and gracilis (4-ST/G) autograft were included. Patients more than 12 months out from their date of ACL injury; with meniscal tears requiring partial meniscectomy involving more than one-third of the meniscus; with increased clinical laxity of the medial collateral ligament (MCL), lateral collateral ligament (LCL), or posterior cruciate ligament (PCL); a previous knee injury; or radiographic evidence of degenerative arthritis were excluded from participation. Of the 557 patients screened, 355 were excluded and 112 declined to participate.^4,18 An uninjured control group was also recruited from the local area via advertisement.^4,18 This group was comprised of 60 participants who were matched to the surgical subjects by age (i.e., 18–50 years), sex, race (i.e., African American, Hispanic, Caucasian, Other), and activity level (i.e., Tegner score ≥ 2). Control participants were excluded if they had a previous knee injury, increased clinical knee laxity of the MCL, LCL, or PCL relative to the control knee, or evidence of degenerative arthritis on radiographic assessment.

ACL Reconstruction/Initial Graft Tension Protocols

ACLs were reconstructed either with a B-PT-B autograft obtained from the central third of the ipsilateral patellar tendon or a 4-ST/G autograft.^4,18 Participants chose the graft type in consultation with their surgeon and all surgeons adhered to the same operative procedures, which entailed use of the transtibial technique for femoral tunnel drilling. Grafts were preconditioned with 20 manual tension cycles prior to fixation. Bone blocks for the B-PT-B autografts were secured using interference screws (Titanium Interference screws; 7×20; 8×20, 9×20; Arthrex, Naples FL), and the 4-ST/G autografts were fixed with cortical fixation on the femur (Endobutton; Smith & Nephew; Mansfield, MA) and a biodegradable interference screw on the tibia (Biointrafix, small or large sheath with 8 or 9 mm tapered screw; Depuy Mytec; Raynham, MA), reinforced at the surgeon’s discretion with a screw and spiked soft tissue washer (ACUFEX Spiked Washer/Tibial Anchor; Smith & Nephew; Mansfield, MA). Grafts were tensioned following a “laxity-based” approach, where the level of graft tension (i.e., low vs. high) was determined indirectly based on the AP laxity value of the index knee relative to the contralateral knee at the time of fixation. Grafts for the low-tension assignment were tensioned with the knee at 0° of flexion such that the AP laxity value of the index knee matched that of the contralateral knee,^4,18 and grafts for the high-tension assignment were tensioned with the knee at 30° of flexion such that the AP laxity value of the index knee was over-constrained by 2 mm relative to that of the contralateral knee.^4,18 The rationale behind this laxity-based approach is that the level of tension applied to recreate normal AP knee laxity would be less than that required to over-constrain the joint within the same patient if the tensioning was performed at the same knee flexion angle. The knee flexion angles used in this protocol contributed to establishing the distinct graft tension conditions, as over-constraining AP laxity by firmly tensioning the graft with the knee at 30° of flexion (i.e., the high-tension assignment), the knee angle where the ACL is under less tension, would result in heightened graft tension across all angles of the extension-flexion cycle compared to the low tension group in which the graft was tensioned in full extension (0°), where the ACL is at the greatest tension. Tibial fixation was partially engaged for both graft types, and AP laxity at 20° of flexion was evaluated using a sterilizable knee arthrometer (KT-1000S; MEDmetric Corp, San Diego CA) and compared with that of the contralateral knee under anesthesia. If the targeted knee laxity value was not achieved within 1mm, fixation was released, and the tensioning procedure was repeated. The laxity value was verified once the fixation procedure was complete. Postoperatively, all participants followed a standardized rehabilitation program designed for return to sport at six months.⁸

Clinical Outcomes

AP laxity values were measured for both knees using a knee arthrometer (KT-1000, MEDmetric Corp).¹⁵ Three manual maximum tests were performed, and the displacement readings averaged. The difference between knees (index - contralateral) was calculated, compared, and reported.

Clinical outcomes were assessed using the 2000 IKDC examination score.²⁴ This score rates knees as normal (A), nearly normal (B), abnormal (C), or severely abnormal (D) based on knee function, symptoms, and range of motion, with the final IKDC examination rating based on the worst overall score. A trained sports physical therapist (SLF) administered all clinical examinations.

Functional Outcome

Participants performed the 1-legged hop test for distance three times per leg, and the trials within each leg were averaged.¹¹ A mean hop deficit value was calculated as the quotient of the average index limb hop values and average contralateral limb hop values multiplied by 100.

Patient-Reported Outcomes

The KOOS⁴¹ and SF-36⁵¹ questionnaires assessed patient-reported outcomes. The KOOS evaluates 5 domains: (1) quality of life (KOOS-QOL), (2) sports and recreation (KOOS-sport), (3) activities of daily living (KOOS-ADL), (4) symptoms (KOOS-symptoms), and (5) pain (KOOS-pain). In addition, the composite KOOS model developed by Englund et al.¹⁶ was used to identify patients with symptomatic osteoarthritis. Patients with a KOOS-QOL value less than or equal to 87.5 and with at least two of the other subscales meeting the model criteria (i.e., KOOS-pain ≤ 86.1, KOOS-symptoms ≤ 85.7, KOOS-ADL ≤ 86.8, and KOOS-sport ≤ 85.0) were designated as having symptomatic osteoarthritis.^16,52 The SF-36 evaluates health status related to physical functioning (PF), physical role limitations (PR), bodily pain (BP), vitality (V), social functioning (SF), emotional role (ER), mental health (MH), and general health (GH).⁵¹

Physical activity levels were monitored using the Tegner activity scale, which grades activity level based on work and sports activity on a scale of 0–10, where 0 represents physical disability due to knee problems and 10 represents national or international level soccer participation.⁴⁸ In addition, a supplemental survey where participants were asked if they had any subsequent knee injuries that required a doctor’s visit, how satisfied they were with their surgical outcome (1=not satisfied, 10=very satisfied), and if they would elect to undergo ACL reconstruction surgery again was administered.

Osteoarthritis Imaging Outcomes

Radiographic Scoring

The overall condition of the knee joint was evaluated using the modified OARSI radiographic score.⁵ Posteroanterior and lateral radiographs of both knees were graded on a scale of zero (normal) to four (severe) based on osteophyte formation and joint space narrowing. In addition, sclerosis, attrition, and ligament calcification were assessed on a dichotomous scale. An experienced musculoskeletal radiologist (HCG) scored all radiographs while blinded to group.

MRI Scoring

PTOA was also assessed via MRI using the WORMS.³⁹ This scoring system uses magnetic resonance sequences to grade 14 independent features in 15 joint regions.³⁹ These features include cartilage signal and morphology, subarticular bone marrow abnormalities, subarticular cysts, subarticular bone attrition, and marginal osteophytes. The conditions of the menisci and cruciate and collateral ligaments, as well as the presence of loose bodies and periarticular cysts, were also assessed as part of this score. The WORMS evaluations were performed by an experienced musculoskeletal radiologist (HCG) who was blinded to group assignment. Details pertaining to the MRI sequences used for WORMS analysis are included below (Table 1).

Table 1.

MRI sequences and acquisition parameters for WORMS analysis.

MRI sequence	Acquisition parameters
Sagittal T1-weighted water-excitation three-dimensional (3D) fast low-angle shot (3D FLASH)	20/7.6 [TR msec/TE msec]; 12° [flip angle]; 160 mm [field of view, FOV]; 1.5 mm/0 [slice thickness/interslice gap]; 80 slices per slab; 130 hz/pixel [bandwidth, BW]; 512×512 [matrix]; right/left [phase encoding axis]; one average of two excitations.
Coronal intermediate-weighted turbo-spin echo (TSE)	3850/29; 7 [echo train length]; 140 mm; 3 mm/0 mm; 41 slices; 352 hz/pixel; 307×384 [matrix]; anterior/posterior; one average.
Sagittal T2*-weighted WE-3D double echo steady state (WE-3D DESS) #	16.3/4.7; 25°; 140 mm; 0.7 mm/0 mm; 185 hz/pixel; 307×384; anterior/posterior; one average.
Sagittal intermediate-weighted TSE with fat-saturation	3460/36; 5 ETL; 160 mm; 3 mm/0 mm; 248 hz/pixel; 314×448; superior/inferior; one average.

Statistical Analysis

Two-way mixed model analyses of variance were used to evaluate sex and group differences in long-term (10-to-12-year) continuous outcome measures. The statistical models included fixed factors representing sex (female, male), group (low-tension, high-tension, and control), and their interaction. If significant interactions were detected, F-tests corresponding to simple effects (i.e., group sex effects within group or group effects within sex) were examined. Pairwise comparisons among group means were performed using Fisher’s Least Significant Difference (LSD) procedure. Linear contrasts were constructed to test for sex differences limiting comparisons to subjects in the two surgical groups (i.e., excluding control subjects). For those outcome measures reported as a difference score between index and contralateral knees (i.e., AP Laxity, OARSI, WORMS) or as a ratio (i.e., 1-legged hop test), the left or right knee was randomly chosen in control subjects to represent the index limb. For limb specific outcomes, paired t-tests were used to compare index to contralateral within groups. All means presented for continuous outcomes are least-squares means (95% CIs).

Chi-square tests were used to evaluate sex and group effects on IKDC examination scores, the frequency of subsequent injuries, and the percentage of patients in which the composite KOOS model⁵² indicated the presence of symptomatic osteoarthritis. Statistical significance was evaluated based on P<.05. All analyses were performed using SAS statistical software (SAS Institute, Cary, NC).

RESULTS

Patient Characteristics

At 10 to 12 years post-ACLR, 32 low-tension (18 females and 14 males), 28 high-tension (16 females and 12 males), and 35 control (15 females and 20 males) subjects were available for follow-up (Fig. 1). There was no significant difference between groups in the rate of loss to follow-up among the available subjects (31%, 36% and 42% for the low-tension, high-tension, and control groups, respectively; P=.492). See the Online Supplement (Table S1) for additional details. Age, weight, number of days from injury to surgery, patient sex, ethnicity, and graft type were not significantly different between groups (Table 2).

Table 2.

Demographic information for participants included in the 10-to-12-year follow-up.

Demographics	Low tension (N=32)	High tension (N=28)	Control (N=35)	P-value
Age in years, mean [SD]	24.8 [9.3]	22.3 [6.9]	26.2 [7.1]	.143
Weight in Kg, mean [SD]	74.0 [14.9]	69.8 [16.7]	69.7 [12.0]	.396
Days to surgery, median [IQR]	90 [62.5,162.5]	99 [56–164]	-	.982
Patient Sex				.429
Females, # (%)	18 (56.3)	16 (57.1)	15 (42.9)	-
Males, # (%)	14 (43.8)	12 (42.9)	20 (57.1)	-
Ethnicity				.502
White (Non-Hispanic), # (%)	31 (96.9)	25 (89.3)	32 (91.4)	-
Other # (%)	1 (3.1)	3 (10.7)	3 (8.6)	-
Graft type				.496
B-PT-B	19 (59.4)	19 (67.9)	-	-
4-ST/G	13 (40.6)	9 (32.1)	-	-

SD = standard deviation, IQR = interquartile range, B-PT-B = bone-patellar tendon-bone graft, 4-ST/G = 4 stranded hamstrings tendon graft.

Clinical Outcomes

The distribution of IKDC examination scores was significantly different across groups (Table 3), with both the low- and high-tension groups scoring significantly worse than the control group (P<.001 and P=.021, respectively). No significant difference was observed between the low- and high-tension groups (P=.063), and there was no evidence of a difference in the distribution of IKDC examination scores between females and males. A detailed presentation of the IKDC examination results can be found in the Online Supplement (Tables S2–S13).

Table 3.

IKDC examination outcomes overall and among female and male patients at 10 to 12 years post-ACLR (Number [%]).

	Overall			Females			Males
IKDC Exam Score	Low tension* (N=27)	High tension* (N=22)	Control (N=27)	Low tension (N=15)	High tension (N=13)	Control (N=11)	Low tension (N=12)	High tension (N=9)	Control (N=16)	Group P-value	Sex P-value	Interaction P-value
A	7 (25.9)	9 (40.9)	22 (81.5)	4 (26.7)	6 (46.2)	9 (81.8)	3 (25.0)	3 (33.3)	13 (81.3)	<.001	.726	--
B	19 (70.4)	8 (36.4)	4 (14.8)	10 (66.7)	4 (30.8)	1 (9.1)	9 (75.0)	4 (44.4)	3 (18.8)
C	1 (3.7)	3 (13.6)	0 (0.0)	1 (6.7)	2 (15.4)	0 (0.0)	0 (0.0)	1 (11.1)	0 (0.0)
D	0 (0.0)	2 (9.1)	1 (3.7)	0 (0.0)	1 (7.7)	1 (9.1)	0 (0.0)	1 (11.1)	0 (0.0)

An asterisk (*) denotes values that are significantly different from the control group.

There were no significant group or sex differences in AP laxity as measured by the knee arthrometer, and no evidence of interaction between these factors (Table 4). However, AP laxity was significantly worse in the index knee as compared to the contralateral knee within the high-tension group (P=.030). No significant differences were detected in AP laxity between the low- and high-tension groups (P=0.611). Additional information is available in the Online Supplement (Table S14).

Table 4.

Knee arthrometer AP laxity (index – contralateral) and 1-legged hop deficit ([(index/contralateral)*100]) outcomes at 10 to 12 years post-ACLR (Mean [95% CI]).

	Overall			Females			Males
Outcomes	Low tension	High tension	Control	Low tension	High tension	Control	Low tension	High tension	Control	Group P-value	Sex P-value	Interaction P-value
AP Laxity	1.2 [−0.2, 2.7] N=27	1.8# [0.2,3.4] N=22	0.1 [−1.4,1.5] N=27	1.6 [−0.3,3.5] N=15	2.1 [0.1,4.2] N=13	−0.2 [−2.4,2.0] N=11	1.0 [−1.2,3.0] N=12	1.5 [−1.0,3.9] N=9	0.3 [−1.5,2.2] N=16	.265	.761	.803
Hop Deficit	96 [91,102] N=25	93# [87,98] N=22	99 [94,104] N=26	94 [87,100] N=15	92 [85,99] N=13	97 [89,105] N=10	99 [91,107] N=10	94 [85,102] N=9	100 [94,107] N=16	.290	.272	.919

A number symbol (#) denotes values that are significantly different between index and contralateral knees.

Functional Outcomes

There were no significant sex or group differences in hop deficit, and no evidence of interaction between these factors (Table 4). Additionally, no differences in hop distance were detected between the low- and high-tension groups (P=.331). However, significant differences in hop distance were observed between the index and contralateral knees within the high-tension group (P=.011). Additional details are available in the Online Supplement (Table S15).

Patient-Reported Outcomes

Group differences were significant for all five KOOS subscales (Table 5). KOOS subscale differences (Mean±SE) for control vs. low- and high-tension groups, respectively, were as follows: KOOS-symptoms: 9±3, 6±3; KOOS-pain: 6±2, 6±3; KOOS-ADL: 3±2, 5±2; KOOS-sport: 9±3, 8±4; KOOS-QOL: 9±4, 16±5. The low-tension group scored significantly worse than the control group for KOOS-symptoms (P=.016), and the low- and high-tension groups both scored significantly worse than the control group for KOOS-pain (P=.008 and P=.013, respectively), KOOS-sport (P=.018 and P=.037, respectively), KOOS-QOL (P=.045 and P<.001, respectively), and KOOS-ADL (P=.049 and P=.008, respectively). No significant differences were observed between the low- and high-tension groups for any of the KOOS subscales (P≥.101). In addition, there was no significant difference between females and males or evidence of any group by sex interaction for these outcomes (Table 5).

Table 5.

Patient-reported outcomes (KOOS, Tegner, Patient satisfaction) at 10 to 12 years post-ACLR (Mean [95% CI]).

	Overall			Females			Males
PRO	Low tension (N=32)	High tension (N=27)	Control (N=35)	Low tension (N=18)	High tension (N=15)	Control (N=15)	Low tension (N=14)	High tension (N=12)	Control (N=20)	Group P-value	Sex P-value	Interaction P-value
KOOS Sym	83* [79,88]	85 [80,90]	92 [87,96]	85 [78,91]	86 [79,93]	90 [83,97]	82 [75,89]	84 [76,92]	93 [87,99]	.038	.881	.670
KOOS Pain	91* [87,94]	91* [87,95]	97 [94,101]	92 [88,97]	93 [88,98]	98 [93,103]	89 [84,94]	89 [83,95]	97 [93,102]	.011	.243	.771
KOOS ADL	96* [93,98]	94* [91,97]	99 [97,102]	97 [94,101]	96 [92,100]	99 [95,103]	94 [90,98]	92 [88,97]	100 [96,103]	.022	.181	.570
KOOS Sport	85* [80,90]	86* [80,91]	94 [89,98]	88 [82,95]	85 [78,93]	91 [84,98]	82 [75,90]	86 [78,94]	96 [90,102]	.033	.993	.302
KOOS QOL	85* [79,92]	78* [71,84]	94 [88,100]	85 [77,94]	78 [70,87]	93 [84,102]	85 [76,94]	77 [67,87]	95 [88,103]	.002	.928	.915
Tegner Score	4.6^ [4.0,5.2]	5.3 [4.6,5.9]	5.1 [4.5,5.7]	3.8 [3.0,4.6]	4.9 [4.1,5.8]	4.7 [3.8,5.6]	5.4 [4.5,6.4]	5.6 [4.6,6.6]	5.5 [4.7,6.2]	.375	.007	.526
Satisfaction	8.9 [8.3,9.4]	9.0 [8.4,9.6]	--	8.9 [8.1,9.6]	9.2 [8.4,10]	--	8.9 [8.0,9.7]	8.8 [7.8,9.7]	--	.821	.580	.633

An asterisk (*) denotes values that are significantly different from the control group.

A caret symbol (^) denotes values that are significantly different between females and males.

PRO = patient-reported outcome; KOOS Sym = KOOS symptoms subscale; KOOS Sport = KOOS sport and recreation function subscale; KOOS QOL = KOOS knee-related quality of life subscale; and KOOS ADL = KOOS activities of daily living subscale. The KOOS-symptoms, Tegner, and Satisfaction outcomes had one additional female subject in the high-tension group, resulting in an overall sample size of 28 for that group. These differences in sample size are not indicated in the table.

Based on the composite KOOS model, ten patients in the low-tension group (31.3%), seven patients in the high-tension group (25.9%), and three patients in the control group (8.6%) had scores indicative of symptomatic osteoarthritis. Overall, no sex or group differences were observed (P=.368 and P=.060, respectively), and no significant differences were detected between the low- and high-tension groups (P=.653). However, when this analysis was performed between the combined tension groups and control group, the group difference was significant (P=.020). See the Online Supplement (Table S16–S17) for additional information.

There were no significant sex or group differences for any of the SF-36 subscales (P ≥ 0.180 and P ≥ 0.221, respectively), and no evidence of interaction between these factors (P≥0.249). Additionally, no significant differences were observed between the low- and high-tension groups on any of the SF-36 subscales (P≥.214). See the Online Supplement (Table S18) for detailed information.

There were no significant overall group differences for the Tegner activity level score (Table 5), and no significant differences were detected between the low- and high-tension groups (P=.176). However, significant differences between females and males were observed for this outcome (Table 5). Although there was no evidence that sex differences were group dependent (P=.523), this difference was most pronounced in the low-tension group in which females scored significantly worse than males (P=.013). Male and female activity levels were not significantly different in either the high-tension group or control group (P=.341 and P=.198, respectively).

There were no significant sex or group differences between low- and high- tension groups (P=.821) for the Satisfaction outcome, and no evidence of interaction between these factors. Satisfaction was generally high, averaging approximately nine on a 10-point scale among the surgical tension groups (Table 5).

Twenty-eight subjects included in the 10- and 12-year follow-up reported a total of 39 subsequent injuries since their initial baseline visit. Of these injuries, five were ACL graft failures and eight were contralateral ACL tears. Of the ACL graft failures, three occurred in the high-tension group and two occurred in the low-tension group. Of the contralateral ACL tears, three occurred in the low-tension group, four occurred in the high-tension group, and one occurred in the control group. See the Online Supplement (Table S19) for a detailed history of subsequent injuries in the affected ACLR and control subjects.

Imaging Outcomes

Significant group effects were observed for the MRI-based WORM difference score (i.e., index-contralateral) (Table 6). The low- and high-tension groups scored significantly worse than the control group (P=.002 and P=.042, respectively), although there was no significant difference between tension groups (P=.374) (Table 6). Mean WORM scores for the index knee were significantly worse than those for the contralateral knee in both the low- and high-tension groups (P=.001 and P=.050, respectively). These knee differences were primarily due to male patients whose index knee had higher scores, indicating a worse overall knee condition, than the contralateral knee in both the low- (P<.001) and high-tension (P=.058) groups. WORM difference scores were not significantly different between females and males when control subjects were included (P=.061) but were significantly higher in males compared to females among the two tension groups exclusively (P=.008; Online Supplement, Table S20). No significant interaction of sex and group was observed (Table 6).

Table 6.

WORM and OARSI difference score outcomes at 10 to 12 years post-ACLR (Mean [95% CI]).

	Overall			Females			Males
Image Score	Low tension	High tension	Control	Low tension	High tension	Control	Low tension	High tension	Control	Group P-value	Sex P-value	Interaction P-value
WORM Score	13.3*,# [6.1,20.6] N=26	8.3*,# [0.2,16.5] N=21	−3.4 [−10.8,4.1] N=25	2.5 [−6.9,11.9] N=15	4.0 [−6.1,14,1] N=13	−1.1 [−12.6,10.4] N=10	24.1 [13.2,35.1] N=11	12.7 [−0.2,25.6] N=8	−5.7 [−15.1,3.7] N=15	.008	.061	.054
OARSI Score	2.4*,^,# [1.1,3.8] N=27	1.0 [−0.6,2.6] N=21	0.0 [−1.4,1.4] N=26	0.5 [−1.3,2.3] N=15	0.1 [−1.9,2.0] N=13	0.1 [−2.0,2.2] N=11	4.4 [2.4,6.4] N=12	1.9 [−0.5,4.4] N=8	−0.2 [−2.0,1.6] N=15	.050	.034	.106

An asterisk (*) denotes values that are significantly different from the control group

a caret symbol (^) denotes values that are significantly different between females and males

a number symbol (#) denotes values that are significantly different between index and contralateral knees.

Significant group differences were observed for the radiograph-based OARSI difference score (i.e., index-contralateral) (P=.050, Table 5), with the low-tension group scoring significantly worse relative to the control group (P=.015). There was no significant difference in this score between the high-tension and control groups (P=.341), and no significant difference was observed between tension groups (P=.179). Additionally, the index and contralateral knees had significantly different OARSI scores within the low-tension group (P=.001), which was primarily due to the male patients within that group whose index knee scored significantly worse than the contralateral knee (P<.001). Furthermore, males had significantly higher OARSI difference scores than females (P=.034; Table 6). This difference was most pronounced in the low-tension group in which male patients scored significantly worse than female patients (P=.006). The difference between females and males remained significant when analyses were limited to surgical patients in the two tension groups exclusively (P=.008; Online Supplement, Table S20). No interaction of sex and group was observed (Table 6).

DISCUSSION

This study sought to evaluate how initial graft tension and patient sex influence PTOA outcomes at 10 to 12 years post-ACLR. The results demonstrate no significant differences between the tension groups, as hypothesized. However, independent group and sex main effects were observed for the IKDC examination, KOOS, WORM difference score, and OARSI difference score outcomes, and the Tegner and OARSI difference score outcomes, respectively. The finding of patient sex-based differences does not support the hypothesis that there would be no differences in outcomes between female and male patients at 10 to 12 years post-ACLR.

No significant differences between the low- and high-tension groups were detected for any of the functional or clinical outcomes assessed in this study. This finding is consistent with several randomized trials that have evaluated the effects of initial graft tension on clinical outcomes after ACL reconstruction. For hamstring tendon grafts, Kim et al. found no significant differences in knee laxity when grafts were tensioned to 78, 117, and 146 N at one-year after surgery.²⁷ For BTB grafts, Yoshiya et al. and van Kampen et al. found no differences in knee laxity based on initial graft tension post-ACLR using BTB grafts.^50,56 Similarly, Chahal et al. recently investigated differences in post-operative outcomes based on graft fixation angle (i.e., 0° versus 30° degrees of flexion), which corresponds to low and high graft tensions, respectively, and determined no differences between groups for AP stability, as well as no differences in any of the KOOS subdomains; although, secondarily, they did find that patients who underwent graft fixation in full extension had higher Marx activity scores and were more likely to reach the minimal clinically important difference for KOOS-pain.¹² Conversely, Nicholas et al. reported significant improvements in knee laxity when BTB grafts were tensioned to 90 N as compared to 45 N, although no differences were detected in other functional outcomes, including the Knee Outcome Survey and single-limb hop test.³⁷ Evidence-based reviews by Arneja et al. and Kirwan et al. summarize the discrepancies concerning the effects of initial graft tension on post-operative outcomes well, concluding that there is no clear trend in terms of statistically significant or clinically relevant differences based on initial graft tension and that there is insufficient evidence to conclude whether postoperative function is improved at any specific tension level, respectively.^6,28 These findings, which align with those of the present study, call into question the clinical relevance of initial graft tension as relates to post-operative outcomes, and warrant future investigation.

The low- and high-tension groups scored significantly worse than the control group for the IKDC examination, KOOS (pain, ADL, sport, and QOL subscales), and WORM difference score outcomes at 10 to 12 years post-ACLR. Collectively, these findings suggest that ACLR failed to mitigate PTOA risk regardless of initial graft tension, which is consistent with the existing literature.^4,18,19 Interestingly, the low-tension group alone scored worse than the control group for the KOOS-symptoms and OARSI difference score outcomes, suggesting that a low initial graft tension may predispose patients to arthrosis that is not reflected by the IKDC examination, KOOS (Pain, ADL, Sport, and QOL subscales), or WORM difference score outcomes. However, among female and male patients, the low-tension group tended to have worse IKDC examination scores relative to the control group (Online Supplement; Table S9–S10), suggesting that this score may have an emerging sensitivity to osteoarthritic changes within this group. In addition, the difference between the high-tension and control groups for KOOS-symptoms approached significance at this time point (P=.059), suggesting that arthrosis may be progressing in this group. This point is consistent with the findings that the index knee scored worse than the contralateral knee within the high-tension group for the AP laxity and 1-legged hop deficit outcomes (Online Supplement; Table S14–S15), which could suggest the progression of arthrosis in the surgical knee within this group. Additional long-term follow-up is warranted to determine whether the KOOS-symptoms trend will emerge as significant at later time points.

Female patients had significantly worse Tegner scores than male patients at 10 to 12 years post-ACLR. This finding suggests that female patients maintained a lower physical activity level than male patients, which could suggest a relative functional deficit between female and male patients postoperatively. However, female sex did not emerge as a risk factor for any of the other outcomes assessed and this result was also observed in the control group, which suggests that this finding may owe to other extrinsic factors unrelated to surgery. In contrast, male patients had significantly worse OARSI difference scores relative to female patients, suggesting that males have more radiographic signs of PTOA relative to females at this post-surgery time point. In addition, there is evidence that this result was primarily influenced by male patients within the low-tension group, suggesting that a low initial graft tension may predispose male patients to worse PTOA. Collectively, the OARSI difference score and Tegner sex-based findings may indicate that a higher level of physical activity post-ACLR promotes PTOA in the affected knee, which is consistent with the results of a long-term study conducted by Curado J et al., where a moderate or strenuous level of physical activity 22 years post-ACLR, specifically engagement in pivot or pivot-contact sports, was identified as a PTOA risk factor.¹³ Importantly, the OARSI sex-based result aligns with the findings that the index knee had significantly worse OARSI and WORM difference score outcomes relative to the contralateral knee among males within the low-tension group, implicating the progression of osteoarthritis in the surgical knee among males within this group. This result is also consistent with the composite KOOS model assessment of symptomatic osteoarthritis, which demonstrated that the difference between the combined tension groups and the control group approached significance among male patients (Online Supplement; Tables S16–S17). Furthermore, there was evidence that this difference was primarily influenced by the low-tension group, again suggesting male sex and low initial graft tension as risk factors for PTOA (Online Supplement; Tables S16–S17). Additionally, the strong trend that male patients in the low-tension group had worse WORM difference scores than female patients further support this finding. As no significant differences between tension groups or interaction of sex and group were observed, these findings bear further investigation.

Examination of OARSI difference and WORM difference scores revealed osteophytes and cartilage degeneration as the primary drivers of poor imaging outcomes at the 10-to-12-year time point. Given this agreement, it is unclear why male sex emerged as a significant risk factor for worse OARSI difference scores, but not worse WORM difference scores. Interestingly, when this analysis was performed among the tension groups exclusively, this difference became significant, suggesting that the effect of male sex on the WORM difference scores may be dependent on initial graft tension (Online Supplement; Table S20). Moreover, the clinical significance of these findings remains unclear, as male sex did not emerge as a risk factor for the other outcomes assessed at this post-surgery time point. Furthermore, the finding of male sex as a potential PTOA risk factor is largely inconsistent with the existing literature, which—albeit controversial—predominantly reports female sex as a PTOA risk factor post-ACLR (or no sex-based effects at all).^26,35 However, a study conducted by Salmon et al. reported a trend where early, radiologic signs of degeneration were higher in males relative to females at seven years post-ACLR.⁴² Additional long-term studies of PTOA risk among female and male patients following ACLR are therefore warranted.

Previous studies have reported female sex as a risk factor for worse KOOS (Sport,² QOL,² and Pain³⁰), knee laxity,^42,47,53 SF-36 (Role Physical, Bodily Pain, and General Health),¹⁷ and IKDC examination^30,53 outcomes post-ACLR, whereas the present study detected no sex differences in these assessments. While the reason(s) for these discrepancies remain unclear, it is important to note that the referenced studies were conducted with short- to mid-term follow-up, while the present study was conducted with longer-term follow-up. Thus, it is possible that the reported sex differences occur at early post-surgery time points and resolve in the long-term, as suggested by a study where minor sex differences were initially observed for KOOS scores at nine months postoperatively but were no longer detectable at the 12-month follow-up.⁴⁹ Furthermore, as the cohort leveraged in the present study was only assessed for sex differences at 10 to 12 years postoperatively, the time point at which these differences emerged is unknown. Future research is therefore required to investigate the emergence and duration patterns for sex differences in these post-operative outcomes.

There are several study limitations to consider. Given the loss to follow-up rate at 10 to 12 years postoperatively (36.7%), the study was sufficiently powered to detect relatively large effect sizes (Cohen’s d=.74). As a result, smaller magnitude differences cannot be ruled out. In addition, although the radiologist was blinded to treatment group, it is possible that the presence of screws and tunnels in the surgical knee biased scoring of the radiograph or MR images. However, the radiologist would not have been able to distinguish the contralateral images between the control and tension groups, or initial graft tension among knees reconstructed with either autograft type. An additional limitation to consider is the use of both B-PT-B and 4-ST/G autografts, which was initially justified because previous studies had shown that the autograft types had similar biomechanical properties^54,55 and post-operative outcomes, and were thus thought not to result in post-operative outcome differences.^22,23,34 However, although both graft types are deemed viable options for primary ACLR, recent meta-analyses comparing the post-operative outcomes of B-PT-B and 4-ST/G autograft use in ACLR demonstrate that 4-ST/G autografts may fail at a higher rate and produce inferior static knee stability relative to B-PT-B autografts.^20,43,45 Furthermore, ACLR with 4-ST/G autografts has also been shown to produce worse outcomes in female patients relative to male patients.^32,46,42 These graft-based differences could potentially confound the findings of this study; however, the proportion of graft types in the two tension groups was similar and thus should have mitigated this effect (Online Supplement; Table S1). Ultimately, the potential effect of graft type on outcomes could not be assessed as graft type was self-selected (i.e., not randomized) and the sample size was insufficient to address this question.

In conclusion, the study supported the hypothesis that ACLR produced inferior outcomes relative to the matched, uninjured control group and that the procedure did not prevent the onset of PTOA in the knee-injured cohort, regardless of initial graft tension. Furthermore, no differences in outcomes were detected between the low- and high-tension groups. However, the results do not support the hypothesis that there would be no difference in outcomes between female and male patients, as sex differences were observed for the Tegner activity level and OARSI difference scores. Cumulatively, these findings suggest that male patients treated with a low initial graft tension may be at higher risk for PTOA postoperatively.