Elbow Ulnar Collateral Ligament Reconstruction and Repair: A Systematic Review and Meta-analysis of Biomechanical Studies

Abstract

Background:

Technical variations in elbow ulnar collateral ligament reconstruction (UCLR) include graft source, graft/tunnel configuration, and humeral and ulnar fixation. While the biomechanical performance of various constructs has been reported, these studies have small sample sizes and compare at most a few technical variations.

Purpose:

To quantitatively synthesize the results of biomechanical investigations of UCLR and repair.

Study Design:

Systematic review.

Methods:

A systematic review and meta-analysis was conducted according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses); included articles were published between 1998 and 2020. Biomechanical metrics were utilized to compute effect sizes (standardized mean difference [SMD]) for quantitative analysis when 2 studies reported the same metric for the same comparison. After our initial search, 1293 studies were identified. Summary effects were estimated in random-effects models, and mixed-effects models were constructed to evaluate the fixed effects of technical variations through meta-regression.

Results:

A total of 24 eligible studies were included, of which 19 were included in the quantitative analysis. Compared with the intact ligament, UCLR had significantly lower ultimate strength (SMD, –1.411; P < .0001) and stiffness (N/mm) (SMD, –3.259; P = .0268), and significantly greater valgus opening at 70° of flexion (SMD, 1.638; P < .0001). Stiffness (N·m/deg), valgus opening angle at 30° and 90° of flexion, and gapping at failure were not significantly different from the intact UCL (all P > .05). There was no significant difference in ultimate strength between docking and Jobe reconstructions (P = .2889). There were no significant differences between repair and reconstruction in ultimate strength, stiffness (N·m/deg), or yield torque (all P > .05).

Conclusion:

Our study demonstrates that, at time zero, UCLR has inferior biomechanical properties compared with the native intact ligament. Biomechanical performance of UCLR was either inferior to the intact UCL (ultimate strength, stiffness [N/mm], and valgus opening at 70° flexion) or not significantly different from it (stiffness [N⋅m/deg], valgus opening at 30° and 90° of flexion, and gapping at failure). There is no difference in biomechanical outcome measurements between docking and Jobe reconstructions, or between UCL repair and reconstruction.

The ulnar collateral ligament (UCL) of the elbow has a serious impact on the careers of overhead athletes, particularly throwers. Given that previous studies have demonstrated the effect of UCL reconstruction (UCLR) on return to sports in adolescent as well as adult athletes, the increase in primary UCL injury is both noteworthy and concerning.^34,38,41 Further, while multiple surgical technical variations of UCLR and UCL repair have been described, reports of the biomechanical performance of these constructs are limited to studies with relatively small sample sizes. ¹⁸ As such, further investigation is warranted.

Common UCLR techniques include the modified Jobe figure-of-8, docking, and hybrid techniques. Classically, the Jobe technique involves passing a palmaris longus graft through tunnels in the ulna and humerus in a figure-of-8 fashion with the graft sutured to itself. ⁷ The docking technique was introduced as an alternative to the Jobe technique. It uses the same ulnar tunnel method as the Jobe technique but requires only a single anterior humeral tunnel and 2 small exit holes. A Krackow or Whip stitch is placed at the end of the graft, which is docked into the humeral tunnel. The ends of the suture are pulled through the exit holes and tied over the bony bridge. ⁷ Similarly, there are various fixation devices, such as interference screws and cortical buttons, that may be used. The literature on outcomes after these techniques is mixed, and no technique has emerged as superior to the others.

Direct UCL repair has specific advantages over reconstruction in restoring normal anatomy, avoiding graft-site morbidity, reducing complications associated with harvesting autografts, and enabling a quicker return to sports. ¹⁷ Repair may be indicated in proximal or distal avulsion tears in biomechanically stable joints without evidence of chronic attritional diseases. ¹⁴ The technique for repair involves both an ulnar tunnel centered at the sublime tubercle and a humeral tunnel, with tunnels located at the centers of the native UCL attachments. Suture anchor implants can be loaded with collagen-coated tape to serve as an internal brace to support the repair. Repair performed with suture tape augmentation has yielded results comparable to those of UCLR. Sutures are used to repair proximal or distal detachment of the UCL and longitudinal split of the ligament. ⁴⁴

This study aimed to quantitatively synthesize the results of biomechanical investigations of UCLR and UCL repair. The authors hypothesized that, regardless of the technique, UCLR would consistently perform inferiorly or at best no different from native intact ligaments, and that there would be no difference in performance between docking and Jobe reconstructions or between reconstruction and repair.

Methods

Eligibility Criteria

Under PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses), biomechanical studies with English-language results of elbow UCLR and/or repair were eligible. To meet the inclusion criteria, eligible studies had to report objective measurements of biomechanical performance and/or properties for at least 1 method/construct/technique of elbow UCLR or repair. Studies with reconstruction results were required to specify the methods of humeral and ulnar fixation, the graft type/source, and the graft/tunnel configuration.

The exclusion criteria for of an entire article were as follows: (1) the article contained no original results for biomechanical analysis of elbow UCLR or repair; (2) it was impossible to determine which results were obtained with which method/construct/technique (eg, authors reported summary statistics for multiple methods/constructs/techniques, which are too different to analyze together for this analysis and did not stratify results for any method/construct/technique); (3) or if the authors did not state a specific method/construct/technique or provide sufficient detail such that the method/construct/technique can be determined for any reported outcome. Any technique was potentially eligible for inclusion. We did not preselect specific techniques for inclusion or exclude any techniques a priori.

The criteria for exclusion of a group/subgroup of results within an otherwise eligible article were as follows: (1) no objective biomechanical results were available for that group/subgroup; (2) it was impossible to distinguish results for a group/subgroup when results could otherwise be determined for at least 1 method/construct/technique in the same article; (3) the method/construct/technique for a group/subgroup was not named or described in sufficient detail to allow proper identification of that method/construct/technique.

Studies were grouped for analysis according to internal comparisons performed (ie, reconstruction vs intact, repair vs intact, reconstruction vs repair, etc).

Information Sources and Search Strategy

A systematic search of the Cochrane Central Register of Controlled Trials (CENTRAL), Cumulative Index to Nursing and Allied Health Literature (CINAHL), Embase, MEDLINE, SPORTDiscus, and Web of Science was performed on March 29, 2020, according to PRISMA using the following terms: (ulnar AND collateral AND ligament OR UCL) AND (repair OR reconstruction) AND Elbow. Included studies were published between 1998 and 2020.

Selection Process

Two authors (A.J.H. and J.H.S.) collected the results of the systematic database searches, screened all titles for relevance, and removed duplicates. The same authors then reviewed abstracts to determine eligibility. Throughout this process, any disagreements were discussed with a senior researcher (A.M.L.) until consensus was reached. These 3 authors (C.C., B.M.D., and N.S.) subsequently reviewed the remaining full-text articles to verify that all inclusion criteria were met and that no articles or groups/subgroups therein met the criteria for exclusion.

Data Collection Process

One author (A.J.H.) independently collected all biomechanical results available in the included studies, organized by repair or reconstruction, and, where applicable, the technical variations of reconstruction. Units for each outcome were recorded. A second, more senior author (A.M.L.) then reviewed and verified these results, and determined which effect sizes were comparable across studies (ie, compatible units, comparable testing conditions, etc).

Data Items

In each included study, the number of specimens with evaluable results was noted, and details about the repair and reconstruction constructs were recorded. For reconstructions specifically, we determined graft source (palmaris longus [PL], hamstring [HS–ie, gracilis or semitendinosus], or extensor digitorum longus [EDL]); graft configuration (triangular [T] or linear [L]); humeral fixation (docked [D], transosseous [TO], interference screw [IS], cortical button [CB], or suture anchor [SA]); and ulnar fixation (TO, IS, CB, or SA). For repairs, we noted augmentation methods (internal bracing)—including humeral and ulnar fixation strategies. We then recorded any reported quantitative biomechanical metrics and corresponding units for native specimens, repairs, and reconstructions. If noted, we also recorded failure modes and the number of specimens that failed in each way.

Effect Measures and Synthesis Methods

Biomechanical metrics were utilized to compute effect sizes for quantitative analysis when at least 2 studies reported the same metric (with the same units of measurement) for the same comparison (eg, native vs reconstruction, native vs repair, reconstruction vs repair). Additionally, since the most commonly performed technical variations for reconstruction utilize triangular graft configurations with transosseous ulnar fixation and either docked or transosseous humeral fixation, biomechanical outcomes were quantitatively evaluated when at least 2 studies reported the same metric for this comparison, as long as the graft source was the same for both groups within each study. Each metric was evaluated separately. Given the inherent but anticipated variability in actual results, likely a result of subtle variations in biomechanical testing conditions from study to study, in each case the effect size for modeling was the standardized mean difference as follows: $SMD=\frac{x_1-x_2}{S{D}_P}$ , where x₁ and x₂ represent sample means and to account and correct for the positive bias in standardized mean difference, ²¹ where n₁ and n₂ and SD₁ and SD₂ are the sample sizes and standard deviations, respectively. Thus, internal comparisons were required for the quantitative methods; no raw results were evaluated quantitatively.

Summary effects were estimated for available comparisons in random effects models with the restricted maximum likelihood (ReML) method. Heterogeneity was quantitatively assessed using the generalized/weighted least squares extension of Cochran’s Q test for which the corresponding P value was reported (denoted P_Q), τ², and I² (expressed as a percentage). Mixed effects models were constructed with suitable reconstruction comparisons to evaluate fixed effects of technical variations (ie, graft type, graft configuration, humeral fixation, and ulnar fixation) as moderators through meta-regression. This was performed when at least 2 studies reported the same biomechanical outcome across at least 2 levels of a given technical variation. Moderator significance was determined using the Q_M test, which measures the variability in observed effect sizes attributable to the moderators. The P value for the Q_M test was denoted P_QM. Pairwise differences between moderator levels were computed via linear hypothesis tests. Models with a moderator were compared with reduced (no moderator) models with a likelihood ratio test and a change in Akaike information criterion (AIC). A reduction in AIC is consistent with a superior fit. Heterogeneity for mixed effects models was quantitatively assessed in the same manner as with random effects models, except that the Q_E-test for residual heterogeneity was employed.

Analysis was conducted in the R statistical environment (R Version 4.0.4; R Foundation for Statistical Computing) with the metafor package. ⁴⁶ For all quantitative statistical tests, the criterion for significance was set as P < .05.

Results

Study Selection

The initial search produced 799 nonduplicated abstracts, with 24 articles ^‡ appropriate for qualitative analysis and 19 articles ^§ for quantitative analysis after full-text review (Figure 1). Three included articles reported results for subgroups (reconstructions) with entirely novel techniques not elsewhere described,^5,32,36 and these subgroups were excluded from analysis.

Figure 1.

PRISMA selection process. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Study Characteristics

All 24 included studies reported reconstruction results. Five included studies contained repair results and compared repair with reconstruction.^{6,15,23,27,45} Sixteen studies actually performed biomechanical tests on the native intact UCL and reported the results. ^‖ Only 1 study with repair results tested ligaments of intact specimens. ⁶ The remaining studies exclusively reported results comparing reconstruction technical variations or reconstruction and repair.^{4,8,15,23,24,27,28,33,45} Characteristics of the included studies are summarized in Tables 1 and 2.

Table 1.

Characteristics of Included Biomechanical Studies with UCLR Results ^a

			Reconstruction Technical Variations
Study (Year)	Journal	N b	Graft	Config	Hum Fix	Uln Fix	Biomechanical Outcomes, Units c
Ahmad et al 2 (2003)	Am J Sports Med	10	PL	L	IS	IA	Mean change in valgus opening angle (reconstructed vs intact), deg
							Stiffness, N/mm
							Ultimate tensile strength, N⋅m c
Armstrong et al 3 (2005)	J Shoulder Elbow Surg	7	PL	L, T	D, IS, TO	CB, IS, TO	Cycles to failure, No.
							Gapping at failure, mm, c Peak load, N
Bernholt et al 5 (2019)	J Shoulder Elbow Surg	12	PL	T	D	TO	Gapping at failure, mm c
							Stiffness, N⋅m/deg c
							Ultimate tensile strength, N⋅m c
Bodendorfer et al 6 (2018)	Am J Sports Med	9	PL	T	D	TO	Gapping at failure, mm c
							Ultimate tensile strength, N⋅m c
							Valgus opening angle, 90° c
Chronister et al 8 (2014)	J Hand Surg	12	PL	L	D, TO	IS, IS	Cycles to failure, No.
							Peak load, N
							Stiffness, N/mm
Ciccotti et al 9 (2009)	Am J Sports Med	10	PL	T	D, TO	TO	Ultimate tensile strength, N⋅m c
Cohen et al 10 (2015)	J Shoulder Elbow Surg	10	PL	T	D	TO	Ultimate tensile strength, N⋅m c
Dargel et al 13 (2015)	J Orthop Sci	10	PL	T	D	TO	Ultimate tensile strength, N⋅m c Valgus opening angle at 0°
							Valgus opening angle at 30°
							Valgus opening angle at 60°
							Valgus opening angle at 90° c
Dugas et al 15 (2016)	Am J Sports Med	9	PL	T	TO	TO	Gapping at 10 N⋅m, mm
							Stiffness, Nm/deg c
							Ultimate tensile strength, N⋅m c
Duggan et al 16 (2011)	J Shoulder Elbow Surg	6	HS	T	D	TO	Articular contact pressure at 30°, MPa
							Articular contact pressure at 90°, MPa Radiocapitellar contact area at 30°, mm2 Radiocapitellar contact area at 90°, mm2
							Valgus opening angle at 30° c
							Valgus opening angle at 90° c
Hechtman et al 20 (1998)	Am J Sports Med	31	PL	L, T	SA, TO	SA, TO	Ultimate tensile strength, N⋅m c
Hurbanek et al 22 (2009)	Am J Sports Med	18	PL	T	D, IS	TO	Ultimate tensile strength, N⋅m
							Valgus opening angle at 30° c Valgus opening angle at 60°
							Valgus opening angle at 90° c
Itami et al 23 (2021)	J Shoulder Elbow Surg	7	PL	T	D	TO	Stiffness, N/mm c
							Ultimate tensile strength, N⋅m c
							Valgus opening angle at 90° c
Jackson et al 24 (2013)	Am J Sports Med	12	EDL	L, T	CB, D	CB, TO	Mean valgus opening angle, deg
							Stiffness, N⋅m/deg c
							Ultimate tensile strength, N⋅m c
							Yield torque, N⋅m c
Jones et al 27 (2018)	Orthop J Sports Med	10	PL	T	TO	TO	Gapping at 10 N⋅m, mm
Large et al 28 (2007)	Arthroscopy	20	HS	L, T	IS, TO	IS, TO	Ultimate tensile strength, N⋅m c
McAdams et al 31 (2007)	J Shoulder Elbow Surg	16	PL	L, T	D, IS	IS, TO	Valgus opening angle at 70°
McGraw et al 32 (2013)	Am J Sports Med	20	PL	T	D, TO	TO	Stiffness, N/mm c
							Ultimate tensile strength, N⋅m c
Morgan et al 33 (2010)	Am J Sports Med	16	HS	L, T	IS, TO	CB, TO	Stiffness, Nm/deg
							Valgus opening angle at 70° d
Paletta et al 35 (1998)	Am J Sports Med	23	PL	T	D, TO	TO	Stiffness, N⋅m, mm/mm
							Strain at maximal moment, mm/mm
							Ultimate tensile strength, N⋅m c
Ruland et al 36 (2008)	Am J Sports Med	10	PL	T	D	TO	Stiffness, N⋅m/deg c
							Ultimate tensile strength, N⋅m c
Seiber et al 39 (2010)	Clin Biomech	18	PL	L, T	IS, TO	IS, TO	Ultimate tensile strength, N⋅m c
							Yield torque, N⋅m c
Shah et al 40 (2009)	J Shoulder Elbow Surg	32	PL	L, T	D, IS, TO	IS, TO	Valgus opening angle at 70° c
Urch et al 45 (2019)	Orthop J Sports Med	8	PL	T	D	TO	Stiffness, N⋅m/deg c
							Ultimate tensile strength, N⋅m c Yield angle, deg
							Yield torque, N⋅m c

^aAm J Sports Med, American Journal of Sports Medicine; Arthroscopy, Arthroscopy: The Journal of Arthroscopic and Related Surgery; CB, cortical button; Clin Biomech, Clinical Biomechanics; Config, configuration (ie, graft configuration); EDL, extensor digitorum longus; HS, hamstring; Hum Fix, humeral fixation; IS, interference screw; J Hand Surg, Journal of Hand Surgery; J Orthop Sci, Journal of Orthopaedic Science; J Shoulder Elbow Surg, Journal of Shoulder and Elbow Surgery; L, linear; Orthop J Sports Med, Orthopaedic Journal of Sports Medicine; PL, palmaris longus; SA, suture anchor; T, triangular; TO, transosseous; UCLR, ulnar collateral ligament reconstruction; Uln Fix, ulnar fixation.

^bTotal number of specimens.

^cOutcomes used in quantitative methods.

^dDid not contribute to quantitative synthesis due to lack of suitable comparisons.

Table 2.

Characteristics of Included Biomechanical Studies with UCL Repair Results ^a

			Reconstruction Technical Variations
Study (Year)	Journal	N b	Augmentation	Hum Fix	Uln Fix	Biomechanical Outcomes, Units
Bodendorfer et al 6 (2018)	Am J Sports Med	9	2-mm FiberTape c	SA d	SA d	Gapping at failure, mm Ultimate tensile strength, N⋅m Valgus opening angle at 90°
Dugas et al 15 (2016)	Am J Sports Med	9	2-mm FiberTape c	SA d	SA d	Gapping at 10 N⋅m, mm Stiffness, N⋅m/deg Ultimate tensile strength, N⋅m
Itami et al 23 (2021)	J Shoulder Elbow Surg	7	2-mm FiberTape c	SA d	TO	Stiffness, N/mm Ultimate tensile strength, N⋅m Valgus opening angle at 90°
Jones et al 27 (2018)	Orthop J Sports Med	10	2-mm FiberTape c	SA d	SA d	Cyclic load gapping, mm
Urch et al 45 (2019)	Orthop J Sports Med	8	1.3-mm SutureTape c	SA d	SA e	Stiffness, N⋅m/deg Ultimate tensile strength, N⋅m Yield angle, deg Yield torque, N⋅m

^aAm J Sports Med, American Journal of Sports Medicine; Hum Fix, humeral fixation; J Shoulder Elbow Surg, Journal of Shoulder and Elbow Surgery; Orthop J Sports Med, Orthopaedic Journal of Sports Medicine; SA, suture anchor; TO, transosseous; UCL, ulnar collateral ligament; Uln Fix, ulnar fixation.

^bTotal number of specimens.

^cArthrex.

^d3.5-mm SwiveLock (Arthrex).

^e2.9-mm short PushLock (Arthrex).

Results of Individual Studies

The biomechanical results reported by the individual studies are presented in Table 3. Sixteen studies provided failure mechanisms, which are summarized in Table 4.

Table 3.

Results of Biomechanical Evaluations of UCLR and Repair ^a

	Results
							Reconstruction
			Intact		Repair			Technical Variations
Study	Biomechanical Outcome	Unit	N	Mean ± SD	N	Mean ± SD	N	G	C	H	U	Mean ± SD	Conclusions
Ahmad et al 2 (2003)	Stiffness ultimate	N/mm	10	42.81 ± 11.6	0	NR	10	PL	L	IS	IS	20.28 ± 12.5	Comparable failure strength to native physiologic elbow kinematics restored
	Tensile Strength	N·m	10	34.29 ± 6.9	0	NR	10	PL	L	IS	IS	30.55 ± 19.24
Armstrong et al 3 (2005)	Cycles to failure	N	7	2536	0	NR	7	PL	T	D	TO	701 ± 181	Inferior peak load to failure with all reconstructions compared with intact ligaments (P < .001). No difference in strength between docking and single- strand reconstruction with CB ulnar fixation (P > .05). Docking and single-strand CB ulnar fixation are stronger than IS or figure-of-8 technique (P < .004). An optimal fixation method for single-strand reconstruction may require an improved IS or a modified CB procedure.
							7	PL	T	TO	TO	703 ± 256
							7	PL	L	IS	IS	454 ± 334
							7	PL	L	IS	CB	333 ± 133
	Gapping to failure	mm	7	2.4 ± 0.9	0	NR	7	PL	T	D	TO	4.21 ± 0.59
							7	PL	T	TO	TO	4.1 ± 0.65
							7	PL	L	IS	IS	3.6 ± 1.26
							7	PL	L	IS	CB	4.48 ± 0.77
	Peak load	N	7	142.5 ± 39.4	0	NR	7	PL	T	D	TO	53 ± 9.5
							7	PL	T	TO	TO	33.3 ± 7.1
							7	PL	L	IS	IS	41 ± 16
							7	PL	L	IS	CB	52.5 ± 10.4
Bernholt et al 5 (2019)	Gapping at failure	mm	12	1.4 ± 0.8	0	NR	12	PL	T	D	TO	1.9 ± 0.8	Reconstruction with an internal brace yielded comparable stiffness and ultimate failure torque to those of the native ligament, superior to standard docking.
	Stiffness	N·m/deg	12	4 ± 0.8	0	NR	12	PL	T	D	TO	3.0 ± 0.4
	Ultimate tensile strength	N·m	12	36.9 ± 10.1	0	NR	12	PL	T	D	TO	18.3 ± 4.1
Bodendorfer et al 6 (2018)	Gapping at failure	mm	9	50.94 ± 15.29	0	NR	9	PL	T	D	TO	47.87 ± 15.16	No significant differences in load to failure, gapping, or valgus opening angle between the native ligament, reconstruction, or repair.
			9	36.93 ± 16.65	9	41.92 ± 11.5	0	NA	NA	NA	NA	NA
	Ultimate tensile strength	N·m	9	28.98 ± 10.02	0	NR	9	PL	T	D	TO	23.14 ± 9.05
			9	29.54 ± 9.3	9	23.28 ± 10.06	0	NA	NA	NA	NA	NA
	Valgus opening angle at 90°	deg	9	26.62 ± 7.05	0	NR	9	PL	T	D	TO	25.21 ± 6.92
			9	19.87 ± 7.98	9	22.55 ± 5.55	0	NA	NA	NA	NA	NA
Chronister et al 8 (2014)	Cycles to failure	N	6	NR	0	NR	6	PL	L	D	IS	575 ± 409	No significant difference between constructs in valgus laxity or graft fixation displacement at failure.
			6	NR	0	NR	6	PL	L	TO	IS	802 ± 584
	Peak load	N	6	NR	0	NR	6	PL	L	D	IS	47
			6	NR	0	NR	6	PL	L	TO	IS	58
	Stiffness	N/mm	6	NR	0	NR	6	PL	L	D	IS	6.7 ± 3.7
			6	NR	0	NR	6	PL	L	TO	IS	6.7 ± 3.6
Ciccotti et al 9 (2009)	Ultimate tensile strength	N·m	20	14.9 ± 6.2	0	NR	10	PL	T	TO	TO	8 ± 4.4	Both methods provide comparable valgus stability to that of the native ligament at flexion angles ≥ 90°.
							10	PL	T	D	TO	8.2 ± 4.5
Cohen et al 10 (2015)	Ultimate tensile strength	N·m	20	20 ± 10.7	0	NR	10	PL	T	D	TO	4.35 ± 3.33	No significant difference in load to failure between elbows reconstructed at 30° vs 90° of flexion.
							10	PL	T	D	TO	4.85 ± 1.52
Dargel et al 13 (2015)	Valgus opening angle at 0°	deg	10	7.3 ± 1.1	0	NR	10	PL	T	D	TO	11.1 ± 2.4	No significant difference in valgus deformation or valgus stability was found when testing graft thickness over the full flexion range of motion.
	Valgus opening angle at 30°	deg	10	NR	0	NR	10	PL	T	D	TO	11.2 ± 2.5
	Valgus opening angle at 60°	deg	10	NR	0	NR	10	PL	T	D	TO	9.9 ± 1.9
	Valgus opening angle at 90°	deg	10	5.9 ± 1.2	0	NR	10	PL	T	D	TO	7.5 ± 1.6
	Valgus opening angle at 120°	deg	10	NR	0	NR	10	PL	T	D	TO	6.9 ± 2.1
Dugas et al 15 (2016)	Gapping at 10 N⋅m	mm	0	NR	9	2.02 ± 1.16	9	PL	T	TO	TO	2.86 ± 2.14	Repair has comparable stiffness and maximum torque at failure to traditional reconstruction techniques. Repair showed greater resistance to gapping than reconstruction.
	Stiffness	N·m/deg	0	NR	9	1.32 ± 0.39	9	PL	T	TO	TO	1.28 ± 0.49
	Ultimate tensile strength	N·m	0	NR	9	23.6 ± 10.8	9	PL	T	TO	TO	20.9 ± 6.58
Duggan et al 16 (2011)	Articular contact pressure at 30° (1.75 N·m torque)	MPa	6	0.37 ± 0.31	0	NR	6	HS	T	D	TO	0.46 ± 0.47	UCL injury increases articular cartilage contact surface pressures and valgus laxity. These values can improve with UCL reconstruction.
	Articular contact pressure at 30° (5.25 N·m torque)	MPa	6	0.47 ± 0.37	0	NR	6	HS	T	D	TO	0.72 ± 0.52
	Articular contact pressure at 90° (1.75 N·m torque)	MPa	6	0.32 ± 0.14	0	NR	6	HS	T	D	TO	0.37 ± 0.19
	Articular contact pressure at 90° (5.25 N·m torque)	MPa	6	0.58 ± 0.28	0	NR	6	HS	T	D	TO	0.69 ± 0.42
	RC contact area at 30° (1.75 N⋅m torque)	mm2	6	117.8 ± 84.1	0	NR	6	HS	T	D	TO	117.3 ± 85.2
	RC contact area at 30° (5.25 N⋅m torque)	mm2	6	157.5 ± 69.7	0	NR	6	HS	T	D	TO	129.5 ± 24.9
	RC contact area at 90° (1.75 N⋅m torque)	mm2	6	128.5 ± 44.1	0	NR	6	HS	T	D	TO	118.3 ± 46.7
	RC contact area at 90° (5.25 N⋅m torque)	mm2	6	120 ± 31.8	0	NR	6	HS	T	D	TO	120.7 ± 21.3
	Valgus opening angle at 30° (1.75 N⋅m torque)	deg	6	1.11 ± 1.09	0	NR	6	HS	T	D	TO	0.49 ± 1.18
	Valgus opening angle at 30° (5.25 N⋅m torque)	°	6	3.62 ± 2.76	0	NR	6	HS	T	D	TO	3.71 ± 1.81
	Valgus opening angle at 90° (1.75 N⋅m torque)	deg	6	1.35 ± 1.15	0	NR	6	HS	T	D	TO	0.33 ± 0.26
	Valgus opening angle at 90° (5.25 N⋅m torque)	deg	6	3.67 ± 1.57	0	NR	6	HS	T	D	TO	2.43 ± 0.99
Hechtman et al 20 (1998)	Ultimate tensile strength	N·m	17	22.7 ± 9	0	NR	17	PL	L	SA	SA	13.6 ± 5.5	The SA method more accurately replicated the native UCL’s normal anatomy and function. Both constructs are weaker than native UCL.
			14	22.7 ± 9	0	NR	14	PL	T	TO	TO	15.4 ± 7.1
Hurbanek et al 22 (2009)	Valgus opening angle at 30°	deg	9	7 ± 4	0	NR	9	PL	T	D	TO	12 ± 4	The addition of an interference screw to the docking method significantly increased the reconstruction’s stiffness and decreased elbow laxity. However, there was no difference in the moment of failure between the 2 reconstructions, and the biomechanical performance was largely similar.
			9	9 ± 6	0	NR	9	PL	T	IS	TO	10 ± 3
	Valgus opening angle at 60°	deg	9	7 ± 2	0	NR	9	PL	T	D	TO	10 ± 3
			9	8 ± 5	0	NR	9	PL	T	IS	TO	8 ± 2
	Valgus opening angle at 90°	deg	9	7 ± 2	0	NR	9	PL	T	D	TO	10 ± 3
			9	7 ± 4	0	NR	9	PL	T	IS	TO	6 ± 2
	Ultimate tensile strength	N·m	0	NR	0	NR	9	PL	T	D	TO	12.2 ± 4.7
			0	NR	0	NR	9	PL	T	IS	TO	14.4 ± 4
Itami et al 23 (2021)	Stiffness	N·m/deg	0	NR	7	4.2 ± 0.4	7	PL	T	D	TO	3.4 ± 0.4	Repair restored stiffness and valgus stability.
	Ultimate tensile strength	N·m	0	NR	7	20.8 ± 1.6	7	PL	T	D	TO	19 ± 1.4
	Valgus opening angle at 90°	deg	0	NR	7	9 ± 1.4	7	PL	T	D	TO	12.1 ± 1.3
	Yield torque	N·m	0	NR	7	15.6 ± 1.8	7	PL	T	D	TO	16.2 ± 1.4
Jackson et al 24 (2013)	Stiffness	N·m/deg	0	NR	0	NR	6	EDL	T	D	TO	2.1 ± 0.8	Both techniques restored valgus laxity to the intact state. No significant difference in stiffness, ultimate torque, and yield torques between the techniques.
			0	NR	0	NR	6	EDL	L	CB	CB	2.3 ± 0.8
	Ultimate tensile strength	N·m	0	NR	0	NR	6	EDL	T	D	TO	25.1 ± 7.1
			0	NR	0	NR	6	EDL	L	CB	CB	26.5 ± 9.2
	Valgus opening angle at 90°	deg	0	NR	0	NR	6	EDL	T	D	TO	9.6 ± 1.9
			0	NR	0	NR	6	EDL	L	CB	CB	10.1 ± 5.2
	Yield torque	N·m	0	NR	0	NR	6	EDL	T	D	TO	18.6 ± 4.4
			0	NR	0	NR	6	EDL	L	CB	CB	18.7 ± 7.8
Jones et al 27 (2018)	Gapping at 10 N⋅m	mm	0	NR	10	3.54 ± 2.48	10	PL	T	TO	TO	6.8 ± 4.11	Repairs with internal bracing had significantly less gap formation than the reconstructed UCLs.
Large et al 28 (2007)	Ultimate tensile strength	N·m	0	NR	0	NR	10	HS	T	TO	TO	22.7 ± 8.7	The Jobe technique was superior to the IS technique in terms of stiffness and failure.
			0	NR	0	NR	10	HS	L	IS	IS	13.4 ± 8.7
McAdams et al 31 (2007)	Valgus opening angle at 70° (cycle 10)	deg	16	4.4	0	NR	8	PL	T	D	TO	11	The docking technique produced significantly greater valgus angles than the native ligament and the IS technique after early cyclic loads.
					0	NR	8	PL	L	IS	IS	7.4
	Valgus opening angle at 70° (cycle 100)	deg	16	4.9	0	NR	8	PL	T	D	TO	17.5
					0	NR	8	PL	L	IS	IS	8.8
	Valgus opening angle at 70° (cycle 100)	deg	16	5.7	0	NR	8	PL	T	D	TO	19.3
					0	NR	8	PL	L	IS	IS	16.8
McGraw et al 32 (2013)	Stiffness	N/mm	20	21 ± 4.5	0	NR	10	PL	T	D	TO	5.3 ± 1	The docking plus technique produced significantly greater stiffness and ultimate load failure than the docking technique.
					0	NR	10	PL	T	D+TO	TO	11.2 ± 3
	Ultimate tensile strength	N·m	20	35 ± 7.3	0	NR	10	PL	T	D	TO	8.6 ± 4.3
					0	NR	10	PL	T	D+TO	TO	20.6 ± 3.7
Morgan et al 33 (2010)	Stiffness	N·m/deg	0	NR	0	NR	8	HS	L	IS	CB	0.822 ± 0.101	Both reconstruction techniques restored valgus stability.
			0	NR	0	NR	8	HS	T	TO	TO	0.874 ± 0.133
	Valgus opening angle at 70°	deg	0	NR	0	NR	8	HS	L	IS	CB	9.83 ± 2.59
			0	NR	0	NR	8	HS	T	TO	TO	10 ± 3.93
Paletta et al 35 (1998)	Stiffness	N·m (mm/mm)	10	301.4 ± 126.7	0	NR	11	PL	T	TO	TO	74.3 ± 38.1	Neither technique replicated native biomechanics, but the docking technique had significantly higher load to failure than the Jobe technique.
					0	NR	12	PL	T	D	TO	80.9 ± 50.2
	Strain at maximal moment	mm/mm	10	0.087 ± 0.042	0	NR	11	PL	T	TO	TO	0.235 ± 0.103
					0	NR	12	PL	T	D	TO	0.295 ± 0.103
	Ultimate tensile strength	N·m	10	18.9 ± 9.1	0	NR	11	PL	T	TO	TO	8.9 ± 3.8
					0	NR	12	PL	T	D	TO	14.3 ± 4.1
Ruland et al 36 (2008)	Stiffness	N·m/deg	10	2.0 (1.8-2.2)	0	NR	10	PL	T	D	TO	0.6 (0.4–0.8)	Reconstructions exhibited inferior biomechanical properties relative to intact ligaments.
	Ultimate tensile strength	N·m	10	23.6 (20.7-26.3)	0	NR	10	PL	T	D	TO	12.8 (10.0–15.6)
Seiber et al 39 (2010)	Ultimate tensile strength	N·m	18	19.1 ± 9.4	0	NR	9	PL	L	IS	IS	15.5 ± 4	Reconstruction produced similar strength compared with the native ligament and improved stability compared with the torn ligament.
					0	NR	9	PL	T	TO	TO	15.5 ± 4
	Yield torque	N·m	18	18 ± 9.3	0	NR	9	PL	L	IS	IS	14.7 ± 4.4
					0	NR	9	PL	T	TO	TO	14.7 ± 4.4
Shah et al 40 (2009)	Valgus opening angle at 70° (cycle 10)	deg	8	3.9 ± 2.4	0	NR	8	PL	L	D	IS	9.5 ± 4.8	Reconstruction with humeral and ulnar IS fixation produced similar valgus angles to intact ligaments and smaller angles than other reconstruction techniques.
			8	4.3 ± 2.4	0	NR	8	PL	T	D	TO	7.9 ± 2.8
			8	2.9 ± 0.9	0	NR	8	PL	T	TO	TO	11.4 ± 8.6
			8	3.8 ± 1.2	0	NR	8	PL	L	IS	IS	5.4 ± 1.5
	Valgus opening angle at 70° (cycle 100)	deg	8	4.8 ± 1.8	0	NR	8	PL	L	D	IS	12.8 ± 6.4
			8	5.2 ± 2.8	0	NR	8	PL	T	D	TO	11.1 ± 5
			8	3.5 ± 1	0	NR	8	PL	T	TO	TO	19.1 ± 11.4
			8	4.5 ± 1.5	0	NR	8	PL	L	IS	IS	6.9 ± 1.9
	Valgus opening angle at 70° (cycle 100)	deg	8	6 ± 2.3	0	NR	8	PL	L	D	IS	17.1 ± 9.1
			8	6.3 ± 3.3	0	NR	8	PL	T	D	TO	14.8 ± 6.2
			8	4.4 ± 1.5	0	NR	8	PL	T	TO	TO	27.8 ± 9.4
			8	5.4 ± 2	0	NR	8	PL	L	IS	IS	8.8 ± 2.7
Urch et al 45 (2019)	Stiffness	N·m/deg	0	NR	8	1.6 ± 0.2	8	PL	T	D	TO	2 ± 0.2	Repair exhibited comparable valgus laxity and rotation of the native ligament, but reconstruction had superior load-to- failure.
	Ultimate tensile strength	N·m	0	NR	8	17.6 ± 1.7	8	PL	T	D	TO	23.9 ± 2.2
	Yield angle	deg	0	NR	8	5.4 ± 0.6	8	PL	T	D	TO	10.2 ± 1
	Yield torque	N·m	0	NR	8	9 ± 1.4	8	PL	T	D	TO	19.1 ± 1.4

^aG, graft source (EDL, extensor digitorum longus; HS, hamstring; PL, palmaris longus); C, configuration (L, linear; T, triangular); H, humeral fixation (CB, cortical button; D, docked; IS, interference screw; SA, suture anchor; TO, transosseous); NR, not reported; U, ulnar fixation (CB, cortical button; IS, interference screw; SA, suture anchor; TO, transosseous); UCLR, ulnar collateral ligament reconstruction.

Table 4.

Biomechanical Construct Failure ^a

		Technical Variations
Study	Arm	G	C	H	U	Construct Failures
Ahmad et al 2 (2003)	Native					Intrasubstance rupture, n = 8 Avulsion from ulnar insertion, n = 2
	Reconstruction	PL	L	IS	IS	Graft rupture, n = 6 Ulnar fracture, n = 2 Graft slippage, n = 1 Screw and graft pullout, n = 1
Armstrong et al 3 (2005)	Reconstruction	PL	T	D	TO	Suture pullout of graft-suture interface, n = not listed
	Reconstruction	PL	T	TO	TO	Suture pullout of graft-suture interface, n = not listed
	Reconstruction	PL	L	PL	CB	Suture pullout of graft-suture interface, n = not listed
	Reconstruction	PL	L	IS	IS	Tendon pullout at tendon-screw interface, n = not listed
Bernholt et al 5 (2019)	Reconstruction	PL	T	D	TO	Graft, n = 9 Knot failure, n = 2 Ulnar tunnel fracture, n = 1
Bodendorfer et al 6 (2018)	Reconstruction	PL	T	D	TO	Ulnar tunnel fracture, n = 2 Midsubstance graft rupture, n = 1 Failure at proximal suture-tendon interface, n = 6
	Repair					Pullout of ulnar anchor, n = 4 Pullout of humeral anchor, n = 1 Suture pullout, n = 2 Ulnar fracture, n = 2
Chronister et al 8 (2014)	Reconstruction	PL	L	D	IS	Suture pullout of graft-suture interface, n = 5 Knot failure over the humeral bone bridge, n = 1
	Reconstruction	PL	L	TO	IS	Intrasubstance graft stretching, n=5 Partial graft-screw slippage, n = 1
Ciccotti et al 9 (2009)	Reconstruction	PL	T	TO	TO	Ulnar avulsion, n = 5 Midsubstance rupture, n = 5
	Reconstruction	PL	T	D	TO	Midsubstance rupture, n = 1 Suture pullout, n = 4 Humeral tunnel fracture, n = 1 Ulnar tunnel fracture, n = 4
Cohen et al 10 (2015)	Reconstruction, 30° of flexion	PL	T	D	TO	Suture pullout at humeral side, n = 2; Tunneling, n = 2 b Bone fracture at the tunnel, n = 4 Graft rupture, n = 2
	Reconstruction, 90° flexion	PL	T	D	TO	Suture pullout at humeral side, n=3 Tunneling, n = 3 Graft rupture, n = 5
Dugas et al 15 (2016)	Reconstruction	PL	T	TO	TO	Ulnar tunnel fracture, n = 3 Humeral shaft/supracondylar fracture, n = 3 Intrasubstance tear, n = 3
	Repair					Ulnar screw pullout, n = 4 Epicondyle screw pullout, n = 2 Humeral shaft/supracondylar fracture, n = 3
Hechtman et al 20 (1998)	Reconstruction	PL	L	SA	SA	Sutures slipping, n = 9 Suture failure, n = 3 Ulnar bone fracture, n = 2 Anchor pullout, n = 2 Intraligament failure, n = 1
	Reconstruction	PL	T	TO	TO	Sutures slipping, n = 9 Humeral fracture, n = 2 Ulnar bone fracture, n = 2 Intraligament failure, n = 1
Hurbanek et al 22 (2009)	Reconstruction	PL	T	DO	TO	Suture pullout of graft-suture interface, n = 8 Ulnar tunnel fracture, n = 1
	Reconstruction	PL	T	IS	TO	Suture pullout of graft-suture interface, n = 7 Ulnar tunnel fracture, n = 1 Midsubstance graft rupture, n = 1
Jackson et al 24 (2013)	Reconstruction	EDL	T	D	TO	Suture pullout of graft-suture interface, n = 5 Ulnar tunnel fracture, n = 1
	Reconstruction	EDL	L	CB	CB	Sutures pull through graft, n = 6
Jones et al 27 (2018)	Reconstruction	PL	T	TO	TO	Ulnar tunnel fracture, n = 7 Humeral tunnel fracture, n = 1 Supracondylar distal humerus fracture, n = 1 Graft failure at the humeral tunnel, n = 1
	Repair					Ulnar screw pullout, n = 6 Ulnar bone tunnel failure, n = 3 Humerus fracture, n = 1
Large et al 28 (2007)	Reconstruction	HS	T	TO	TO	Epicondylar fracture, n = 3 Ulnar tunnel fracture, n = 1 Graft stretching, n = 6
	Reconstruction	HS	L	IS	IS	Pullout from ulnar tunnel, n = 3 Pullout from humeral tunnel, n = 4 Humeral fracture, n = 1 Graft stretching, n = 3
McGraw et al 32 (2013)	Reconstruction	PL	T	D	TO	Suture pullout, n = 4 Midsubstance graft rupture, n = 4 Suture rupture, n = 1 Humeral tunnel fracture, n = 1
Paletta et al 35 (1998)	Reconstruction	PL	T	TO	TO	Suture failure, n = 1 By suture-tendon interface failure, n = 12 By ulnar tunnel fracture, n = 2
	Reconstruction	PL	T	D	TO	Suture failure, n = 12 Bone tunnel fracture, n = 2
Ruland et al 36 (2008)	Reconstruction	PL	T	D	DO	Suture pullout, n = 8 Knot failure, n = 1 Knot pulled through the bridge, n=1

^aG, graft source (EDL, extensor digitorum longus; HS, hamstring; PL, palmaris longus); C, configuration (L, linear; T, triangular); H, humeral fixation (CB, cortical button; D, docked; IS, interference screw; SA, suture anchor; TO, transosseous); U, ulnar fixation (CB, cortical button; IS, interference screw; SA, suture anchor; TO, transosseous).

^bTunneling noted when both sides of reconstruction became loose.

Results of Syntheses

Comparison With Intact Ligaments

Compared with native intact ligaments (Table 5), reconstructions demonstrated significantly lower ultimate strength and stiffness (N/mm) (Figure 2), and significantly higher valgus opening angle at 70° of flexion (Figure 3). Stiffness (N·m/deg), valgus opening angle at 30° of flexion, valgus opening angle at 90° of flexion, and gapping at failure were not significantly different between intact ligaments and repairs.

Table 5.

Biomechanical Comparison of UCLR Versus Native Intact UCL ^a

					Heterogeneity
Outcome	SMD	SE	95% CI	P	PQ	τ2	I 2, %
UTS	−1.411	0.316	−2.031 to −0.791	<.0001	.0006	0.767	77.25
Stiffness, N/mm	−3.259	1.472	−6.143 to −0.374	.0268	.0043	3.812	87.72
Stiffness, N·m/deg	−6.529	5.115	−16.553 to 3.496	.2018	<.0001	50.421	96.27
VOA at 30°	0.269	0.478	−0.669 to 1.206	.5743	.0596	0.361	56.20
VOA at 70°	1.638	0.000	1.637 to 1.638	<.0001	<.0001	0.248	47.90
VOA at 90°	0.045	0.460	−0.857 to 0.946	.9229	.0019	0.743	73.63
GAF	0.8294	0.631	−0.408 to 2.066	.1888	.0052	1.037	78.64

^aNative ligament as the reference group.ES, effect size; GAF, gapping at failure; SMD, standardized mean difference; VOA, valgus opening angle; UCL, ulnar collateral ligament; UCLR, UCL reconstruction; UTS, ultimate tensile strength.

Figure 2.

Forest plot demonstrating summary effect analysis for ultimate strength and stiffness comparing UCLR with intact ligaments. Config, configuration; RE, random effects; SMD, standardized mean difference; UCLR, ulnar collateral ligament reconstruction.

Figure 3.

Forest plot demonstrating summary effect analysis for gapping at failure and valgus opening angle comparing UCLR with intact ligaments. Config, configuration; RE, random effects; SMD, standardized mean difference; UCLR, ulnar collateral ligament reconstruction.

Only 1 study compared intact ligaments with repairs. ⁶ Those authors did not observe a significant difference in gapping, valgus opening angle at 30° of flexion, or ultimate strength between repaired and reconstructed specimens and their respective intact states.

UCL Reconstruction Technical Variations

Graft Choice

Effects of graft source (palmaris longus or hamstring) were analyzable for valgus opening angle at 30° of flexion and valgus opening angle at 90° of flexion (Table 6). Graft choice demonstrated a significant predictive effect on the valgus opening angle at 90° of flexion (P_QM = .0490), with a decrease in AIC and likelihood ratio testing, implying a better fit when accounting for graft choice (P = .0269).

Table 6.

Effect of Graft Source on Biomechanical Properties, Reconstructions Versus Intact UCL ^a

	Specific Moderator Effects						Heterogeneity			Model Fit
Outcome	Graft	SMD	SE	95% CI	P	PQM	PQE	τ2	I 2, %	ΔAIC b	LRT	P
VOA at 30°	PL	0.710	0.420	−0.113 to 1.533	.0908	.2113	.0654	0.118	29.81	−0.15	2.15	.1424
	HS	−0.251	0.502	−1.234 to 0.733	.6174
VOA at 90°	PL	0.434	0.316	−0.186 to 1.053	.1698	.0490	.0133	0.257	50.19	−2.90	4.90	.0269
	HS	−1.082	0.532	−2.124 – −0.041	.0417

^aAIC, Akaike Information Criterion; HS, hamstring; LRT, likelihood ratio test; PL, palmaris longus; SMD, standardized mean difference; UCL, ulnar collateral ligament; VOA, valgus opening angle.

^bDifference in AIC, full model (ie, with moderator) – reduced model (ie, no moderator).

Valgus opening angle at 30° of flexion was not significantly different from the intact ligaments for either graft type; however, valgus opening angle at 90° of flexion was significantly lower with hamstring grafts when compared with the corresponding native ligament (–1.082; SE, 0.532 [95% CI, –2.124 to −0.041]; P = .0417).

Linear hypothesis testing demonstrated no difference between palmaris longus and hamstring grafts for valgus opening angle at 30° of flexion, but a significantly greater valgus opening angle at 90° of flexion with palmaris longus grafts (1.516; SE, 0.619 [95% CI, 0.304- 2.729]; P = .0142) (Table 7).

Table 7.

Predicted Difference in Biomechanical Properties Between UCLR Graft Sources ^a

Outcome	LH	Est	SE	95% CI	P
VOA at 30° of flexion	PL – HS	0.961	0.654	−0.322 to 2.243	.1420
VOA at 90° of flexion	PL – HS	1.516	0.619	0.304 to 2.729	.0142

^aEst, estimate; HS, hamstring; LH, linear hypothesis; PL, palmaris longus; UCLR, ulnar collateral ligament reconstruction; VOA, valgus opening angle.

Graft Configuration

Effects of graft configuration (triangular or linear) were analyzed with respect to ultimate strength, valgus opening angle at 70° of flexion, and gapping at failure (Table 8). The triangular configuration was designed to replicate the native anatomy of the UCL on the ulna while involving less bone removal from the humerus. The linear configuration was designed to replicate the central fibers of the anterior bundle of the UCL, which were thought to be essential to stability, and which were neglected by the figure-8 and triangular configurations. ²⁵ Graft configuration was associated with significant predictive effects for ultimate strength (P_QM < .0001) and valgus opening angle at 70° of flexion (P_QM < .0001). However, the corresponding likelihood ratio tests were not consistent with better model fit (all P > .05).

Table 8.

Effect of Graft Configuration on Biomechanical Properties, Reconstructions vs Intact UCL ^a

	Specific Moderator Effects						Heterogeneity			Model Fit
Outcome	Config	SMD	SE	95% CI	P	PQM	PQE	τ2	I 2, %	ΔAIC b	LRT	P
UTS	Tri	−1.411	0.326	−2.048 to −0.773	<.0001	<.0001	.0007	.786	77.32	2	.0001	.9926
	Lin	−1.417	0.394	−2.190 to −0.644	.0003
VOA at 70°	Tri	1.931	0.179	1.581 to 2.282	<.0001	<.0001	.0001	.224	45.36	−1.75	3.75	.0528
	Lin	1.359	0.167	1.032 to 1.685	<.0001
GAF	Tri	0.897	0.709	−0.492 to 2.286	.2055	.4400	.0022	1.327	82.66	1.83	0.17	.6823
	Lin	0.632	0.868	1.070 to 2.334	.4667

^aAIC, Akaike Information Criterion; Config, configuration; GAF, gapping at failure; Lin, linear; LRT, likelihood ratio test; PL, palmaris longus; SMD, standardized mean difference; Tri, triangular; UCL, ulnar collateral ligament; VOA, valgus opening angle.

^bDifference in AIC, full model (ie, with moderator) – reduced model (ie, no moderator).

Both triangular (–1.411; SE, 0.326 [95% CI, –2.048 to −0.773]; P < .0001) and linear (–1.417; SE, 0.394 [95% CI, –2.190 to −0.644]; P = .0003) configurations had lower ultimate strength than the intact UCL. Valgus opening angle at 70° of flexion was significantly greater than the intact UCL with triangular configuration (1.931; SE, 0.179 [95% CI, 1.581 to 2.282]; P < .0001) and linear configuration (1.359; SE, 0.167 [95% CI, 1.032 to 1.685]; P < .0001). Gapping at failure was not significantly different from the intact UCL for either configuration (triangular, P = .2055; linear, P = .4667).

The difference between triangular and linear configurations was not significant for any analyzable effect size (Table 9).

Table 9.

Predicted Difference in Biomechanical Properties Between UCLR Graft Configurations ^a

Outcome	LH	Est	SE	95% CI	P
UTS	Tri – Lin	0.007	0.297	−0.576 to 0.590	.9817
VOA at 70° of flexion	Tri – Lin	0.573	0.345	−0.104 to 1.250	.0973
GAF	Tri – Lin	0.265	0.640	−0.989 to 1.519	.6786

^aEst, estimate; GAF, gapping at failure; LH, linear hypothesis; Lin, linear; Tri, triangular; UTS, ultimate tensile strength; UCLR, ulnar collateral ligament reconstruction; VOA, valgus opening.

Humeral Fixation

Effects of humeral fixation were analyzable for ultimate strength, valgus opening angle at 70° of flexion, and gapping at failure (Table 10). Humeral fixation was associated with significant predictive effects for ultimate strength and valgus opening angle at 70° of flexion (both P_QM < .0001), although fit parameters did not necessarily indicate better models.

Table 10.

Effect of Humeral Fixation on Biomechanical Properties, Reconstructions Versus Intact UCL ^a

	Specific Moderator Effects						Heterogeneity			Model Fit
Outcome	H Fix.	SMD	SE	95% CI	P	PQM	PQE	τ2	I 2, %	ΔAIC b	LRT	P
UTS	TO	−0.854	0.375	−1.588 to −0.119	.0227	<.0001	.0009	0.433	64.11	4.24	2.58	.4614
	D	−1.780	0.273	−2.315 to −1.245	<.0001
	IS	−0.297	0.540	−1.356 to 0.762	.5825
	SA	−1.249	0.629	−2.481 to −0.017	.0469
VOA at 70°	TO	2.205	0.380	1.462 to 2.949	<.0001	<.0001	.0005	0.140	34.11	0.82	3.18	.2041
	D	1.704	0.214	1.284 to 2.124	<.0001
	IS	1.215	0.255	0.715 to 1.715	<.0001
GAF	TO	1.452	1.196	−0.892 to 3.795	.2247	.4797	.0012	1.331	83.27	3.75	0.25	.8806
	D	0.881	0.726	−0.543 to 2.301	.2253
	IS	1.091	0.983	−0.836 to 3.018	.2671

^aAIC, Akaike Information Criterion; D, docking; GAF, gapping at failure; H Fix, humeral fixation; IS, interference screw; LRT, likelihood ratio test; SA, suture anchor; SMD, standardized mean difference; TO, transosseous; UCL, ulnar collateral ligament; VOA, valgus opening angle.

^bDifference in AIC, full model (ie, with moderator) – reduced model (ie, no moderator).

Transosseous (–0.854; SE, 0.375 [95% CI, –1.588 to −0.119]; P = .0227), docking (–1.780; SE, 0.273 [95% CI, –2.315 to −1.245]; P < .0001), and suture anchor fixation (–1.249; SE, 0.629 [95% CI, –2.481 to −0.017]; P = .0469) were each associated with significantly lower ultimate strength than the intact UCL, while interference screw fixation was not significantly different (P = .5825). Valgus opening angle at 70° of flexion was significantly greater than that of the intact UCL with transosseous fixation (2.205; SE, 0.380 [95% CI, 1.462 to 2.949]; P < .0001), docking fixation (1.704; SE, 0.214 [95% CI, 1.284 to 2.124]; P < .0001), and interference screw fixation (1.215; SE, 0.255 [95% CI, 0.715 to 1.715]; P < .0001). Gapping at failure was not significantly different from the intact UCL for transosseous, docking, or interference screw fixation (all P > .05).

When evaluating linear combinations of humeral fixation, ultimate strength was significantly lower with docking versus interference screw fixation (–1.483; SE, 0.597 [95% CI, – 2.653 to −0.313; P = .0129), and valgus opening angle at 70° of flexion was significantly greater with transosseous versus interference screw (0.990; SE, 0.458 [95% CI, 0.092 to 1.887]; P = .0306) (Table 11).

Table 11.

Predicted Difference in Biomechanical Properties According to UCLR Humeral Fixation ^a

Outcome	LH	Est	SE	95% CI	P
UTS	TO – D	0.926	0.504	−0.062 to 1.914	.0662
	TO – IS	−0.557	0.712	−1.953 to 0.839	.4343
	TO – SA	0.395	0.843	−1.256 to 2.047	.6390
	D – IS	−1.483	0.597	−2.653 to −0.313	.0129
	D – SA	−0.531	0.667	−1.839 to 0.777	.4264
	IS – SA	0.952	0.800	−0.616 to 2.521	.2341
VOA at 70° of flexion	TO – D	0.502	0.433	−0.347 to 1.350	.2465
	TO – IS	0.990	0.458	0.092 to 1.887	.0306
	D – IS	0.489	0.335	−0.168 to 1.145	.1449
GAF	TO – D	0.571	1.210	−1.801 to 2.943	.6371
	TO – IS	0.360	1.162	−1.917 to 2.638	.7565
	D – IS	−0.211	1.005	−2.181 to 1.759	.8339

^aD, docking; Est, estimate; GAF, gapping at failure; IS, interference screw; LH, linear hypothesis; SA, suture anchor; TO, transosseous; UTS, ultimate tensile strength; UCLR, ulnar collateral ligament reconstruction; VOA, valgus opening angle.

Ulnar Fixation

Effects of ulnar fixation were analyzable for ultimate strength and valgus opening angle at 70° of flexion (Table 12). Ulnar fixation was associated with significant predictive effects on both effect sizes (ultimate strength, P_QM < .0001; valgus opening angle at 70°, P_QM < .0001), but fit statistics were not consistent with better models.

Table 12.

Effect of Ulnar Fixation on Biomechanical Properties, Reconstructions vs Intact UCL ^a

	Specific Moderator Effects						Heterogeneity			Model Fit
Outcome	U Fix	SMD	SE	95% CI	P	PQM	PQE	τ2	I 2, %	ΔAIC b	LRT	P
UTS	TO	−1.446	0.305	−2.044 to −0.848	<.0001	<.0001	.0012	0.648	73.01	2.75	1.25	.5344
	IS	−1.035	0.493	−2.02 to −0.068	.0360
	SA	−1.684	0.520	−2.703 to −0.666	.0012
VOA @ 70°	TO	1.931	0.179	1.581 to 2.282	<.0001	<.0001	.0001	0.224	45.36	-1.75	3.75	.0528
	IS	1.359	0.167	1.032 to 1.685	<.0001

^aAIC, Akaike Information Criterion; CB, cortical button; GAF, gapping at failure; IS, interference screw; LRT, likelihood ratio test; SA, suture anchor; SMD, standardized mean difference; TO, transosseous; U Fix, ulnar fixation; UCL, ulnar collateral ligament; VOA, valgus opening angle.

^bDifference in AIC, full model (ie, with moderator) – reduced model (ie, no moderator).

Ultimate strength was significantly lower than the intact UCL with transosseous fixation (–1.446; SE, 0.305 [95%, –2.044 to −0.848]; P < .0001), interference screw fixation (–1.035; SE, 0.493 [95% CI, –2.002 to −0.068]; P = .0360), and suture anchor fixation (–1.684; SE, 0.520 [95% CI, –2.703 to −0.666]; P = .0012). Valgus opening angle at 70° of flexion was significantly greater for both transosseous fixation (1.931; SE, 0.179 [95% CI, 1.581 to 2.282]; P < .0001) and interference screw fixation (1.359; SE, 0.167 [95% CI, 1.032 to 1.685]; P < .0001).

Linear combinations revealed no significant differences in ultimate strength or valgus opening angle at 70° of flexion (Table 13).

Table 13.

Predicted Difference in Biomechanical Properties According to Ulnar Fixation ^a

Outcome	LH	Est	SE	95% CI	P
UTS	TO – IS	−0.412	0.469	−1.330 to 0.507	.3799
	TO – SA	0.238	0.453	−0.649 to 1.125	.5988
	IS – SA	0.650	0.644	−0.612 to 1.911	.3127
VOA at 70° of flexion	TO – IS	0.573	0.345	−0.104 to 1.250	.0973

^aCB, cortical button; CII, confidence interval; Est, estimate; GAF, gapping at failure; IS, interference screw; LH, linear hypothesis; SA, suture anchor; TO, transosseous; UTS, ultimate tensile strength; VOA, valgus opening angle.

Docking Versus Jobe Reconstruction

Two studies directly compared docking reconstruction (triangular graft configuration, docked humeral fixation, and transosseous ulnar fixation) with Jobe reconstruction (triangular graft configuration, transosseous humeral fixation, and transosseous ulnar fixation),^9,35 and only ultimate strength was analyzable. The difference in ultimate strength between docking and Jobe techniques was not significant (0.6740; SE, 0.6356 [95% CI, –0.5717 to 1.9198]; P = .2889; P_Q = .0476; τ2 = 0.602; I² = 74.51%) (Figure 4).

Figure 4.

Forest plot demonstrating summary effect analysis comparing ultimate strength between docking reconstruction (triangular graft configuration, docked humeral fixation, and transosseous ulnar fixation) and Jobe reconstruction (triangular graft configuration, transosseous humeral fixation, and transosseous ulnar fixation). Config, configuration; RE, random effects; SMD, standardized mean difference.

Reconstruction Versus Repair

Three effect sizes (ultimate strength, stiffness (N·m/deg), and yield torque) were suitable for quantitative comparisons of biomechanical performance between repaired and reconstructed specimens. No differences in performance were observed across these effect sizes (Table 14, Figure 5).

Table 14.

Biomechanical Comparison of UCL Repair Versus UCLR ^a

					Heterogeneity
Outcome	SMD	SE	95% CI	P	PQ	τ 2	PQ
UTS	−0.350	0.866	2.046 to −2.046	.6863	<.0001	2.673	89.94
Stiffness, N·m/deg	0.015	1.070	−2.081 to 2.111	.9890	<.0001	3.102	90.59
YT	−0.016	1.881	−3.703 to 3.671	.9933	<.0001	6.692	94.54

^aReconstruction as the reference group.SMD, standardized mean difference; UTS, ultimate tensile strength; UCLR, ulnar collateral ligament reconstruction; YT, yield torque.

Figure 5.

Forest plot demonstrating summary effect analysis comparing stiffness (N·m/deg), ultimate strength, and yield torque between UCL repair and reconstruction. Config, configuration; RE, random effects; SMD, standardized mean difference; UCL, ulnar collateral ligament.

Discussion

The major findings of our study demonstrated that the biomechanical performance of UCLR was not superior to that of the native UCL. UCLR demonstrated inferior ultimate strength (P≤ .0001), reduced stiffness (N/mm) (P = .03), and greater valgus opening at 70° of flexion (P≤ .0001) compared with the intact UCL. No significant differences were observed in stiffness (N·m/deg) (P = .20), valgus opening at 30° (P = .57) or 90° of flexion (P = .92), or gapping at failure (P = .19) between the 2 techniques. There was no evidence that any specific graft source, graft configuration, or ulnar fixation strategy provided improved patient outcomes. Additionally, no differences in biomechanical performance were identified between the docking and Jobe reconstruction techniques or between UCL repair and reconstruction.

This study reviews previous biomechanical investigations of UCLR and repairs, focusing on their most common configurations. We hypothesized that our review would reveal inferior, or at best biomechanically equivalent, performance of UCLR compared with the native, intact UCL.Our findings support this hypothesis, showing inferior biomechanical properties (ultimate tensile strength standardized mean difference [SMD], –1.411; P < .001; Stiffness SMD, –3.259; P = .0268; VOA and 70° SMD, 1.638; P < .0001) or no significant difference (P > .05) in UCLRs compared with native UCL tissue (Table 5).

We also hypothesized that there would be no difference in biomechanical performance between docking and Jobe reconstruction, or between reconstruction and repair. Our findings again supported this. Although valgus opening angles varied by graft type (1.151; P = .0142) (Table 7), ultimate strength did not differ between interference screw fixation and native tissue (–0.297; SE, 0.540 [95% CI, –1.356 to 0.762]; P = .5825). Similarly, no difference in gapping at failure was observed compared with intact UCL tissue for transosseous, docking, or interference screw fixation (P > .05 in all) (Table 10). However, these fixation methods exhibited significantly lower ultimate tensile strengths compared with native UCL tissue (transosseous P = .0227; P < .0001; suture anchor fixation P = .0469) (Table 10).

Furthermore, no significant biomechanical differences were found among graft configurations (P > .05 in all) (Table 9), ulnar fixation methods (P > .05 in all) (Table 13), or between docking or Jobe reconstructions (P = .02889). Likewise, reconstruction and repair did not differ biomechanically (P > .05 in all) (Table 14). These findings are particularly relevant given that the docking (triangular graft configuration, docked humeral fixation, transosseous ulnar fixation) and Jobe (triangular graft configuration, transosseous humeral fixation, and transosseous ulnar fixation) reconstructions represent the most commonly utilized techniques for UCLR, while repair is increasingly performed for select patients. ⁷

UCL Reconstruction Versus Native Intact UCL

In biomechanical analysis of UCLR and native UCL, ultimate tensile strength and ultimate strength and stiffness (N/mm) of UCLR constructs are significantly less (P < .0001 and P = .0268, respectively), and valgus opening angle at 70° of flexion is significantly greater than the intact UCL (P < .0001). Stiffness (N·⋅m/deg), valgus opening angle at 30° and 90° of flexion, and gapping at failure are not significantly different from the intact UCL (all P> .05). UCLR does not restore or exceed native ligament biomechanics. This supports previous work by Ciccotti et al, ⁹ which demonstrated that the modified Jobe and Docking techniques provide valgus laxity similar to that of intact ligaments at >90° of flexion.

UCLR Technical Variations

Graft Source

Two graft sources—palmaris longus and hamstring tendon—were evaluated quantitatively in the mixed effects models for valgus opening angle at 30° and 90° of flexion. Hamstring grafts had significantly less valgus opening than the native ligament at 90° flexion (P = .0417), and significantly less relative valgus opening at 90° flexion than palmaris grafts upon linear hypothesis testing (P = .0142). However, valgus opening angle alone may not be an ideal biomechanical metric compared with stiffness and ultimate strength, which were not available for analysis. Therefore, no clear advantage of one graft source can be concluded.

Graft Configuration

Valgus opening at 70° of flexion was significantly higher and ultimate strength was significantly lower than native ligaments for both linear and triangular configurations (P < .001 in all cases), and there was no difference between constructs upon linear hypothesis testing (P > .05 for all outcomes). Gapping at failure, a characteristic of the failure state rather than a biomechanical property, did not differ significantly from the native ligament with either configuration. These findings suggest no biomechanical superiority of one configuration over the other.

Humeral Fixation

Effects of humeral fixation (docking, transosseous, interference screw, or suture anchor) were evaluated for ultimate strength, valgus opening angle at 70° of flexion, and gapping at failure. Gapping did not differ between fixation methods (all P < .05). The only fixation method not associated with a different ultimate strength relative to the native ligament was interference screw fixation (P = .5825). Valgus opening at 70° of flexion was significantly greater than the intact UCL for all available fixation methods (all P < .0001). Upon linear hypothesis testing, docking fixation was associated with significantly lower ultimate strength than interference screw fixation (P = .0129), and transosseous fixation was associated with a significantly higher valgus opening at 70° of flexion than interference screw fixation (P = .0306). While these findings offer limited support for humeral interference screw fixation, previous authors have raised concerns about increased clinical risk of medial epicondyle fracture associated with humeral interference screw fixation, and these results should be cautiously interpreted. ⁷

Ulnar Fixation

Effects of ulnar fixation were evaluated for ultimate strength (transosseous, interference screw, or suture anchor) and valgus opening at 70° of flexion (transosseous or interference screw fixation). All fixation methods produced lower ultimate strength (all P < .04) and greater valgus opening (all P < .0001) than the native ligament, with no difference between fixation methods. These findings, therefore, do not suggest any biomechanical benefit of any one fixation strategy.

Docking Versus Jobe Reconstruction

We observed no significant difference in ultimate strength in comparisons between docking (triangular configuration, docked humeral fixation, and transosseous ulnar fixation) and Jobe techniques (triangular configuration, transosseous humeral fixation, and transosseous ulnar fixation) (P = .2889). Our findings are consistent with recent clinical studies^18,19 and meta-analyses,^29,30 which have shown no difference between these 2 most frequently performed UCLR techniques in terms of outcomes or complications.

Drawing direct comparisons between biomechanical studies and clinical significance remains challenging. However, analyzing clinically focused studies on UCLR in the context of the above findings allows clinical parallels to be drawn. For example, a previous review of 6 retrospective studies including 2153 UCLRs examined outcomes using palmaris longus and hamstring grafts. ⁴³ Authors found similar return-to-sport rates (84.59% vs 80.8%) and return-to-same-level-of-play rates (81.96% vs 80.76%) between those who underwent UCLR with palmaris longus and hamstring grafts, respectively. Graft choice did not change revision rates or patient-reported outcome scores (Timmerman-Andrew, Conway-Jobe). Notably, 2 of the 3 included studies that reported Kerlan-Jobe Orthopaedic Clinic scores also found no significant difference between groups.

Another study has reported comparable outcomes across UCLR techniques. ¹⁹ Among 556 baseball players who underwent UCLR (30.2% docking technique vs 51.2% Jobe technique vs 18.6% other), the overall return-to-play rate was 79.9%, and return-to-same-level of play was 71.2%. No significant differences were observed in return-to-play rate (80.1% vs 82.4%, docking vs Jobe groups respectively; P = .537), return-to-same-level of play rate (73.7% vs 73.1%, docking vs Jobe groups respectively; P = .914), subsequent elbow injury rate among those who returned to play (43.9% vs 47.9%, docking vs Jobe respectively; P = .440), subsequent elbow surgery rate (10.5% vs 14.8%, docking vs Jobe groups respectively; P = .203), or revision UCLR rate (2.9% vs 6.2%, docking vs Jobe groups; P = .128). Time to return to play was also similar between groups (mean 433.8 days vs 433.9 days, docking vs Jobe groups respectively, P = .995) and time to return to the same level of sport (505.2 days vs 525 days, docking vs Jobe groups respectively, P = .384).

These similar clinical outcomes, regardless of graft choice or fixation method, align with the biomechanical findings of the present study, which do not indicate a clear superiority of any single technique over another. More extensive research is needed before strong recommendations can be made regarding optimal UCLR methods.

Reconstruction Versus Repair

Reconstruction and repair were evaluated for stiffness (N·m/deg), ultimate strength, and yield torque, and no significant differences were observed (all P > .05). Early clinical results of UCL repair were poor, with less durable and more unpredictable outcomes than reconstruction.^7,12,26 However, more recent technical developments, such as the use of internal brace augmentation, have led to renewed interest and good outcomes in appropriately indicated patients. ¹⁴ All repair constructs described by the included studies in the present analysis involved modern techniques, including internal brace augmentation, and the current synthesis found no biomechanical differences between repair and reconstruction.

Authors previously conducted a review of 8 biomechanical and 9 clinical studies comparing UCL repair and reconstruction, with 5 finding no significant difference between augmented UCL repair and reconstruction, while 1 reported higher stiffness in the repair group. ⁴² Of the 7 studies examining gap formation, 5 found less gap formation in the repair group, while 2 reported no difference. In the same study, peak torque results were mixed, with 2 of 4 studies noting no significant difference, one reporting higher torque in the repair group, and another reporting lower. Similarly, failure-load results were varied, with 5 studies finding no significant difference and 2 reporting lower failure loads in the repair groups. These results align with the present study and further reinforce the absence of a definitive biomechanical difference between reconstruction and repair.

Authors also highlight clinical outcomes in UCL repair with suture augmentation, analyzing 201 suture-augmented UCL repairs across 4 studies. ⁴² The rate of return to the previous level of play was 93%, and the mean return time ranged from 3.8 to 7.4 months. Minor complications occurred in 11.9% of cases, including ulnar nerve paresthesia, medial elbow pain, or superficial wound complications, but only 3.4% of patients required a return to the operating room. Notably, only 1 patient (0.02%) experienced a UCL retear.

Comparisons with UCLR by Griffith et al ¹⁹ highlight differences in clinical outcomes. Specifically, for baseball pitchers, the mean return-to-sport time was 518.2 ± 202.6 days (17.3 ± 6.8 months), significantly longer than the UCL repair cohort of the previously referenced study. ⁴² Additionally, the rate of subsequent elbow surgery was 10.5% for the docking technique and 14.8% for the modified Jobe technique (P = .203). However, further research is needed before definitive conclusions can be drawn regarding the superiority of 1 technique over the other, as the cohorts between these studies were not controlled for.

Previous Works

An early systematic review by Watson et al ⁴⁷ included results of 7 biomechanical studies, but only a limited qualitative synthesis was undertaken. In a more recent systematic review, Saltzman et al ³⁷ summarized the results of 23 biomechanical studies involving 397 elbow specimens. They reported no significant differences across techniques for all analyzed measures, but noted that since each of the 23 included studies used a different method, a more extensive quantitative synthesis was not possible.

The present study builds on previous authors’ work in several key ways. First, despite including a similar number of systematically identified articles (n = 24) as Saltzman et al, ³⁷ our statistical methodology enabled quantitative synthesis of 19 studies. Relatedly, our approach allowed us to separately evaluate different technical components of UCLR (ie, graft source, graft configuration, humeral fixation, and ulnar fixation). Finally, our synthesis included biomechanical assessments of UCL repair in addition to reconstruction.

This study also discredits the misconception that UCLR (Tommy John) surgery can be performed by creating a new ligament biomechanically superior to native anatomy.^1,11,48 For instance, Ahmad et al ¹ found that 30% of coaches, 37% of parents, 51% of high school athletes, and 26% of collegiate athletes believed that UCLR should be performed to enhance performance in the absence of injury, and that 28% of players and 20% of coaches felt that UCLR could enhance performance beyond preinjury levels. ¹ Our synthesis of the results of 153 specimens from 10 biomechanical studies demonstrated that reconstruction was not as strong as the native intact ligament. Furthermore, results of meta-regression revealed this to be the case regardless of graft configuration or fixation method.

Limitations

Our study was not without limitations. First, relatively small sample sizes were observed across all included studies, including this study. Biomechanical studies have limited specimen and testing resources, underscoring the importance of studies such as the present work that synthesize results from the existing literature. The limited sample size also increases the risk of type 2 statistical error and constrains the generalizability of findings to clinical practice. While our statistical methods were more flexible than those of previous studies and provided the framework for quantitative syntheses, insufficient sample sizes across the included studies to test actual combinations of moderators without variable redundancy preclude determining a hypothetically biomechanically optimal construct. Second, utilization of SMD as an effect size has both strengths and limitations. SMD—as employed herein—represents Hedge’s g, which is commonly used and intuitive to people familiar with basic statistics; it is essentially Cohen’s d corrected for SMD positive bias.^21,46 Unfortunately, some may find it difficult to interpret SMD, particularly in a biomechanical context. Nevertheless, in the context of the present study, more readily interpreted effect sizes, such as mean difference, were not appropriate given the extent of heterogeneity. Third, biomechanical studies by design are most representative of time-zero characteristics, and a synthesis of these studies is accordingly most representative of the same. Fourth, as a synthesis of biomechanical studies, our work should not be construed to imply anything about actual clinical outcomes, as biomechanical differences (or the lack thereof) do not necessarily translate into clinical implications. Lastly, approach/exposure of the medial elbow (ie, muscle splitting vs muscle elevating) represents a fifth technical variation, which was not explored by the included studies and was therefore not a part of our synthesis, but could have biomechanical and/or clinical consequences.

Conclusion

Our study found that synthesis of the available biomechanical evidence demonstrates that UCLR does not result in superior biomechanical properties when compared with the native intact ligament. Biomechanical performance of UCLR was either inferior to (ultimate strength, stiffness [N/mm], and valgus opening at 70° of flexion) or not significantly different from (stiffness [N⋅m/deg], valgus opening at 30° and 90° of flexion, and gapping at failure) the intact UCL. There is no difference in biomechanical performance between docking and Jobe reconstructions, or between UCL repair and reconstruction.