Neo-Natal Castration Leads to Subtle Differences in Porcine Anterior Cruciate Ligament Morphology and Function in Adolescence

There are an ever-increasing number of pediatric athletes participating in middle and high school sports [1], accompanied by an increase in pediatric anterior cruciate ligament (ACL) injuries [2–4]. Furthermore, during adolescence there is a difference in the rate of ACL injuries between males and females, with females being 1.4 times more likely to tear their ACL overall [5]. In specific sports, like basketball [6–8] or soccer [9], females are over three times more likely to tear their ACL [10,11]. This has been attributed to differences in aspects of sexual maturity [10,11] or gendered expectations in training, sport selection, and coaching [12,13].

Both extrinsic and intrinsic factors may contribute to sex differences in injury rates. Extrinsic factors include social acceptance of strength training programs for women compared to men [14]. Intrinsic factors include anatomical differences [15–17] or the onset of spikes and cycles of sex hormones [18–21]. The female sex hormones estradiol and progesterone have gathered the most attention with some groups showing associations between peaks of estradiol and progesterone to knee joint laxity and time of ACL injury [13,18,20,22]. However, strong evidence for a direct correlation between estradiol levels and injury risk does not yet exist [20,21]. While a major focus has been on uncovering catabolic effects of female sex hormones to understand the increased risk of ACL injury, an understudied factor is the influence of testosterone, the primary male sex hormone, on developing male ACLs.

It is known that there are androgen receptors present within the human male ACL [23], suggesting that the ACL is an androgen-responsive tissue. In humans, males who tore their ACLs have higher levels of testosterone, indicating that testosterone may play some direct or indirect role in increasing ACL injury risk [24]. To study testosterone in animal models, surgical castration has been performed to drastically decrease total serum testosterone by over 95% [25,26]. For example, adult rats castrated at 12 weeks of age (postpuberty) had 25-fold lower testosterone levels at 5 weeks postcastration than age-matched controls. Although the ACLs in the castrated rat group had a similar size, they had 15% lower load-to-failure values and 18% lower ultimate stresses than intact males [26]. This study provides some evidence that testosterone may play a protective role in ligament function in adults, especially failure properties of the ACL [26]. Thus, it may be possible that both high and low levels of testosterone may impact the ACL, yet the evidence is still sparse, particularly during skeletal growth.

Large animal models provide a useful tool to study the ACL. Our group, along with others, has used a porcine model to understand structural and functional changes of the ACL throughout skeletal growth, including progression through adolescence [27–32]. We have shown similar increases in length and CSA of the ACL and increases of AM bundle contribution within the human and porcine ACL, making this an appropriate model to study changes during adolescence [27,33,34]. The size of the porcine ACL diverges in males and females around the point of sexual maturity, mainly within the two distinct bundles of the ACL, the anteromedial (AM) and posterolateral (PL) bundles [35]. The AM bundle in porcine males and females increase in size throughout adolescence, yet the PL bundle size plateaus near the onset of puberty. Functionally, there is also a shift in bundle function when resisting anterior loads, where the AM bundle becomes the primary stabilizer over the PL bundle in both males and females [35]. The time of these changes is sex-dependent, with the plateau in PL bundle size occurring earlier and functional changes occurring later in females. These sex specific changes, and their timing around adolescence, motivate the examination of the role of sex hormones as a potential cause of these changes. It is still unclear, though, if male sex hormones, such as testosterone, or female sex hormones, such as estradiol and progesterone, drive changes in males and females. Therefore, the objective of this study was to determine how neo-natal castration, which reduces serum testosterone concentration, influences the structure and function of the male porcine ACL during youth and early adolescence. We did not have an a priori hypothesis, due to the limited previous work looking at castration effects on male ACL development in large animal models. To study these potential effects, we imaged stifle (knee) joints in intact and castrated male pigs across growth and development to analyze geometric properties of the ACL and its bundles and tested them mechanically to determine ACL function.

Thirty-six male Yorkshire crossbreed pigs were raised to 3 months (juvenile), 4.5 months (early adolescent), or 6 months (adolescent). Six of each age group were left sexually intact (boars), while the other six were castrated (barrows) at 2 weeks old (n = 6 animals/age/treatment). A sample size of six animals was selected based on previous work where we were able to distinguish statistical differences between male and female pig ACL CSAs [35]. The boar data presented in this study have been re-analyzed from a previous study [35] to be compared with barrow data. All animals were bred at the North Carolina State University Swine Education Unit, and all experimental protocols were approved by the NC State University IACUC. Following euthanasia, hind limbs were removed, and the stifle (knee) joints were stored at −20 °C in saline-soaked gauze.

To determine geometric data, stifle joints were thawed and imaged using a 7.0-T Siemens Magnetom MRI scanner (Siemens Healthineers, Erlangen, Germany) using a 28-channel knee coil and a double-echo steady-state sequence (DESS, flip angle: 25 deg, TR: 17 ms, TE: 6 s, voxel size: 0.42 × 0.42 × 0.4 mm). Joint size data, including bicondylar width, tibial CSA, anterior-posterior (AP) tibial width, as well as ACL, AM, and PL bundle CSAs, orientations, lengths, and volumes were recorded after segmentation using a commercial software (simplewarescanip, synopsys, Mountain View, CA) and processing using matlab (MathWorks, Natick, MA) scripts. Geometric properties measured had a high repeatability (interclass correlation coefficient (ICC): 0.74–0.89) [35]. Briefly, The AM and PL bundles were masked separated as determined by their separation by the lateral meniscus insertion on the tibia and contrast on the scan and were exported as surface meshes. The ACL segmentations were created by the joining of the AM and PL bundles. The meshes were then aligned to the z-axis, transverse planes introduced at 1 mm increments, and the middle 50% of the ACL and bundles were averaged to calculate CSA. Length was determined by the distance between centroids of the femoral and tibial insertion sites. Morphological ACL characteristics, including CSA, length, and volume, were normalized to bicondylar width to account for overall joint size.

Stifle joints were thawed and trimmed. The femur and tibia were mounted in custom molds using a fiberglass reinforced epoxy (Everglass, Evercoat, Cincinnati, OH) to prepare for robotic biomechanical testing. Stifle joints were then mounted to a robotic testing system (KR300 R2500, KRC4, Kuka, Shelby Charter Township, MI) with a universal force-moment sensor (Omega160 IP65, ATI Industrial Automatic, Apex, NC, resolution of 0.25 N and 0.0125 Nm), following previously described methods [35]. All systems were controlled through the simvitro software knee module (Cleveland Clinic, Cleveland, OH).

During each test, the femur was fixed to the testing platform and the tibia was mounted to the robotic system. Anatomical points on each leg were recorded using a three-dimensional digitizer (G2X Micro-Scribe, Amherst, VA, ±0.23 mm accuracy) to establish an anatomic coordinate system. The joints were then flexed from 40 deg (approximately full extension in pigs) to 90 deg while minimizing joint forces to establish a passive path during the test. AP forces were then applied to the joints at 60° of flexion, while minimizing forces in other directions and recording kinematic data. Maximum loads were scaled across age (40 N, 80 N, and 100 N for 3-month, 4.5-month, and 6-month, respectively) based on the femoral cortical CSA. The recorded kinematics were then repeated, and the forces were measured. Joint tissues were then removed sequentially (capsule, AM bundle, then PL bundle), and the intact kinematics were repeated to determine the force contribution of each tissue through the principle of superposition [36,37]. After the AM bundle was transected, the same loads were applied to the joint while new kinematics were recorded to ensure that the remaining PL bundle was under tension for in situ measurements.

Force-translation curves were then plotted across AP loading. The engagement point of the ACL was determined by the point of inflection in the force-translation curves, which separated the toe region and the linear region in anterior loading [38,39]. The toe region was fit using an exponential curve, and the higher loads above the inflection point were fit to a line. Joint neutral zone (also sometimes named “slack”) and stiffness were then calculated using Matlab scripts, as previously described [29,38,40] (Fig. S1 available in the Supplemental Materials on the ASME Digital Collection). The in situ stiffness values for the ACL and AM bundle were calculated under the intact kinematics, where the PL bundle in situ stiffness was calculated under the AM deficient state to ensure PL engagement. AM and PL contribution percentages to resisting anterior loading were calculated by dividing the anterior force for each bundle by the total ACL contribution to anterior loading and multiplying by 100. To compare boar and barrow bundle contributions across age, the difference in AM and PL percent contribution was plotted over time, with negative values indicating greater PL bundle forces and positive values indicating greater AM bundle forces.

Overall effects of age and surgical castration on bicondylar width, CSA, length, volume, in situ stiffness, neutral zone, and difference in bundle percent contribution were assessed using two-way ANOVAs for each variable. All data were normally distributed. No animals were excluded from this work. Multiple comparisons were made between ages and boars and barrows at each age. The type I error rate was controlled using Holm-Sidak's posthoc method. The differences in AM and PL bundle percent contribution were analyzed using one sample t-tests comparing the means to zero. Overall significance was set to α = 0.05 for all tests.

From MRI scans, joint bicondylar width, tibial plateau CSA, and AP tibial width were recorded as metrics of joint size between the boars and barrows (Fig. 1 and Table S1 available in the Supplemental Materials on the ASME Digital Collection). Across all three measurements of joint size, age had statistically significant effects on overall knee joint size (p < 0.0001 for all measures). Specifically, bicondylar width increased by 20% and 21% from 3-months to 6-months for the boars and the barrows, respectively (Figs. 1(a) and 1(d)). Tibial plateau CSA increased by 64% and 63% from 3-months to 6-months for boars and barrows, respectively (Figs. 1(b) and 1(e)). Similar changes were seen for AP tibial width, which increased by 25% and 32% for boars and barrows between the 3-month and 6-month age groups (Figs. 1(c) and 1(f)). Surgical castration had significant effects on joint size (bicondylar width: p = 0.001, tibial plateau CSA: p = 0.0007, AP tibial width: p < 0.0001). Barrow bicondylar width was 8% and 7% smaller at 3 months (adj p = 0.03) and 6 months (adj p = 0.02) but was not statistically significantly smaller at 4.5 months (adj p = 0.40). There were statistically significant (adj p = 0.003) differences at 6 months in tibial plateau CSA between boars and barrows with the barrow plateau CSA being 13% smaller on average. All barrow AP tibial widths at each age group were significantly smaller than the boars (3-month: 10% smaller, adj p = 0.002; 4.5-month: 5% smaller, adj p = 0.05; 6-month: 6% smaller, adj p = 0.02).

In addition to overall joint size, the morphology of the ACL and its corresponding AM and PL bundles were recorded. CSAs, lengths, and volumes for the ACL and the bundles were segmented from the MRI scans. For both boars and barrows, there were significant age effects seen for ACL, AM, and PL CSAs (p < 0.0001 for all) (Fig. 2(a) and Table S2 available in the Supplemental Materials). When comparing boars and barrows, there was no statistically significant effect of castration on ACL, AM bundle, or PL bundle CSA (p = 0.12, p = 0.20, p = 0.66). All CSA values were normalized to the bicondylar width of the joint to account for overall joint size (Fig. 2(b)). After normalizing, the effect of age persisted for the ACL, AM, and PL CSAs (p < 0.0001 for all), but the lack of substantial effect due to castration stayed the same (p = 0.81, p = 0.70, p = 0.27). ACL and bundle orientation were also recorded in the sagittal plane, and the orientation of the ACL and the bundles increased across age but were not statistically significant between boars and barrows (Fig. S2 available in the Supplemental Materials).

Age had similar effects on ACL and bundle lengths, steadily increasing from 3 months to 6 months (p < 0.0001 for all) (Fig. 3(a) and Table S3 available in the Supplemental Materials). There was also an effect of castration on the ACL length (p = 0.0018) and PL bundle length (p = 0.0002), but not the AM bundle length. The boars had on average 8% longer ACLs across all age groups tested. The boars also had longer PL bundles at 3 months (17% greater, adj p = 0.005). After normalizing to joint size, the statistically significant difference in ACL lengths across ages and between boars and barrows disappeared (Fig. 3(b)). The age effect also disappeared for the AM bundle and PL bundle lengths. However, when comparing normalized boar and barrow ACL and bundle lengths, there was still a statistically significant difference in the PL bundle length (p = 0.012), with barrows having a statistically significant shorter PL bundle at 4.5 months (11% shorter, adj p = 0.02).

While ACL and bundle CSA and length showed some minor differences between boars and barrows across the ages tested, the ACL and bundle volumes showed more substantial changes between boars and barrows. A similar age effect was observed in ACL, AM bundle, and PL bundle volumes (p < 0.0001 for all), which all increased during growth (Fig. 4(a), Table S4 available in the Supplemental Materials on the ASME Digital Collection). Castration also had a statistically significant impact on ACL (p = 0.002), AM bundle (p = 0.003), and PL bundle (p = 0.04) volume. Comparing boars and barrows across ages, the boars had on average 19% larger ACLs, 25% larger AM bundles, and 15% larger PL bundles than the barrows. The age effect was still seen after normalizing to the bicondylar width of the joints, but the effects of castration were no longer statistically significant for the ACL (p = 0.06) and PL bundle (p = 0.11) but persisted in the AM bundle (p = 0.04) (Fig. 4(b)).

Using a 6-degree-of-freedom robotic testing system, force-translation responses showed similar trends in both boars and barrows, where the slopes increased with age, indicating an increase in overall joint stiffness (p < 0.0001) (Figs. 5(a) and 5(b)). There were no statistically significant differences in overall joint stiffness, however, between boars and barrows (p = 0.10) (Fig. 5(c)). Joint neutral zone was calculated as the distance between engagement points of the ACL and the PCL. Both age (p = 0.03) and castration (p = 0.009) had overall significant effects on joint neutral zone (Fig. 5(d)). The barrows had shorter neutral zone lengths than the boars, and neutral zone increased throughout growth. The neutral zone of the boar group increased by 0.6 mm from 3 months to 4.5 months, then decreased by 0.2 mm to 6 months. Normalized neutral zone values showed no statistically significant effects of age (p = 0.40) or castration (p = 0.17) (Table S5 available in the Supplemental Materials). Similar changes were seen in the joint laxity, reported as APTT (Fig. 5(e) and Table S6). Both age (p = 0.0008) and castration (p = 0.008) had pronounced impacts on APTT, with barrows having lower APTT values than the boars, specifically 19% shorter APTT at 3 months (p = 0.03). Both groups had increased APTT values across the age groups tested. However, once normalizing to joint size, these age and castration effects disappeared (Fig. 5(f)).

Similar effects were seen in terms of ACL and bundle function. In situ stiffness increased with age (ACL: p < 0.0001, AM bundle: p < 0.0001, PL bundle: p = 0.003) but did not change due to castration (ACL: p = 0.11, AM bundle: p = 0.76, PL bundle: p = 0.15) (Figs. 6(a)–6(c) and Table S7 available in the Supplemental Materials). ACL in situ stiffness values for the boars increased by 111% from 3 to 6 months, and the barrow in situ stiffness values increased by 91%. Normalizing in situ tissue stiffness using length and CSA showed similar results for the ACL and AM bundle as in situ stiffness, but PL bundle normalized in situ stiffness did not change across age (Fig. S3 available in the Supplemental Materials). The ACL was the primary contributor to resisting anterior drawer in both the boars and barrows, but the two bundles of the ACL had distinct contributions to resisting load that varied across age (Fig 6(d) and Table S8 available in the Supplemental Materials). At 3 months, the AM bundle contributed slightly more to resisting anterior drawer than the PL bundle for both boars and barrows. From 3 months to 4.5 months, the AM bundle in the boars became the primary stabilizer in the ACL, taking on 85% of the total force applied to the ACL during anterior drawer. This continued to increase up to 91% in the 6-month boars. In the barrows, the AM bundle contributed 70% and 77% at 4.5 and 6-months, , respectively. There was a statistically significant age effect seen (p = 0.012), with AM bundles becoming more dominant across growth for both boars and barrows. Furthermore, all 4.5-month and 6-month differences in percent bundle contribution were statistically greater than 0, indicating higher AM bundle contribution. There was no statistical significance between boars and barrows (p = 0.24), yet there was a 9% increase in AM bundle contribution for barrows between 3 and 4.5 months, whereas in boars, the contribution of the AM bundle increased by 47%.

This work explored the influence of neo-natal surgical castration on ACL morphology and function later in youth and early adolescence in a male porcine model. Surgical castration lowers serum testosterone levels by over 95% [25] and is a standard practice in the swine industry to control for meat quality [41]. The ACL increased considerably in size and in situ stiffness across skeletal growth for both intact males (boars) and castrated males (barrows). Relative to changes in growth, however, there were only subtle differences between the two boars and barrows. Barrows had, on average, smaller overall knee joints along with smaller absolute ACL, AM bundle, and PL bundle volumes, but no difference in ACL or bundle CSA. When accounting for joint size, any differences disappeared. Functionally, the AM bundle percent contribution increased across growth for both groups, yet the boars had a more dramatic shift from equal bundle percent contribution to a large increase in AM bundle percent contribution.

While joint size has not been previously characterized between boars and barrows, it has been shown that barrows have lighter tibias than boars despite similar density and length measurements [42], indicating greater bone thickness in boars. Differences in body composition between boars and barrows have been well-characterized, with barrows having more fat and less bone and skin weight than boars [42–44]. This finding is similar in humans, where supraphysiological doses of testosterone lead to more fat-free mass and muscle size in men [45]. The smaller joints in the barrow group, however, are not clinically significant as there were no changes in absolute ACL CSA even with the smaller joint. When looking at changes across skeletal growth, the increase in ACL length, CSA, and volume matches previous findings in both pigs [28,35] and humans [46–48]. The minor changes in ACL morphology observed in this study, especially after normalizing to the overall joint size are similar to other findings directly relating intact and castrated animals [26]. A previous study measured ACL CSA after 12-weeks in rats castrated as adults and intact male rats and found that the ACL CSA did not differ between the two groups. The volume of the ACL and the AM and PL bundles had more substantial changes between boars and barrows, with barrows having smaller ACLs, AM bundles, and PL bundles than boars. The more substantial changes in ACL, AM bundle, and PL bundle volume could simply be a result of magnifying smaller changes seen in the CSA and length or could be of a result of larger differences near the entheses of the ACL. However, normalizing to joint size also reduced these differences in volume, indicating that surgical castration may influence absolute size, but may not influence normalized joint properties that can be used to compare individuals.

Regarding mechanics, both boars and barrows had increasing ACL, AM bundle, and PL bundle in situ stiffnesses across skeletal growth, yet no significant differences were observed between boars and barrows. While the stiffness of the joint and the in situ stiffnesses of the tissues increased throughout skeletal growth, there were also minor increases across growth for boars and barrows in absolute joint laxity as measured by APTT, but similar laxity values once normalized to joint size. Furthermore, the barrows had on average less lax joints and smaller neutral zones, especially at 3 months, yet this disappeared after normalizing to joint size. The normalized laxity values for the boars and barrows only varied 5.8% in the boars and 5.1% in the barrows across ages. Human studies often report raw values of APTT at static loads across ages, such as one study that showed that skeletally mature men have a decreased knee laxity compared to immature men where female knee laxity does not change [49]. This differs in relation to this study where scaled loads were applied, and laxity values were similar within boars and barrows. If non-scaled loads were applied, then there might have been similar decreases in absolute laxity across skeletal growth.

Previously, the pig model has been used to detect differences in bundle function across skeletal growth and found that the AM bundle takes on more of a dominant load over the PL bundle as adolescence proceeds [35]. At 3 months, the AM bundle contribution of the barrows (64%) is closer to the boars (58%) than females (45%). By 6 months, the AM bundle contribution in the barrows (77%) matches the AM bundle contribution in female pigs (76%), but is lower than boars (91%). Temporally, the female AM bundle contribution increases steadily from 3- to 4.5- to 6-months, boars have a rapid rise between 3- and 4.5-months then plateau, and barrows have higher AM bundle contribution already at 3-months, which gradually rises over time. Although there are no statistically significant differences in the normalized morphological or in situ stiffness values of the ACL or its bundles (potentially due to limited sample size), in qualitative comparisons across ages, the bundles in the barrow behave more similar to female pigs than to intact male pigs (Fig. S4 available in the Supplemental Materials on the ASME Digital Collection). When looking at the bundle function across skeletal growth and overall joint size, the timing seems to support an earlier maturation in boars than in barrows, yet CSA and in situ stiffness values seems to indicate no difference in ACL maturation. Perhaps testosterone has greater influence on the microstructural properties of the collagen fibers in the ACL, as shown by a decrease in collagen content and increase in collagen fiber diameter in castrated male rabbits [50]. Given that the rise of testosterone in boars starts around 4-months [51], the slight difference in timing of functional changes could be influenced more by testosterone especially between the 3-month and 4.5-month range. Additional variables could also be at play when making comparisons between gilts, boars, and barrows, such as fluctuations in other female sex hormones [18–22].

Results from this work suggest that surgical castration at an early age has minor impacts on the ACL structure and function several months later into early adolescence. The current results cannot explain the continued rise in ACL, AM bundle, and PL bundle CSA in intact male pigs in early adolescence compared to a plateau in ACL and PL bundle CSA in female pigs over the same age period. Other factors exist, such as the equilibrium of the conversion between androgens and estrogens via the aromatase enzyme [52]. With this conversion, castration can also decrease serum estrogens in males. With other studies showing that females have higher joint laxity than males, which is a known risk factor for ACL injury [53,54], previous work in this pig model showing a 38% greater APTT for female pigs at 18-months [35], and this work showing similar normalized laxity values for boars and barrows, future work should focus on the need to directly measure multiple hormonal factors in females.

There are some limitations to the present study that may provide new pathways to understanding the role of testosterone in the developing male ACL. A primary limitation is that while the pig is a good model for understanding human knee function due to its similar anatomical structures and size, no animal model completely replicates the human condition. This study also had a limited sample size. Thus, although we found no statistically significant differences between the boar and barrow ACL structure and function, we cannot rule out that differences may be found with enhanced statistical power. Another limitation of this study is the loss of stabilizing muscles surrounding the knee during applied loads. Testosterone has been shown to have more drastic impacts on muscle mass and strength [55,56], so there still could be a greater impact of testosterone on in vivo joint stability between boars and barrows due to greater muscle stabilization. When looking at the functional characteristics of the ACL and its bundles, only subfailure properties were assessed. A previous study in rats discovered that castrated rats had lower load-to-failure forces and ultimate tensile stresses than intact males, yet these failure properties were not assessed in this current study [26]. Furthermore, there was a high variability in the barrow bundle percent contributions. Due to limitations of specimen collection, there was also a lack of serum testosterone measurements in the boars, so associative analyses could not be done. Longer-term data in older animals would perhaps show a long-term impact of testosterone after its normal rise at 4 months. Sex hormones are generally complex in nature, with independent hormones rarely impacting macroscopic properties, and studying the interaction between fluctuating levels of serum sex hormones will prove useful in attempting to understand sex-specific differences throughout skeletal growth and ACL injury risk.

In conclusion, surgical castration had subtle impacts on joint and ACL morphology and function across skeletal growth. Across skeletal growth, the boars and barrows had increasing joint sizes, ACL CSA values and volumes, and in situ stiffnesses. Between boars and barrows, the joint size, ACL, AM bundle, and PL bundle volumes were smaller in the barrow group than the boars across ages, but not the ACL CSA, in situ stiffness, or bundle contribution. The barrows had a more gradual shift in bundle function toward AM-dominance in resisting load than boars, although this was not statistically significant due to high variability in the barrows and a small sample size. Thus, these results show that early and sustained loss in testosterone leads to some differences in ACL morphology but may not influence measures that may indicate increased injury risk, such as CSA or force in the bundles in response to applied loads. This work is beneficial in understanding that testosterone may not impact ACL performance and injury risk, although further work is needed.

We would like to thank the Swine Education Unit at North Carolina State University, Laboratory Animal Resources at North Carolina State University, and the Biomedical Research Imaging Center at the University of North Carolina—Chapel Hill for their contributions to this work. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

• Division of Graduate Education (Grant No. DGE-1746939; Funder ID: 10.13039/100000082).

• National Institute of Arthritis and Musculoskeletal and Skin Diseases (Grant Nos. F31AR077997 and R01AR071985; Funder ID: 10.13039/100000069).

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.