Thirty-nine elite competitive alpine skiers (skier group), all members of the Swiss Alpine Ski Team, and 38 age-matched healthy controls were recruited for the conduction of this case–control study. In winter, i.e., during the competition period, the training of elite competitive alpine skiers typically includes 1 to 3 on-snow competitions and/or on-snow training sessions, as well as 7 to 14 off-snow training sessions per week (Gilgien et al., 2018). In summer, i.e., during the off-season training period, they train an average of 5–9 sessions on-snow and 6 to 18 sessions off-snow per week, resulting in a total of 14–21 h of training per week (Gilgien et al., 2018). During the first part of the off-season training period, in which also the present study was conducted, there is a particular focus on physical conditioning with 2–4 sessions of endurance training, 2–4 sessions of strength training, 1,2 sessions of explosive strength training and plyometrics, 2,3 sessions of agility and motor training, 3–5 sessions of stability and mobility training, and 1,2 sessions of cross training in other sports (Gilgien et al., 2018). The following inclusion criterion was applied: Physically active defined as more than one session of intense physical activity or more than 30 min of moderate physical activity per week. The number of participants recruited was limited by the number of active national competitive alpine skiers as part of the Swiss Alpine Ski Team. All skiers were monitored for health problems in a two-weekly interval with the Oslo Trauma Research Center (OSTRC) questionnaire for a minimum of 9 months prior to and up to the day of testing. Exclusion criteria were: 1) manifested signs of tendon disease or systemic pathology relevant to the purpose of the study; 2) acute PT tendon complaints; or 3) “severe” traumatic knee injuries leading to long-term training absences of more than 28 days (Fuller et al., 2006) within the 9 months preceding the US assessments. The medical records were verified and updated during an additional check-up by a medical doctor prior to SWV assessments. Nine subjects from the skier group were excluded, as they met exclusion criteria; 4 were excluded due to exclusion criteria 2, and 5 were excluded due to exclusion criteria 3. SWE assessments in the skier group were performed at the start of the off-season training period in May. Controls were particularly recruited to be healthy and not having experienced patellar tendon-related knee pain within the last 12 months. All subjects were informed of the aim of the study and provided written consent. The study was approved by the local ethics committee (KEK-ZH-NR: 2017-01395).

US assessments and data processing were performed exactly as described in our validation study, which revealed good interoperator reliability (ICC of 0.736) and very good interday reliability (ICC of 0.904) (Götschi et al., 2021). One assessment was performed for each tendon by an experienced examiner; in total, 4 examiners were involved. The participant was in the supine position with adjustable support underneath the knee and a knee flexion angle of approximately 20°, and the right patellar tendon was examined (Martinoli, 2010). A US device (Aixplorer^® Ultimate, Supersonic Imagine, Aix-en-Provence, France) capable of high frame-rate US acquisition equipped with a linear transducer (SL18-5) was used. The US transducer was held perpendicular to the skin, and parallel alignment with the fiber axis was ensured by observing and maximizing the fibrillar appearance of the intratendinous structure. Position and orientation of the US transducer was tracked throughout the scanning procedure with optical markers attached to it and an optical tracking system (FusionTrack 500, Atracsys LLC) (Figure 1A). 2D SWE and US brightness-mode (B-mode) data were projected into 3D space and aggregated using a weighted averaging scheme to yield a 3D voxel representation of the measurement (MATLAB 2020a; The MathWorks, Inc., Natick, MA, United States ). The tendon was then segmented in the US B-mode volume, and the SWE volume was masked accordingly (3D slicer, version 4.10.2, (Kikinis et al., 2014). To allow analysis of distinct PT regions, the masked SWE reconstruction was divided into three regions with two cuts positioned at a distance of 10 mm from the proximal and distal insertion (Figure 1B), analogous to the approach described in our validation study (Götschi et al., 2021). Briefly, the area of interest was identified using landmarks placed during manual segmentation of the volume. Accordingly, the most distal point of attachment of the patella to the patellar tendon and the most proximal point of attachment of the tibia were marked for the proximal and distal sections of the PT. Further analysis was performed based on mean values for the respective regions.

To estimate the association of group membership while correcting for other potentially confounding variables with regional SWV, a multivariate approach is needed. Given the sample size available, including all factors of interest (including interaction effects) in one linear model may yield erroneous estimates due to overfitting. Hence, we opted for an alternative approach where we determined the linear model for inference by selecting the set of predictors with the best predictive performance in a leave-one-out validation scheme (maximum adjusted R ²). Consequently, for factors not included in the final models, no effect estimates are available. Univariate comparisons between groups were conducted with independent samples t tests. Statistical analysis was performed with R (Team, 2021), and statistical significance was set at α = 0.05.

The baseline characteristics of both study groups are summarized in Table 1.

Linear regression modeling revealed group membership to be the only determining factor of SWV over the entire study population for all assessed tendon regions (overall, proximal, mid-portion, and distal), with skiers displaying higher SWV (Table 2).

When assessing each study group separately, in the skier group, both subject age and sex were retained in the final regression model for the midportion region of the PT (Table 3). In the control group, no predictors for mid-portion SWV were identified.

Figure 2 visualizes shear wave velocity differences between groups as well as the significant sex effect in the skier group and the absence thereof in the control group.

We found that the regional difference in SWV between the distal and mid-portion PTs differed between the skier and control groups (p = 0.023). The two groups, however, did not differ with regard to the regional delta between the proximal and PT mid-portions (p = 0.599, Figure 3).

Physiological tendon hypertrophy and remodeling in response to altered loading regimes allows the tendon to cope with increased training volume and intensity. Moreover, optimization of muscle tendon unit performance requires both muscle and tendon to adapt, whereas the tendon is responsible for effective force transmission. In this context, lower tendon hysteresis has been reported in ski jumpers and runners compared to control subjects (Wiesinger et al., 2017). Pathological adaptations, on the other hand, lead to deterioration of the tendon structure, limiting the capacity for physical performance and increasing the risk of tendon rupture (Cook and Purdam, 2009; Ooi et al., 2016; Yasui et al., 2017). Competitive alpine skiers are exposed to tremendous loads both during off-season training and in the competition season (Supej and Holmberg, 2010; Hydren et al., 2013; Gilgien et al., 2018). High accelerations combined with forward pushed, highly flexed knees induce large forces acting on the knee joint and, in particular, the PT (Alhammoud et al., 2020). Moreover, compared to other sports, particularly in competition situations the demands on the PT in alpine skiers are relatively high (90–120 s of high-intensity loading of the quadriceps muscles with peak ground reaction forces of up to four times body weight) (Supej and Holmberg, 2010). Therefore, off-snow training modalities are also designed to meet these unique demands, making extensive adaptations of the PT in skiers theoretically sound. Finally, such extraordinary demands can overwhelm the system’s capacity for adaptation, which may manifest in a high prevalence of overuse complaints in this cohort (Fröhlich et al., 2020; Fröhlich et al., 2021). A closer understanding of the adaptive processes in response to long-term repetitive loading with regard to PT material properties is crucial in prevention and early detection strategies.

When assessing intercohort differences, adjustment of cofounding variables such as demographic factors is necessary. To do so, the selection of linear models with the best predictive performance is a suitable method (Stone, 1974). In our dataset, the effect of the tendon loading regime (i.e., group membership) outweighed biological factors such as weight, height, sex, age, tendon CSA, and tendon length. Zhang et al. reported lower stiffness in the PT of athletes; however, the inclusion criteria for the athletic cohort was a minimum of 4 h of exercise per week; hence, the inclusion criteria for the control and athlete groups are not comparable with our study (Zhang et al., 2015). Other studies agree with our finding: Increased PT SWV in athletes has been reported when compared to nonathletic cohorts (Zhang et al., 2015; Bayliss et al., 2016; Mersmann et al., 2017; Cristi-Sanchez et al., 2019; Dirrichs et al., 2019; Selcuk Can et al., 2021). A quantitative comparison of relative magnitude from our study shows a 1.17 times higher SWV in PT in skiers compared to the control group, whereas it is 1.47 when calculated with the results from Selcuk Can et al. (Selcuk Can et al., 2021). The difference may be explained by the different activity levels of the control and athlete groups in the two studies, but nevertheless, the two factors are in a similar range.

Furthermore, increased tendon SWV values have been reported to result from altered loading regimes (Malliaras et al., 2013; Bohm et al., 2015), confirming the tendon’s ability to adapt to load and supporting our finding that the loading regime is the main factor inducing physiological tendon adaptation. Interestingly, a sex-specific difference in SWV was only found in the skier group, which may indicate that high loads must be present to elicit sex-specific adaptation mechanisms. Some studies suggest that there is a sex-specific difference in mechanical tendon properties (Kubo et al., 2003; Onambele et al., 2007; Tas et al., 2017), while others do not (Burgess et al., 2009; O’brien et al., 2010; Morrison et al., 2015); hence, in the literature, there is no consensus in this regard. On the one hand, hormone-based effects on tendon adaptation have been described: Acute exercise-induced increase of tendon collagen synthesis is inhibited by estrogen (Magnusson et al., 2007) and is therefore diminished in magnitude and time, compared with men (Miller et al., 2007). On the other hand, male subjects are both heavier and suspected to be able to generate higher knee extension moments than female subjects, leading to increased PT forces and subsequently inducing PT adaptation. In agreement with this hypothesis, Arampatzis et al. found a positive correlation between maximal knee extension moment and PT stiffness (Arampatzis et al., 2007). Another study found increased PT stiffness in heavy loaded legs, whereas no effect was found in light loaded legs (Kongsgaard et al., 2007). Furthermore, no significant correlation between knee extension torque and PT stiffness has been found in subjects with no specific training engagement (Mannarino et al., 2018), thus supporting our hypothesis. The literature states that female sex is a protective factor regarding patellar tendinopathies, referring to team sport athletes (Morton et al., 2017). Since the difference in PT SWV was observed in competitive alpine skiers but not in the control group, it may be assumed that lower forces applied in female recreational athletes and, therefore, overuse complaints at the PT occur less frequently.

In the current study, different regional patterns in the PT SWV were found between skiers and controls, with significant differences between the delta from the mid-portion to the distal region. Functionally and clinically, proximal regions, mid-portions, and distal regions differ, and thus, it is expected that SWV as a quantification for tendon adaptation (i.e., collagen cross-linking) reflects the heterogeneous regional characteristics. Interestingly, Kongsgaard and others found an increase in proximal PT CSA after light and heavy training in healthy untrained men but a distal increase in CSA only after heavy training, with the training consisting of knee extension exercise (Kongsgaard et al., 2007). Since our study included only physically active subjects and yet the athletic cohort performs heavier physical activity, our finding is reflected by the results from Kongsgaard et al. (Kongsgaard et al., 2007). The precise structural adaptations that underlie elevated tendon SWV are currently unclear; nevertheless, the nonuniform adaptation behavior along the PT in response to the loading regime cannot be neglected. Thus, further research should implement a more sophisticated view regarding loading magnitude with regard to regional adaptations.

This study has limitations to be considered. Physical activity levels of control subjects were not quantified on a continuous scale, and thus, there are potentially highly active subjects included in the control group. Since high-level athletes usually have more muscle mass than recreational athletes, BMI between these groups significantly differs, hence representing a cofounding factor. Furthermore, a challenge in assessing tendon SWV is the alignment between the transducer and tendon fiber orientation. Despite greatest caution in aligning the transducer to the fiber orientation, situations where the fibers are not colinear over the entire measurement frame cannot be compensated for. Finally, tendon loading in the hours prior to or during the measurement was not controlled for; therefore, we asked the subjects to remain relaxed in the measurement position for 5 minutes and refrain from strenuous physical activity 24 h prior to the assessment. An interesting aspect that cannot be concluded from the present study and requires further investigation is whether there are differences in the SWV profiles of skiers in the different periods of a season.