A cohort of collegiate men's volleyball athletes were monitored across the competitive sport season to identify neuromuscular performance and tendon adaptations of the knee. External and internal loads were assessed for each practice and match. The training and competition schedule for included time points is represented in Figure 1. Participants arrived at the laboratory on the same day each week between 7:00–11:00 am and had refrained from physical exercise. In addition to sport practices, athletes participated in the same periodized resistance training program led by a Certified Strength and Conditioning Specialist^®. Upon arrival to the laboratory, standardized locations on the patellar tendon were marked for each participant. Tissue properties were assessed using ultrasonography to quantify tendon thickness and echogenicity, and myotonometry to quantify tendon stiffness. Next, participants were instructed to perform a supervised ∼10 min standardized warm-up, which consisted of dynamic exercises with incrementally increasing intensity and ending in maximal effort practice trials of the tested movements. Data collection began with the countermovement jump and concluded with the drop vertical jump. Countermovement jump variables included jump height, peak force, and reactive strength index-modified (RSImod) while drop jump variables included jump height, peak force, and reactive strength index (RSI).

Sixteen NCAA Division I men's volleyball athletes (age: 20.3 ± 1.2 years, height: 193.5 ± 6.5 cm, bodyfat %: 13.18 ± 4.72%, body mass: 84.3 ± 8.0 kg) participated in the data collection. The inability to participate in maximal effort performance testing throughout the course of the data collection excluded one athlete from participation. Players were medically cleared for intercollegiate athletic participation by the University's sports medicine staff, had the risks and benefits explained to them beforehand, signed an institutionally approved consent form to participate, and completed a medical history form. This study was conducted according to the Declaration of Helsinki guidelines and procedures were approved by the University's Institutional Review Board for use of human subjects in research.

Data analysis was performed using statistical software (R Studio, version 1.4.1717, https://www.r-project.org). Descriptive statistics are reported as mean ± standard deviation. Intra-class correlation coefficients (ICC) and standard error of measurement (SEM) were used to assess intra-rater reliability for all patellar tendon measurements. To assess intra-rater reliability, the two-way mixed effects, absolute agreement, and multiple rater/measurements model was evaluated using the following formula (44):(4)ICC(3,k)=MSR−MSEMSR+MSC−MSEnThe estimated ICC and the 95% confidence internal were reported. ICCs were interpreted as: Poor (<0.5), moderate (0.5–0.75), good (0.75–0.9), excellent (>0.9) (45). SEM was calculated using the following formula:(5)SEM=SDpooled×1−ICCRepeated measures correlations (RMCORR) were assessed using the rmcorr package. RMCORR is a robust analysis strategy enabling the assumption of independence to be relaxed in a repeated measures design (46). RMCORR is also robust to missing data representing conceptual similarity to a multilevel null model allowing for random intercepts while evaluating common slopes representing intra-individual associations (46). Considering these advantages, RMCORR better represents the research question compared to repeated measures analysis of variance (ANOVA) or simple linear regression or correlation, which defers to aggregated data or inter-individual differences, respectively. Assumptions of linearity, homoscedasticity, and normality of residuals were assessed for each predictor variable. Correlation coefficients for RMCORR (r-rm) and Pearson's correlation coefficients (r) were interpreted as: trivial (<0.1), small (0.1–0.3), moderate (0.3–0.5), large (0.5–0.7), very large (>0.7) (47). Significance was reported as p < 0.05. Workload variables were assessed corresponding to the week of the testing procedures as well as time lagged by one week to allow for causal insights on dependent variables. Workload – tendon structure and workload – neuromuscular performance associations were calculated and reported as between-subject correlations (r) and within-subject correlations (r-rm). Plots for each r-rm association are provided in Supplementary Figures 1–4.

Results for RMCORR are presented in Figures 4A,B for countermovement jump and drop vertical jump, respectively. Significant associations were exhibited between countermovement jump peak force and jump count (r-rm = −0.26; small negative, p = 0.02) of the current week as well as workload efficiency lagged (r-rm = −0.31; moderate negative, p = 0.03). No other significant associations were established for countermovement jump performance. Drop vertical jump height exhibited a significant association with jump count lagged (r-rm = −0.30, moderate negative. p = 0.03). No other drop vertical jump variables were associated with any workload metrics. Pearson's correlation coefficients are presented in Figure 5.

Results for patellar tendon structure are reported in Figures 4B,C representing dominant and non-dominant limb, respectively. Echogenicity of the dominant limb exhibited significant associations with jump count lagged (r-rm = 0.32; moderate positive. p = 0.003) and sRPE lagged (r-rm = 0.24; moderate positive. p = 0.03). Thickness of the dominant limb exhibited a significant association with jump count lagged (r-rm = −0.24; small negative. p = 0.03). No significant associations were exhibited in stiffness of the dominant limb and any workload metrics. Echogenicity of the non-dominant limb exhibited significant associations with jump count (r-rm = −0.23; small negative. p = 0.01) and sRPE (r-rm = −0.30; moderate negative. p = 0.001). Thickness of the non-dominant limb exhibited a significant association with workload efficiency (r-rm = 0.20; small positive. p = 0.03). Stiffness of the non-dominant limb exhibited a significant association with jump count lagged (r-rm = −0.21; small negative. p = 0.04). Pearson's correlation coefficients are presented in Figure 5.

Accumulated exposure may result in fatigue, which is characterized by decrements in force production and kinematic alterations in movement strategy. Previous literature has demonstrated sensitivity of RSImod and flight time-to-contraction time ratio to indicate fatigue when exposed to high workloads (16, 48). However, current findings were not in support of those previously reported as a lack of intra-individual associations with RSI, RSImod, and workload metrics existed. Interestingly, a moderate negative intra-individual association between countermovement jump peak force and workload efficiency lagged was shown indicating a tradeoff of decreased force with a concomitant increase in workload efficiency of the previous week. This may have been influenced by training decisions outside of the monitored sport activity (e.g., resistance training). Considering workload efficiency is a ratio, changes in constituent parts relative to one another may influence the outcome. For example, if a higher increase in external workload relative to internal workload exists yet both parameters increase, workload efficiency will still improve. Perhaps this disparity in performance impairments is more affected by mechanical fatigue than perceived exertion. However, underlying contextual factors may exist and considering a lack of association between internal and external workload independently, whether this is an adaptation or not remains inconclusive.

Small to moderate negative associations were exhibited between countermovement jump height and drop vertical jump height with athlete workload metrics. In support of previous literature reporting variation in jumping demands is dependent upon player role, small negative associations between RSI and load metrics were exhibited (49). In the current analysis, athletes with lower jump counts displayed higher jump heights and higher reactive strength capacity. Back row players are characterized by higher jumping volumes during training and competition compared to front row players, who perform fewer jumps yet require higher intensity jumps for tasks such as blocking and spiking (50) A few possibilities exist to explain the small effect sizes seen in the current research. Slow (>250 ms) stretch-shortening cycle movements require unique force expression compared to fast (<250 ms) stretch-shortening cycle movements and may exhibit unique adaptations (51). A negative association with drop vertical jump height, but not countermovement jump height, may indicate fast stretch shortening cycle performance is impaired following higher workloads. Previous findings suggest countermovement jump height may remain stable as a result of technical alterations in order to maintain performance outcomes (52). The shorter duration movement may be more sensitive to fatigue and not provide adequate time for compensation. Another possible explanation is that a non-linear association exists between loads and neuromuscular performance indicators. Exposure and response varies across individuals and across time and thus, may be better supported as a non-linear association (53). More complex analysis techniques such as linear mixed models (54, 55) or time to event analysis (56, 57) may be suitable to address potential insights into individual slopes and time varying associations.

Habitual plyometric loading may elicit positive tendon adaptations provided the stimulus is adequate and ample recovery afforded (29, 58). Conversely, maladaptation may occur in response to suboptimal loading patterns leading to hypoechoic regions and increases in thickness (26). Tendon thickening has been established as an adaptive mechanism allowing for more efficient transfer of mechanical energy (59) with supporting evidence that patellar tendon thickness is associated with higher countermovement jump height (34). Acute inflammatory responses may partially explain the early stages of tendon pathology (60). Preseason training has been characterized by the highest training loads compared to other phases of competition in collegiate volleyball athletes (61). The current data support a decrease in thickness, which may indicate a reduction in inflammation from the high stress of preseason training. Thickness of the non-dominant patellar tendon exhibited a small positive association with workload efficiency indicating a decreased perception of effort with a reduction in thickness possibly influenced by a reduction in inflammation. Psychophysiological assessments, such as sRPE, are influenced by the current status of the athlete. Further, inflammation is linked to pain in tendinopathy literature possibly influencing perceived exertion (60, 62). In the dominant patellar tendon, an increase in echo intensity and decrease in thickness were associated with increased jump count lagged. Increase in echogenicity was associated with increased sRPE lagged while echogenicity of the non-dominant patellar tendon exhibited small and moderate negative associations with jump count and sRPE. Ground reaction forces upwards of 4× bodyweight can be experienced in a landing following an attack in volleyball (63) Attackers have been shown to prefer unilateral landing characterized by landing contact of the opposite side of the swing arm 68% of the time (64). The aforementioned findings may explain the observed inter-limb discrepancies in the current study and emphasize the need to develop more precise methods of measuring mechanical load (65).

Negative inter-individual associations were exhibited between echogenicity of non-dominant and dominant patellar tendon and sRPE, suggesting perceived exertion may be impacted by patellar tendon structure. A negative inter-individual association was exhibited between echogenicity of the non-dominant patellar tendon and jump count, suggesting a healthier tendon on the non-dominant limb in those who perform fewer average jumps. Overuse injury is more dependent upon magnitude of loading than frequency of loading (14). Thus, frequent unilateral landing of the non-dominant side in conjunction with large ground reaction forces when landing from higher jump heights may explain the observed differences in dominant compared to non-dominant adaptations.

Stiffness of the non-dominant patellar tendon exhibited a small negative intra-individual association with jump count lagged. Pathological patellar tendon tends to exhibit a decreased stiffness, which may lead to higher strain in the tendon (26). High strain is characterized by larger deformation when exposed to mechanical forces, which may be responsible for risk of structural damage (2). Intra-individual associations with time lagged external load and stiffness and with current weak external load and echogenicity may indicate temporal differences in the onset of structural changes in response to load. To date, patellar tendon pathology has been identified through unique characteristics present in pathological conditions (22, 26, 66). Limited research has examined the latency of structural and mechanical abnormalities in patellar tendinopathy (30). While abnormalities coexist in pathological conditions, the time course of onset and recovery may differ. The current research provides novel insight into intra-individual associations with external and internal loads, neuromuscular performance, and patellar tendon structural properties. Further, the findings support the utility of assessing and monitoring ultrasound imaging metrics relevant to patellar tendinopathy in volleyball athletes.

Monitoring athlete load to mitigate risk of overuse is worthwhile in order to determine optimal load prescription. However, the utility of ultrasound image analysis to assess tendon pathology remains in question. The current analysis provides support for associations between athlete load and patellar tendon structure, but requires further evidence to identify relevant contextual factors. It is recommended that the robustness of the current structural properties along with other relevant image metrics of tendon pathology be examined. Since tendon pathology demonstrates an array of abnormalities with each likely occurring on a unique timeline, investigation into the time course and development of abnormalities may help guide detection and risk mitigation.

The current research is not without limitations. Time lagged workload metrics included 7 of the 9 time points with additional missing data depending upon the individual. Although RMCORR is robust to missing data, a complete dataset would provide stronger evidence. Further, RMCORR is limited to linear associations and although suitable for addressing the fundamental research question presented in the current research, more complex modeling could be used to evaluate time-varying, non-linear trends in load exposure, training responses, and associations. Quantifying mechanical load experienced in the tissue is theoretically crucial to understanding adaptations of the corresponding tissue. This poses limitations on interpretation and may reduce the effectiveness of determining associations with structural adaptations. The accelerometers used in the current study are worn close to the athlete's center of mass and detect movements as a system. Jumping and landing tasks often impose unilateral demands that cannot be accurately depicted; therefore, reliance on systemic, whole-body surrogate measures may account for the small effects observed. Injury data was unavailable at the time of this data collection. Previous injury is known as an influential factor for future pathology, particularly in patellar tendinopathy, and is recommended to include as an intrinsic risk factor when evaluating injury risk and mitigation strategies. Asymptomatic tendon as well as unreported symptomatic tendon may influence these findings and is recommended to be further investigated.