30 patients suffering from end-stage gonarthrosis (male = 18, female = 12, age = 63.5 ± 8.1 y, height = 176.4 ± 7.2 cm, weight = 86.9 ± 16.1 kg) participated in this study. Patients were randomly assigned into one of three groups: 1) Control-Group (CON): This group followed the standard clinical treatment without a specialized prehabilitation protocol. 2) Active-Control-Group (AC): The second group followed the standard clinical treatment and participated in a 6-week prehabilitation protocol with a sham-BFR application. 3) BFR-Group (BFR): The third group followed the standard clinical treatment and participated in a 6-week prehabilitation protocol with additional BFR application. The standard clinical treatment consists of the surgery and 7 days of hospitalization with daily physical therapy, which was followed by 3 weeks of inpatient rehabilitation.

The study design consists of a prospective, single blinded, parallel study design to determine the influence of prehabilitation on pre- and postoperative skeletal muscle mass and strength. While the CON underwent routine clinical practice without prehabilitation, the other groups completed an identical 6-week prehabilitation protocol, one with additional BFRE (BFR-Group) and one with a sham-BFR application (AC-Group) before TKA, to reduce a potential bias in expectations of the intervention effect between groups. Preliminary visits were conducted before the start of the study to familiarize the patients with the cycling ergometer, testing protocols and tourniquet pressures.

Measurement time points were 6 weeks- (baseline), 3 weeks- (3w-Prehab) and 5 days before surgery (Pre-OP), as well as 3- (3m-Post-OP) and 6-months (6m-Post-OP) postoperatively.

At each time point, vital parameters like blood pressure (RR) and heart rate (HR) were recorded. Anthropometric data consist of recording body weight (BW), body height (BH) as well as the circumference of both thighs and calves for estimation of skeletal muscle mass of the lower extremities. To visualize the femoral circumference an anatomic reference line was drawn between the spina iliaca anterior superior and the margo superior patella. At 50% (FC-50) length of this reference line, the femoral circumference of both legs was determined. The calf circumference was determined after multiple measurements at the largest diameter. The measurement was performed three times at all points and the mean value was determined. Furthermore, the circumference of both knees was recorded to illustrate swelling pre- and postoperative. Measurement points were above the margo superior patellae, in the middle of the patella as well as above the apex patellae.

Muscular function in this study was analyzed by a unilateral six-repetition maximum test (6RM) of the leg-extension and leg-curl machine (Paoli et al., 2013). Following a warmup, a maximum of five trials separated by 5 min of recovery was allowed to obtain a true 6RM. With an accuracy of 1.00 kg, the highest load that the subject was able to lift six times to a knee extension of approximately zero degrees was accepted as 6RM.

The functionality of the participating patients was determined by active knee joint mobility (range of motion, ROM) as well as the chair-rising- (CRT) and 6-min walking test (6MWT). The ROM of the knee joint was determined in the supine position. A goniometer was used to measure the active extension and flexion of the knee joint (Bade et al., 2014). In CRT, the subject sits on an ordinary chair without armrests. With arms crossed on the chest, the test person performs as many stand-up and sit-down movements as he or she can manage within 30 s (Gill and McBurney, 2008). A complete repetition was then scored after the test person was in full extension while standing, as well as with a leaning back in a sitting position. The 6MWT is used to estimate and monitor cardiovascular and pulmonary performance below the anaerobic threshold. Participating patients walk for 6 minutes along a 20 m walkway. The goal for the patient is to complete as much distance as possible in the given time window. Individual pace changes and pauses are allowed. After 6 minutes, the time is stopped, and the walking distance is written down in meters.

To assess the subjectively experienced functional status, pain perception and quality of life of the patients, the Knee Injury and Osteoarthritis Outcome Score (KOOS) was used. KOOS contents of 42 questions in five different dimensions: Symptoms (seven questions), pain (nine questions), activities of daily living (17 questions), functionality in sports and recovery (five questions) and quality of life-related to the affected knee (four questions).

Statistical analyses were performed using SPSS (SPSS, v.27, Chicago, IL, United States). Normal distribution and homogeneity of variance were verified using Shapiro-Wilk and Levene’s Test, respectively. Potential group differences in baseline and surgical data (i.e., age, height, body weight, duration of surgery, etc.) were assessed using one-way ANOVAs. To compare the changes in measures over time among groups, two-way repeated-measures ANOVAs (rANOVAs; time x group) were performed. In case of significant time × group interactions, separate one-way repeated measures ANOVAs were used to analyze the simple main effects for time within each group. The mean differences between groups (i.e., simple main effects for group) within each time point were assessed using separate one-way ANOVAs. If the main effects for time or group were detectable, post hoc-tests with Bonferroni correction were performed to check which factor levels differ significantly from one another. For the interaction and main effects, the partial eta squared η_p ² was calculated as effect strength measure and interpreted as follows (Cohen, 1988): a η_p ² ≥ 0.01: small effect, ≥ 0.06: medium effect, and ≥0.14: large effect. For all results, an alpha level of 0.05 was interpreted as statistically significant. To reduce investigator bias, data analysis was blinded to the evaluating researchers.

Table 2 summarizes the training data during phase 1 (session 1–6) and 2 (session 7–12) of the prehabilitation including exercise intensity, mean HR and RR during training and individual LOP measured before training. All patients in the training groups completed all planned training sessions during the 6-weeks prehabilitation period. The current study documented a dropout rate of 0% and no exercise or BFR-related adverse events. For exercise intensity, there were a significant time × group interaction (p < 0.001, η_p ² = 0.599) and time effect (BFR-group: p < 0.001, η_p ² = 0.902; AC-group: p < 0.001, η_p ² = 0.900). Post hoc-tests revealed that the training intensity in both training groups significantly increased during the 2. phase of prehabilitation. No statistically noticeable changes in physiological measures during the training period were detected in any group (Table 2).

Table 3 summarizes the femoral- and calf-circumference of both operated- (OP) and non-operated (NonOP) legs and their percent difference during the prehabilitation- and post-operative-period. Significant time × group interaction effects were indicated for all measured femoral circumference values (p < 0.001, 0.350 < ηp2 < 0.494). Further analyses revealed significant time effects in the BFR-group (p < 0.001, 0.674 < η_p ² < 0.754) and CON (p ≤ 0.001, 0.627 < η_p ² < 0.728). The AC-group did not indicate any time effects (p ≥ 0.078, 0.178 < η_p ² < 0.262). For the calf circumference, no changes were detected in any group despite of the significant interaction effect (p = 0.018, η_p ² = 0.205) and main time effect in CON for the NonOP leg (p = 0.013, η_p ² = 0.426). Post hoc-tests revealed that the femoral circumference of both legs significantly increased in the BFR-group already at 3w-Prehab (p = 0.002). In addition, the BFR-group showed a further improvement in the femoral circumference of the OP leg after the prehabilitation period (p = 0.006). The CON did not indicate any changes in the femoral circumference during and after the prehabilitation period (p = 0.131). At 3m-Post-OP, both CON and BFR-group showed significantly decreased femoral circumference of both legs compared to the Pre-OP-level (p ≤ 0.017), but the BFR-group still remained above the baseline level (p = 0.023). At 6m-Post-OP, all femoral circumference in the BFR-group increased again (p ≤ 0.030) with significantly higher values compared to the baseline level (p < 0.001). In contrast, the CON demonstrated significantly reduced femoral circumference of both legs at 6m-Post-OP with significantly lower values compared to the pre-operative level (p ≤ 0.05).

Regarding the percent difference between the OP and NonOP leg, significant time × group interaction effects were observed for the femoral circumference (p = 0.014, η_p ² = 0.186), while for the calf circumference, there were no significant main or interaction effects (Table 3). Further analyses on percent difference in the femoral circumference revealed a significant time effect only for the CON (p = 0.002, η_p ² = 0.469) and a significant group effect at the baseline (p = 0.037, η_p ² = 0.217). Despite the absence of the significant time effects, both training groups indicated a decreased percent difference in the femoral circumference between both legs following the prehabilitation, which increased again during the post-operative period but not beyond the baseline level (Table 3). In contrast, the percent difference in the femoral circumference between both legs in the CON continually increased during the postoperative study period. It was higher at 6m-Post-OP than the baseline level (p = 0.013).

The knee circumference measured at three different places of both OP and NonOP legs during the prehabilitation- and postoperative-period are presented in Table 4. Regarding the knee circumference, we found no time × group interaction effects (p > 0.059, 0.060 < η_p ² < 0.164) except for the upper knee circumference of the Non-OP leg (p = 0.010, η_p ² = 0.213), which indicated no further time or group effects (Table 4).

Further analyses revealed a significant time effect only for the lower knee circumference of the OP leg (p = 0.017, η_p ² = 0.072), indicating that knee circumference increased at 3m-Post-OP and returned to baseline level at 6m-Post-OP.

Regarding the knee swelling accessed by the percent difference between OP and NonOP knee, a significant time × group interaction effect was found only for upper knee (p = 0.049, η_p ² = 0.172). For the swelling measured at middle knee, we found a significant group effect (p = 0.046, η_p ² = 0.211) indicating lower values in the BFR-group compared to other groups. For lower knee, found a significant time effect (p < 0.001, η_p ² = 0.373) with a continually decreased swelling during the prehabilitation period, which increased at 3m-Post-OP and returned to baseline level at 6m-Post-OP.

Further analyses for the swelling measured at upper knee detected significant time effect only in the CON (p = 0.017, η_p ² = 0.445). In addition, there were significant group effects at baseline (p < 0.012, η_p ² = 0.280) and 3m-Post-OP (p < 0.049, η_p ² = 0.200). At baseline, the upper knee swelling was significantly higher in the AC-group that the CON (p = 0.043), while the BFR-group did not differ to other groups (p ≥ 0.093). The CON demonstrated a greater increase in the upper knee swelling with higher level at 3m-Post-OP compared to the BFR-group (p = 0.046), which significantly decreased again at 6m-Post-OP (p = 0.044). Despite being statistically non-significant, both BFR- and AC-groups demonstrated decreased knee swelling values at 6m-Post-OP compared to the baseline level (Table 4).

Table 5 summarizes the ROM assessed during active extension and flexion. For all ROM measurements, there were no significant time × group interaction effects (p ≥ 0.403, 0.043 < η_p ² < 0.073). Further, significant time effects were detected only for OP leg (extension: p = 0.012, η_p ² = 0.138; flexion: p < 0.001, η_p ² = 0.468) indicating a continually improved ROM during overall study period. No main group effects were found for all ROM measures (p ≥ 0.169, 0.048 < η_p ² < 0.128).

For 6-MWT (Figure 2A), we found a significant time × group interaction effect (p = 0.012, η_p ² = 0.209). Further analyses revealed a significant time effect only in BFR-group (p < 0.001, η_p ² = 0.677). Post hoc-tests for BFR-group indicated a significant improvement in 6-MWT already at 3w-Prehab compared to baseline (390 ± 82 m to 431 ± 69 m, p = 0.034), which increased further after the prehabilitation (to 456 ± 58 m) with a significantly higher level to the baseline- and 3w-Prehab-level (p ≤ 0.048). At 3m-Post-OP, the BFR-group showed a significant deterioration in 6-MWT compared to Pre-OP (to 426 ± 73 m, p = 0.05), but it pronounced recuperated at 6m-Post-OP (to 464 ± 58 m, p = 0.002). Consequently, the BFR-group demonstrated a significantly higher ability in 6-MWT at 6m-Post-OP compared to baseline- (p = 0.004) and 3w-Prehab-level (p = 0.007).

Regarding the CRT (Figure 2B), there was a significant time x group effect (p = 0.007, η_p ² = 0.205). In addition, we found significant time effects in BFR- (p < 0.001, η_p ² = 0.671) and AC-group (p < 0.001, η_p ² = 0.596), whereas no changes occurred in the CON. According to Post hoc-tests, the patients in the BFR-group significantly improved in the CRT already at 3w-Prehab compared to the baseline (8.90 ± 2.08 reps. to 11.20 ± 2.35 reps., p = 0.012). In comparison, the AC-group showed a significant improvement only after the prehabilitation period (9.90 ± 2.51 reps. to 12.00 ± 2.49 reps., p = 0.006). At 3m-Post-OP, the AC-group exhibited a significant deterioration in the CRT compared to Pre-OP (to 10.00 ± 2.98 reps., p = 0.002), which pronounced improved at 6m-Post-OP again (to 11.30 ± 2.50 reps., p = 0.019), and was significantly higher to baseline level (p = 0.026). In contrast, the BFR-group indicated no statistically significant change in the CRT at 3m-Post-OP (12.10 ± 2.73 reps. to 10.90 ± 2.77 reps., p = 1.00), which was still higher compared to the baseline level (p = 0.038). At 6m-Post-OP, the BFR-group improved again (to13.30 ± 2.31 reps., p = 0.003) remaining above the baseline- and 3w-Prehab level (p ≤ 0.004).

The results regarding the muscular strength of both OP and NonOP legs and their percent difference during the prehabilitation and post-operative period are presented in Table 6. Significant interaction effects were observed for all measured muscle strength indices (p < 0.001, 0.567 < η_p ² < 0.625). Further analyses revealed a significant main time effect for all muscle strength measures in all groups (p < 0.05, 0.298 < η_p ² < 0.916). In addition, there were significant group effects for all measured leg strength indices (p ≤ 0.046, 0.204 < η_p ² < 0.676) excepting for leg extension of OP leg at baseline (p = 0.077, η_p ² = 0.173) and at 3w-Prehab (p = 0.230, η_p ² = 0.103). Post hoc-tests revealed that both training groups significantly improved in all measured leg strength indices already at 3w-Prehab (BFR-group: p ≤ 0.01; AC-group: p ≤ 0.026). At Pre-OP, the BFR-group indicated more pronounced improvements in all leg strength measures (i.e., 3w-Prehab to Pre-OP: p < 0.001) with significantly higher values compared to other groups (p < 0.05). No changes occurred in the CON during the prehabilitation period (p = 0.100). At 3m-Post-OP, significant reductions in leg strength measures were observed in BFR-group (p ≤ 0.01) except for the leg extension of the NonOP leg (p = 0.308), which were still above the baseline level (p ≤ 0.01). The AC-group indicated significantly decreased muscular strength in both leg-extension (only in NonOP leg) and -curl (in both legs) at 3m-Post-OP even to the baseline level (p ≤ 0.05). Similarly, the patients in the CON showed a significant decrement in leg extension of both OP and NonOP legs at 3m-Post-OP with a lower value compared to the pre-operative level (p ≤ 0.031). At the same time, no changes occurred in leg curl (p ≥ 0.187). At 6m-Post-OP, both training groups significantly improved again in all leg strength measures (BFR-group: p ≤ 0.029; AC-group: p ≤ 0.031) with a significant difference to baseline- (BFR-group: p < 0.001 for all measures; AC-group: p ≤ 0.029 only for leg curl of both legs) and to 3w-Prehab-level (only in BFR-group: p ≤ 0.038 for all measures). The CON also indicated significant improvements but only in muscular strength of OP leg (leg extension: p = 0.005; leg curl: p = 0.007). Consequently, there were significant differences in leg strength between BFR-group and other groups during the overall post-operative period (AC-group: p ≤ 0.032; CON: p ≤ 0.020) excepting for the leg extension of OP leg at 3m-post-OP between BFR-group and CON (p = 0.265).

Regarding the strength deficit of the OP leg accessed by the percent difference between OP and Non-OP leg during leg-extension and -curl, we found no significant interaction (p ≥ 0.063, 0.134 < η_p ² < 0.150) and time effects (p ≥ 0.105, 0.038 < η_p ² < 0.081). There were significant main group effects (p ≤ 0.003, 0.366 < η_p ² < 0.481) with lower values in BFR-group compared to other groups.

The analysis on the KOOS (Figure 3) demonstrated significant time × group interaction effects for all evaluated dimensions (p ≤ 0.004, 0.268 < η_p ² < 0.416). Further analyses revealed a significant main time effect for all measures of KOOS in all groups (CON: p ≤ 0.004, 0.475 < η_p ² < 0.907; BFR-group: p < 0.001, 0.869 < η_p ² < 0.978; AC-group: p ≤ 0.001, 0.571 < η_p ² < 0.951). In addition, there were significant group effects for all measures of KOOS (p ≤ 0.049, 0.200 < η_p ² < 0.581) excepting for the dimension symptoms at baseline, 3w-Prehab, 3m-Post-OP, and 6m-Post-OP (p ≥ 0.207, 0.029 < η_p ² < 0.110) and quality of life-related to the affected knee at baseline (p = 0.398, η_p ² = 0.066).

Post hoc-tests for the KOOS related to symptoms (Figure 3A) revealed a significant improvement only in the BFR-group during (45.0 ± 5.4 to 54.2 ± 3.9, p < 0.001) and after the prehabilitation period (to 60.8 ± 3.7, p < 0.001) with a significant higher value compared to other groups at Pre-OP (CON: 51.7 ± 4.7; AC-group: 48.6 ± 9.5, p ≤ 0.01). No difference was observed between CON and AC-group (p = 0.893). At 3m-Post-OP, the BFR-group indicated a significant lower KOOS related to symptoms (47.2 ± 3.0) compared to 3w-Prehab- (p = 0.021) and Pre-OP-level (p < 0.001), but still similar level to other groups (CON: 45.0 ± 4.5; AC-group: 46.1 ± 7.7). At 6m-Post-OP, the KOOS related to symptoms increased in all groups (CON: 63.4 ± 5.1; BFR-group: 67.1 ± 3.6; AC-group: 65.2 ± 9.0) with a significantly higher value compared to previous level (p ≤ 0.05).

The KOOS related to pain (Figure 3B) in the CON was significantly higher at baseline compared to other groups (CON: 48.7 ± 3.9; BFR-group: 41.1 ± 4.3; AC-group: 44.2 ± 3.8, p ≤ 0.048), but there was no difference between both training groups (p = 0.299). The BFR-group significantly improved the KOOS related to pain during (to 52.8 ± 4.1, p < 0.001) and after the prehabilitation period (to 57.6 ± 3.4, p < 0.001), while no changes occurred in the CON (at 3w-Prehab: 48.0 ± 4.1; at Pre-OP: 47.6 ± 5.4, p = 1.00). The AC-group showed an improvement in the KOOS related to pain only after the prehabilitation (i.e., at Pre-OP to 49.8 ± 5.2) with a significant difference to baseline (p = 0.033). Consequently, the BFR-group exhibited significant higher KOOS related to pain compared to other groups both at 3w-Prehab (p ≤ 0.048) and at Pre-OP (p ≤ 0.003). During the post-operative period, all groups indicated further improvements in the KOOS related to pain with a significant higher value to all pre-operative time points (at 3m-Post-OP: 60.8 ± 6.9 vs. 67.8 ± 3.5 vs. 61.7 ± 6.8; at 6m-Post-OP: 70.0 ± 4.7 vs. 76.2 ± 3.6 vs. 71.1 ± 7.9 in CON, BFR-, AC-group, respectively). In addition, the BFR-group indicated a significant higher KOOS related to pain during the post-operative period compared to the CON (p ≤ 0.050 at both 3m- and 6m-Post-OP), whereas the AC-group did not differ to any other groups (p ≥ 0.10).

The analysis on the KOOS related to the activities of daily living (Figure 3C) revealed a continuous improvement in both training groups during the overall study period (BFR-group: 45.5 ± 4.2 to 53.5 ± 6.1 to 57.9 ± 3.7 to 63.7 ± 5.1 to 71.9 ± 3.1; AC-group: 46.7 ± 3.5 to 50.5 ± 3.8 to 52.8 ± 3.4 to 59.0 ± 3.5 to 67.8 ± 3.1; p ≤ 0.047), while the CON showed a significant improvement only during the post-operative period (49.2 ± 4.2 to 49.0 ± 4.4 to 49.4 ± 6.0 to 58.7 ± 4.5 to 65.6 ± 4.6; p < 0.001). Moreover, the BFR-group indicated significant higher KOOS related to activities of daily living compared to other groups at Pre-OP (p ≤ 0.05), 3m- (p ≤ 0.05), and 6m-Post-OP (p ≤ 0.05), whereas no differences were observed between CON and AC-group (p ≥ 0.325).

Regarding the functionality in sports and recovery (Figure 3D), the CON showed a higher KOOS compared to BFR-group at the baseline (24.0 ± 3.2 vs. 19.5 ± 3.7; p = 0.024), but the AC-group indicated no difference to any other groups (20.5 ± 3.7; p ≥ 0.105). Only in the BFR-group, the functionality in sports and recovery already improved at 3w- Prehab (to 27.5 ± 3.5; p = 0.001) with a significant higher value compared to AC-group (22.5 ± 3.5; p = 0.024). At the Pre-OP, both training groups demonstrated higher sports and recovery functionality than the baseline level (BFR-group: to 29.5 ± 3.7; AC-group: to 24.0 ± 3.9; p ≤ 0.013). No changes occurred in the CON during (to 25.0 ± 4.1) and after (to 25.0 ± 4.1) the prehabilitation period. At the 3m-Post-OP, only in the BFR-group, the functionality in sports and recovery was higher compared to the baseline level (to 31.0 ± 5.7; p < 0.001). At the 6m-Post-OP, the functionality in sports and recovery in the BFR-group was significant higher compared to each of all other time points (38.5 ± 3.4; p ≤ 0.003), whereas the CON (32.0 ± 6.7) and AC-group (29.5 ± 6.0) demonstrated a significant higher value only compared to the baseline- (p ≤ 0.031) and 3m-Post-OP-level (p ≤ 0.037). After the Pre-OP until 6m-Post-OP, the functionality in sports and recovery was higher in the BFR-group compared to other group (p ≤ 0.05).

The KOOS related to the quality of life-related to the affected knee (Figure 3E) increased in both training groups already at 3w-Prehab (BFR-group: 21.9 ± 3.3 to 33.8 ± 4.4; AC-group: 24.5 ± 5.3 to 28.5 ± 5.5; p ≤ 0.017), and thus the BFR-group indicated higher value compared to the CON (25.5 ± 8.6; p = 0.024). At Pre-OP, the BFR-group demonstrated a more pronounced improvement in the quality of life-related to the affected knee (to 40.0 ± 3.2; p = 0.011) with a higher value compared to other groups (CON: to 25.6 ± 10.0; AC-group: 30.9 ± 5.2; p ≤ 0.017). During the post-operative period, all groups demonstrated an increased quality of life-related to the affected knee compared to each of other previous time points (CON: to 45.6 ± 9.2 to 54.4 ± 8.99; BFR-group: to 52.5 ± 6.8 to 69.4 ± 4.6; AC-group: to 44.4 ± 4.6 to 55.0 ± 4.0; p ≤ 0.032), whereas the BFR-group still exhibited higher values compared to other groups (p ≤ 0.05) except for 3m-Post-OP in the CON (p = 0.119).