What makes long-term resistance-trained individuals so strong? A comparison of skeletal muscle morphology, architecture, and joint mechanics

106 107 The greater muscular strength of long-term resistance-trained (LTT) individuals is often 108 attributed to hypertrophy but the role of other factors, notably maximum voluntary specific tension 109 (ST), muscle architecture and any differences in joint mechanics (moment arm) have not been 110 documented. The aim of the present study was to examine the musculoskeletal factors that might 111 explain the greater Quadriceps strength and size of LTT vs untrained (UT) individuals. LTT (n = 16, age 112 21.6 ± 2.0 years) had 4.0 ± 0.8 years of systematic knee extensor heavy-resistance training 113 experience, whereas UT (n = 52; age 25.1 ± 2.3 years) had no lower-body resistance training 114 experience for > 18 months. Knee extension dynamometry, T1-weighted magnetic resonance images 115 of the thigh and knee and ultrasonography of the Quadriceps muscle group at 10 locations were 116 used to determine Quadriceps: isometric maximal voluntary torque (MVT), muscle volume (QVOL), 117 patella tendon moment arm (PTMA), pennation angle (QΘP) and fascicle length (QFL), physiological 118 cross-sectional area (QPCSA) and ST. LTT had substantially greater MVT (+60% vs UT, P<0.001) and 119 QVOL (+56%, P<0.001) and QPCSA (+41%, P<0.001) but smaller differences in ST (+9%, P<0.05) and 120 moment arm (+4%, P<0.05), and thus muscle size was the primary explanation for the greater 121 strength of LTT. The greater muscle size (volume) of LTT was primarily attributable to the greater 122 QPCSA (+41%; indicating more sarcomeres in parallel) rather than the more modest difference in FL 123 (+11%; indicating more sarcomeres in series). There was no evidence in the present study for 124 regional hypertrophy after LTT. 125


Introduction
facilitates independence and functional mobility (18, 53) with ageing. Participation in resistance 145 training (RT) is well known to increase strength and therefore is widely recommended on an on-146 going/continuous (i.e. long-term) basis for individuals of all ages as well as numerous patient groups 147 (3, 49, 51, 72). Hence long-term RT individuals are known to be substantially stronger than untrained 148 controls (UT; (9, 56), a functional difference that is often attributed to their larger muscle size (i.e. 149 greater volume or cross-sectional area [CSA] due to hypertrophy). However, the role of other 150 morphological and mechanical differences that may also influence strength, notably specific tension 151 (i.e. force per unit area), muscle architecture and joint moment arm have been poorly documented. 152 In fact, long-term systematic RT (i.e. multiple years) has been shown to result in substantially 153 greater muscle size compared to untrained controls (+70-76% greater Biceps Brachii anatomical CSA 154 [ACSA; (9, 52)]; +85% greater Quadriceps volume; (38)), but whether an increase in muscle size is 155 accompanied by similar, smaller or no changes in maximum voluntary specific tension (ST) remains 156 unknown. Furthermore, the extent to which increases in overall muscle size (volume) after long-157 term RT are due to increases in either sarcomeres in parallel (i.e. increased physiological CSA; PCSA) 158 and/or in series (i.e. fibre/fascicle length) has not been examined. Finally, the extent of region-159 specific hypertrophy, both between constituent muscles and along their length, after long-term RT 160 remains to be elucidated. Therefore, a rigorous assessment of muscle size (ACSA, PCSA and Volume), 161 ST  RT does exist it would be expected to be pronounced in long-term RT individuals that exhibit 187 substantially larger muscles, however, this has not been examined. 188 The structural remodelling of muscle morphology in response to RT can be observed by 189 examining muscle architecture, specifically Pennation Angle (Θ P ) and Fascicle Length (F L ). Numerous 190 studies have found Θ P to: increase after RT (1, 10, 57), and after RT interventions (12, 15, 16, 67); or 191 be higher in resistance-trained vs. untrained individuals on a cross-sectional basis (39, 47, 70). An 192 increase in Θ P may facilitate an increase in the contractile material attaching to the 193 tendon/aponeurosis, independent of any change in ACSA. However, the increase Θ P also has a 194 negative effect on force generating capacity by reducing the transmission of force between the 195 fibres and the tendon/aponeurosis (8). These contrary effects of Θ P on the force generating capacity 196 of the muscle are theoretically best reflected by effective PCSA (Q EFF PCSA) that accounts for both the 197 number of sarcomeres in parallel and force transmission to the aponeurosis/tendon. 198

199
The changes in F L after short-term RT remain controversial with reports of no change in F L 200 (isometric RT: (6); or conventional isoinertial RT [lifting and lowering]: (14, 26, 29, 30, 80) and 201 increased F L (isometric: (65); isoinertial: (7, 78). One study of long-term heavy RT individuals (RT 202 history: 12.4 ± 5.4 yrs [mean ± SD]) observed no difference in F L compared to controls (70). The 203 controversy surrounding the architectural changes, especially F L , after RT could in part be due to 204 heterogenous architectural changes throughout the muscle after RT (14, 59) in a similar manner, and 205 potentially linked to region specific hypertrophy. Therefore, comprehensive architectural 206 measurements throughout the muscle may clarify whether F L changes after long-term RT. 207 torque production (17, 77) in untrained controls (74, 79). For some muscles it has been suggested 210 that muscle growth after RT may cause an advantageous increase in the moment arm by positioning 211 the tendon further from the joint centre (79). Although the anatomy of the patella and patella 212 tendon wrapping around the distal femur, mean that this may be unlikely for the Quadriceps, the 213 contribution of any differences in moment arm to the strength in long term RT individuals compared 214 to untrained individuals is unknown. 215

216
The aim of the present study was to determine the factors that explain the greater strength 217 and larger muscle size (volume) of long-term RT individuals (LTT) vs untrained (UT) individuals. This 218 involved a comprehensive comparison of Quadriceps morphology and mechanics, specifically: 219 measures of muscle size (Q VOL , QACSA MAX , QPCSA, Q EFF PCSA) and regional hypertrophy/muscle mass 220 distribution (between and along the Quadriceps muscles) with MRI, agonist muscle ST (accounting 221 for antagonist co-activation, moment arm and Q EFF PCSA), muscle architecture (F L and Θ P ) at 10 sites 222 throughout the Quadriceps with ultrasound imaging, and moment arm also assessed with MRI. It 223 was hypothesised that: (i) the anticipated greater strength of LTT vs UT would be due to both their 224 greater muscle size (Q VOL , QACSA MAX , QPCSA, Q EFF PCSA ) and higher ST; (ii) the greater muscle volume 225 of LTT would be due to higher PCSA rather than greater F L (i.e. sarcomeres in parallel not in series); 226 and (iii) there would be marked regional hypertrophy between and along constituent Quadriceps 227 and follow-up oral discussion) systematic, progressive heavy RT of the quadriceps ~3 x/wk for ≥3 244 years (mean ± SD, 4 ± 1 years; range, 3-5 years), involving completion of several knee extensor 245 exercises (e.g. squat, lunge, step-up, knee extension and leg press) within an individual session, and 246 with the primary aim of developing maximum strength. The RT of this group had not been 247 experimentally supervised although some of these participants had received variable coaching 248 (technique and programming) support. Participation in weight classified or predominantly 249 endurance sports was an exclusion criteria to avoid these potential confounders of morphological 250 adaptation. Of the LTT group, resistance training was the only systematic physical activity of 50% 251 (n=8), 38% (n=6) were national level rugby union players, with the remaining 12% (n=2) competing in 252 powerlifting/body building. Use of androgenic-anabolic steroids was an exclusion criterion for all 253 participants. Many individuals in the LTT group reported regular use of nutritional supplements (e.g. 254 whey protein and creatine).  The analogue force signal was amplified (x370; A50 amplifier, Force Logic UK, Berkshire, UK) 294 and sampled at 2,000 Hz using an A/D converter (Micro 1401; CED, Cambridge, UK) and recorded 295 with Spike 2 computer software (CED). In offline analysis, force signals were low-pass filtered at 500 296 Hz using a fourth order zero-lag Butterworth filter (54), gravity corrected by subtracting baseline 297 force, and multiplied by lever length, the distance from the knee joint space to the centre of the 298 ankle strap, to calculate torque. 299 300 Surface electromyography (EMG) of the hamstring muscles (Biceps Femoris Long Head and 301 Semitendinosus) was recorded using a wireless EMG system (Trigno; Delsys Inc., Boston, MA). Skin 302 preparation (shaving, abrading, and cleansing with 70% ethanol) was conducted before single 303 differential Trigno Standard EMG sensors (Delsys Inc., Boston, MA; fixed 1-cm interelectrode 304 distance) were placed on the Biceps femoris long head and Semitendinosus at 45% of thigh length 305 above the popliteal fossa. Sensors were placed parallel to the presumed orientation of the 306 underlying fibres. EMG signals were amplified at source (x300; 20 to 450-Hz bandwidth) before 307 further amplification (overall effective gain, x909), and sampled at 2,000 Hz via the same A/D 308 converter and computer software as the force signal, to enable data synchronization. In offline 309 analysis, EMG signals were corrected for the 48-ms delay inherent to the Trigno EMG system. has demonstrated a mean within-participant coefficient of variation for repeat Quadriceps muscle 363 volume measurements using the same protocol 12 weeks apart with a control group to be 1.7% (11). 364 Inter-and intra-rater reliability for Q VOL calculated from the repeated analysis of five MRI scans was 365 1.2 and 0.4%, respectively.  (pairwise ANOVA contrasting only two muscles) was also performed. Effect Size (ES) for absolute 513 difference data was calculated as previously detailed for between-subject study designs (50) and 514 classified as follows: <0.20 = "trivial," 0.20-0.49= "small," 0.50-0.79 = "moderate," or ≥ 0.80= 515 "large." P values were corrected for multiple tests using the Benjamini-Hochberg procedure (13) 516 with a false detection rate of 5%, and significance was defined as adjusted P<0.05. For the whole 517 cohort (i.e. data pooled from both LTT and UT groups, n=68) the relationships between 518 musculoskeletal variables and MVT were first assessed with independent Pearson's product moment 519 correlations, and then stepwise multiple regression analysis was performed, with only the significant 520  Table  550 2), and mean F L of each individual muscle was longer (VM: +12%, ES=0.7; VL: +13%, ES=1.0; and RF: 551 +12%, ES=0.8; all P<0.05) or showed a tendency to be longer (VI: +7%; P=0.06, ES=0.8) for LTT than 552 UT. The outcome of the ANOVA revealed a constituent muscle (VL, VM, VI, RF) x group (LTT, UT) 553 interaction effect (i.e. bigger differences between groups for some muscles than others; P=0.03), 554 and post-hoc analysis showed larger differences between UT and LTT in the VM, VL and RF 555 compared to VI (pairwise ANOVA with only two muscles; group x muscle interaction; All P≤0.008). 556 Considering the specific measurement sites, 6 out of 10 sites showed greater F L of LTT vs UT (VM PRX , 557 VI PRX, VI MID , RF MID , VL DIS and VL PRX sites; all P<0.001), with a tendency to be longer for RF PRX (P=0.06) 558 and no differences at the remaining 3 measurement sites (all P>0.15; Figure 4A). 559 560 QΘ P was 13% greater in LTT than UT (P<0.001, ES=0.7; Table 2 Tendon force was 54% greater in LTT than UT (5576 ± 905 N vs 8564 ± 1410 N; P<0.001, ES= 2.6). 574 There was 8% greater ST of the Quadriceps in LTT than UT (33.3 ± 4.5 N.cm 2 vs 36.1 ± 5.3 N.cm 2; 575

Factors that explain the greater strength and muscle mass (volume) of Long-term RT individuals. 579 580
The difference in strength between LTT and UT (+60%) in comparison to the differences 581 between the groups in a range of underpinning musculoskeletal variables, specifically those 582 variables that were each significantly greater in LTT than UT, are shown in Figure 5. Of the 583 musculoskeletal variables, the largest differences were in the muscle size indices (Q VOL +56%; 584 QACSA MAX +50%) which therefore provide the primary explanation for the greater strength of LTT. 585 This greater muscle size of LTT in combination with a more modest difference in QΘ P (+12%) resulted 586 in a difference in Q EFF PCSA (+40%), which alongside other smaller contributions from ST (+8%) and 587 moment arm (+4%) appears to explain the strength difference. The greater muscle volume of LTT vs 588 UT (Q VOL +56%) appeared to be primarily due to increased QPCSA (+41%) with a much smaller 589 contribution of QF L (+11%; Figure 5). Bivariate correlations for the whole cohort (i.e. both groups, 590 n=68) were found between all musculoskeletal variables and MVT (Q VOL r= 0.90 ( Figure 6 was accompanied by both a greater quantity of skeletal muscle and higher ST. However, the 604 differences between LTT vs UT for the indices of muscle size (e.g. ranging from volume +56% to 605 Q EFF PCSA 41%) were substantially larger than was the case for ST (+8%), or in fact PTMA (+4%), and 606 thus muscle size was the primary explanation for the greater strength of LTT. For the second 607 hypothesis the greater Q VOL (+56%) of LTT was due primarily to enhanced QPCSA (41%), indicating 608 more sarcomeres in parallel, although we also found convincing evidence for greater QF L (+11%), and contrary to our third hypothesis, we found no evidence for regional hypertrophy / muscle mass 611 distribution between or along the constituent Quadriceps muscles. 612

613
The difference in MVT of LTT vs UT in the current study was substantial (+60%), but 614 somewhat lower than observed in one previous study (+77%: (70)). The greater MVT of LTT was 615 accompanied by both a greater quantity of skeletal muscle and higher specific tension, although it 616 was clear from the magnitude of the differences that the indices of muscle size (e.g. volume +56%, 617 Q EFF PCSA +41%) were substantially larger than was the case for ST (+8%), or in fact PTMA (+4%), and 618 thus muscle size was the primary explanation for the greater strength of LTT. The importance of 619 muscle volume for strength was reinforced by our regression analysis of the whole cohort that found 620 muscle volume was the only determinant of MVT, alone explaining 81% of the variance in strength. 621 Several other studies have found substantially greater muscle size of long-term resistance-trained 622 participants (70% to 86% (9,37,46,49), but none have previously examined maximum voluntary 623 specific tension to investigate the contribution of force per unit area to the enhanced strength of 624 LTT. 625 626 We found modest differences in specific tension (+8%), even after the average 4 years of 627 regular, heavy RT of LTT. Whilst no previous studies have examined the specific tension of LTT 628 individuals, after short-term (9 weeks) RT maximum voluntary specific tension has been reported to 629 increase by 20% (30), which is clearly somewhat contrary to the more modest 8% difference we have 630 found for LTT vs UT. However, it is notable that Erskine et al., (30) reported average isometric 631 strength gains ~2-fold greater than we have found (31% vs 11.5-18.2%, (31, 32)) with almost 632 identical training regimes and the same number of training sessions, and this discrepancy likely 633 explains the large increase in specific tension they have reported. Nonetheless, numerous short-634 term RT studies have shown greater increases in strength/force than cross-sectional area, indicating 635 an increase in the specific tension (23, 27, 29, 42, 45, 61, 69, 83). Increased specific tension could be 636 attributable to changes in neuromuscular activation (e.g. increased agonist activation (10, 60)) or an 637 increase in the intrinsic contractile specific tension, perhaps due to a shift in muscle fibre phenotype 638 (20) or alterations in muscle architecture (24). Moreover, the modest difference we have found in 639 specific tension after LTT suggests that increases in specific tension that occur with RT may be 640 relatively limited, and thus the underpinning mechanisms for increased maximum voluntary specific 641 tension (i.e. increased agonist neuromuscular activation or intrinsic contractile specific tension) are 642 also relatively small. 643 first report to quantify the contribution of these different aspects of muscle morphology to the 647 enhanced muscle mass of substantially hypertrophied human muscle, and it is clear that muscle 648 growth primarily occurs due to an increase in the contractile material arranged in parallel with a 649 smaller contribution from increased sarcomeres in series. To provide a comprehensive assessment 650 of Quadriceps muscle architecture we measured Θ P and F L at 10 sites within the Quadriceps, which 651 revealed LTT to have a greater QΘ P (+13%) and QF L (11%) than UT. A greater QΘ P facilitates the 652 attachment of more contractile material, and thus the application of more force, to the 653 tendon/aponeurosis (i.e. as reflected by PCSA; (40, 45, 47, 61)), independently from any increase in 654 muscle ACSA or volume, although force transmission to the tendon is increasingly compromised 655 (according to the cosine of Θ P ). Overall a greater QΘ P is thought to be beneficial for isometric force 656 +34%: (39)), which are clearly a larger difference than we found in the present study (QΘ P : +11%). 660 This contrast may indicate an anatomical specificity to muscle architectural changes after RT or site-661 specific differences. Furthermore, the findings of the present study are surprisingly similar to the 662 increases in Θ P observed following short-term lower body RT (2, 10, 26, 35); perhaps suggesting that 663 changes in lower body Θ P may not continue to adapt with prolonged RT and could predominantly 664 occur in the early phase of a training program (i.e. first 3 months). 665

666
The possibility of F L increases after RT, largely based on short-term RT studies, has been 667 controversial (7, 16, 26, 29, 30, 64, 78, 82). Using architecture measurements at 10 sites throughout 668 the Quadriceps we found the LTT group to have an 11% greater QF L compared to UT. One previous 669 study of LTT vs UT reported no differences between their groups (39), however they assessed F L at 670 only one site, equivalent to the VL MID site of our experiment, where we also observed no differences 671 between LTT and UT ( Figure 4A). In contrast, we found a clear difference for 3 out of 4 of the 672 individual muscles (VM, VL, and RF) a tendency for a difference in the fourth (VI), and over the whole 673 muscle group QF L showed a highly significant difference with a large effect size (+11%, P<0.01 ES 674 1.2). We also found quantitative evidence for a training group (LTT vs UT) by constituent muscle 675 interaction for F L , demonstrating inhomogeneous adaptations to LTT. Thus, it seems likely that the 676 regional variability in F L changes, the error associated with a single measurement site, the 677 assessment at 10 sites throughout the Quadriceps muscle group indicates that QF L does increase 680 with prolonged RT. Interestingly, based on geometric modelling it has recently been argued that 681 relatively modest changes in F L can have disproportionately large effects on ACSA and muscle 682 volume (46). In essence, longer (extended) fascicles due to the addition of sarcomeres in parallel 683 appears to result in a disproportionately larger increases of sarcomeres in parallel and therefore 684 could be a key explanation for the differences in muscle size (ACSA, PCSA and volume) we have 685 observed. 686 Whilst Θ P did not show such strong evidence for inhomogeneous adaptations to LTT (no 687 training group x muscle interaction effect) there were a range of differences when comparing the 688 four constituent muscles (Θ P 8-15%; F L 6-13%). Therefore, this study further highlights the need for 689 multiple sites to comprehensively quantify architectural differences or changes after training as 690 single sites may be difficult to replicate (36) and as seen in the present study and others, a single site 691 measurement similar to VL MID is not reflective of overall architecture differences across the 692 Quadriceps muscle group following RT (26, 35). 693

694
Despite the 56% greater muscle volume of LTT vs UT we found no evidence for regional 695 hypertrophy either between the constituent Quadriceps muscles or along their length. Previous 696 short-term RT studies, documenting relatively limited hypertrophy, have however, repeatedly 697 reported non-uniform regional hypertrophy, both between and along the individual Quadriceps 698 muscles, although curiously the pattern of regional hypertrophy has been surprisingly diverse (i.e. 699 which muscles and locations had the greatest hypertrophy (26, 35, 76, 37, 43, 44, 57, 58, 61, 69, 700 75)). In the current study, we scanned the entire length of the thigh to accurately identify the ends 701 of the bone and subsequently define the precise position of each of a large number of axial images 702 (slices per muscle: VM, 23-26; VI, 24-27; VL, 24-27; RF, 23-26) relative to those absolute landmarks in 703 order to carefully quantify regional differences in muscle size. In addition, we recently found a mean 704 within-participant coefficient of variation for repeat Quadriceps muscle volume measurements using 705 the same protocol 12 weeks apart with a control group to be 1.7%, indicating the reliability of our 706 measurements (11). In contrast, previous studies typically used a small number of slices and 707 positioned slices based on relatively imprecise surface anatomical measurements. Therefore, 708 previous reports of regional hypertrophy may have been confounded by the inconsistent location of 709 the images. Alternatively, as the LTT individuals in the current study had been doing a range of 710 different training practices it is conceivable that this may have resulted in diverse individual 711 along the femur (Figure 3) were no more variable for LTT than UNT groups. In summary, given the 715 careful methods and large difference in muscle volume in the current study without any evidence for 716 regional hypertrophy it seems likely that this phenomenon may have been overestimated by 717 previous studies. 718 In addition to morphological changes in the muscle, joint mechanical properties such as 719 PTMA may make a small contribution to maximal torque production (17, 77). In the present study, 720 PTMA was 5% greater in LTT compared to UT. In other muscle groups it has been suggested that 721 muscle hypertrophy may result in biomechanically advantageous increases in leverage of muscular 722 force application (5, 73, 74, 79). However, for the Quadriceps the anatomy of the patella and patella 723 tendon wrapping around the distal femur, mean that this is unlikely to be the case. In addition, when 724 PTMA was normalized to height there was no difference between the groups indicating that the 4% 725 greater height of LTT group was in large part responsible for their greater PTMA. 726

727
There are a number of limitations within the current study that should be recognized. Whilst 728 the current cross-sectional study design provided a pragmatic approach to examining the substantial 729 adaptations that occur after LTT. However, due to the cross-sectional nature of the current study 730 and the extensive, retrospective RT background (mean 4 years RT) of these participants we have 731 relatively limited information regarding their exact training (e.g. precise loads, types of contractions, 732 periodization). Nonetheless these participants all had the primary goal of increasing maximum 733 strength, were demonstrably stronger than controls (+60%) and we excluded participants involved in 734 activities (e.g. weight category and endurance sports) that might compromise morphological 735 adaptations to RT. A repeated measurement design on the same participants before, potentially 736 during, and after a prolonged period of RT is clearly a stronger design. Although this approach would 737 be practically challenging, there are very few supervised RT studies of ≥6 months duration, it would 738 facilitate an in-depth examination of the time course of adaptations to prolonged RT and could be 739 informative for a number of the measures investigated in the current experiment (e.g. specific 740 tension, architecture, regional hypertrophy). The acquisition of clear T1 MR images along the whole 741 thigh (~25 minutes) is not compatible with measurements during contraction, and in our experience, 742 it is also challenging to record clear ultrasound images of all the constituent muscles during MVCs 743 (55). Thus, the imaging measurements of muscle size, architecture, and moment arm within the 744 current experiment were made at rest in order to facilitate precise measurements. In addition, due 745 maximum contraction (55). Whilst we have recently found LTT to have a stiffer patella tendon 750 compared to UT, the greater strength of this group appears to produce similar muscle shortening, 751 and thus presumably architectural changes, at MVC (56). Therefore, we are not aware of any 752 systematic effects that might interact with these potential confounders and influence the 753 comparison of LTT and UT groups within the current study. 754 Finally, the use of B-mode ultrasound presents a number of methodological issues when 755 quantifying muscle architecture in vivo (For a review see (36)). In the present study by using a 756 relatively long probe (92 mm vs commonly used 40-60 mm) we were able to minimize the need for 757 extrapolation of fascicle trajectory beyond the recorded image (typically <10% of the measured F L 758 was extrapolated). Architecture measurements were also performed in the knee isometric 759 dynamometer with a knee angle of 115 O (i.e. the same knee joint angle as the strength 760 measurements), and this longer muscle length relative to rest explains why F L was longer in the 761 present study than in some previous reports (35, 71). However, we are conscious that ultrasound 762 images are a 2-D representation of a complex 3-D structure and recommend that future work utilize 763 more sophisticated 3-D techniques (e.g. diffusion tensor MRI). 764

765
In conclusion, the present study demonstrates that the larger Quadriceps strength of LTT 766 individuals was primarily due to greater muscle size with smaller differences in specific tension and 767 moment arm, and thus muscle size was the primary explanation for the greater strength of LTT. The 768 greater muscle volume (+56%) of LTT was due primarily to enhanced PCSA (41%), indicating more 769 sarcomeres in parallel, although we also found convincing evidence for greater QF L (+11%), 770 indicating a modest difference in sarcomeres in series. Finally, there was no evidence for regional 771 hypertrophy either between or along the Quadriceps muscles after long-term RT