Physiological demands of running at 2-hour marathon race pace.

The requirements of running a 2 hour marathon have been extensively debated but the actual physiological demands of running at ~21.1 km/h have never been reported. We therefore conducted laboratory-based physiological evaluations and measured running economy (O2 cost) while running outdoors at ~21.1 km/h, in world-class distance runners as part of Nike's 'Breaking 2' marathon project. On separate days, 16 male distance runners (age, 29 ± 4 years; height, 1.72 ± 0.04 m; mass, 58.9 ± 3.3 kg) completed an incremental treadmill test for the assessment of V̇O2peak, O2 cost of submaximal running, lactate threshold and lactate turn-point, and a track test during which they ran continuously at 21.1 km/h. The laboratory-determined V̇O2peak was 71.0 ± 5.7 ml/kg/min with lactate threshold and lactate turn-point occurring at 18.9 ± 0.4 and 20.2 ± 0.6 km/h, corresponding to 83 ± 5 % and 92 ± 3 % V̇O2peak, respectively. Seven athletes were able to attain a steady-state V̇O2 when running outdoors at 21.1 km/h. The mean O2 cost for these athletes was 191 ± 19 ml/kg/km such that running at 21.1 km/h required an absolute V̇O2 of ~4.0 L/min and represented 94 ± 3 % V̇O2peak. We report novel data on the O2 cost of running outdoors at 21.1 km/h, which enables better modelling of possible marathon performances by elite athletes. Using the value for O2 cost measured in this study, a sub-2 hour marathon would require a 59 kg runner to sustain a V̇O2 of approximately 4.0 L/min or 67 ml/kg/min.


Introduction
There is considerable scientific and public interest in the requirements of running a 26.2 mile 46 (42.195 km) marathon in less than 2 hours (36,37,61), as was recently accomplished by 47 Eliud Kipchoge of Kenya in an exhibition event in Vienna. Traditional physiological factors 48 that have been proposed to exert an important influence in this regard include the runner's 49 maximal oxygen (O 2 ) uptake (V O 2 max), the fraction of the V O 2 max that can be sustained 50 during the marathon which is, in turn, related to the lactate threshold (LT) or critical speed 51 (CS), and the O 2 cost of submaximal running (i.e., running economy in units of ml of 52 O 2 /kg/km), (28,35,37). Other important 'external' factors include the course profile, 53 environmental conditions (altitude, ambient temperature, relative humidity and wind speed), 54 pacing strategy, drafting, pre-and in-race nutrition, and footwear and apparel (9,23,24,36). 55 To run a marathon in under 2 hours, an elite distance runner must be able to sustain a 56 metabolic steady-state while running at just over 13.1 mph (i.e. ~4 minutes and 34 seconds 57 per mile) or 21.1 km/h (i.e. ~2 minutes and 50 seconds per km). To our knowledge, the O 2 58 cost of running outdoors at sea level at ~21.1 km/h has never been reported. This is 59 understandable given that there are presumably very few athletes in the world capable of 60 running at this speed in a metabolic steady-state, which is a necessary condition for the valid 61 assessment of running economy (48). Estimating the O 2 cost of running at 21.1 km/h by 62 extrapolating the V O 2 -running speed relationship established at lower speeds (typically 15-19 63 km/h) in less highly-trained athletes might not be appropriate, especially given the difference 64 in air resistance which is evident between treadmill and outdoor running at higher speeds (32, (36). 69 The purpose of this study was to investigate the O 2 cost and physiological demand (i.e.,  set at a 1% gradient (32). Prior to testing, a resting blood sample was drawn from a fingertip 120 for the assessment of baseline blood lactate concentration. The athlete was then fitted with a 121 telemetric heart rate (HR) monitor (Polar S610, Kempele, Finland (Exeter) or Wahoo TickrX, 122 Atlanta, GA (Beaverton)) and allowed to perform his individual warm-up regimen including it has been reported that they provide measurements of V O 2 that are reliable and valid 131 relative to the gold standard Douglas bag method (1,25,51).

132
The starting speed for the treadmill test was 17 km/h. Each stage was 3 min in duration and 133 the belt speed was increased by 1 km/h until 19 km/h and 0.5 km/h thereafter until the athlete 134 reached volitional exhaustion (i.e., when he could not complete a stage or declined the 135 opportunity to start a new one). For the first stage of the test, the belt speed was increased to 136 17 km/h and following the command of "3-2-1-GO" the athletes commenced running, having 137 previously stood still for 60 s with their feet astride the moving treadmill belt. A fingertip       cost of running for the athletes across the full range of speeds is shown in Figure 2B. The 246 mean O 2 cost of running was similar across the speeds studied but, at each speed, there was 247 considerable inter-individual variability (with a range of ~170-220 ml/kg/km; Figure 2B).

248
The individual blood [lactate]-running speed relationships are presented in Figure 3A, with 249 the response of a representative athlete highlighted in Figure 3B and the group mean ± SD LT 250 and LTP shown in Figure 3C. The group mean LT occurred at 18.9 ± 0.4 km/h, which 251 corresponded to 83 ± 5 % V O 2 peak, 166 ± 9 b/min (87 ± 5 % HR max) and a blood [lactate] 252 of 2.2 ± 0.8 mM. The group mean LTP occurred at 20.2 ± 0.6 km/h which corresponded to 92 253 ± 3 % V O 2 peak, 181 ± 8 b/min (94 ± 2 % HR max) and a blood [lactate] of 4.6 ± 1.3 mM. 254 We took the opportunity to evaluate V O 2 kinetics during the first stage of the incremental test 255 (i.e. step test from standing at rest to running at 17 km/h; see Figure 1). 1.10 ± 0.12 kcal/kg/km. In the second part of the test, as instructed, the athletes maintained a 263 speed of 21.0 ± 0.2 km/h. At this speed, the group mean V O 2 was 4.11 ± 0.37 L/min (70 ± 6 264 ml/kg/min; 95 ± 3 % V O 2 peak) and the group mean O 2 cost was 191 ± 19 ml/kg/min (P<0.01 265 compared to the lower speed). Nine athletes were able to accelerate in the final lap to achieve 266 a speed of 22.5 ± 0.8 km/h, V O 2 peak of 4.20 ± 0.28 L/min (71.9 ± 6.1 ml/kg/min) and HR 267 max of 185 ± 10 b/min. 268 It was notable that not all athletes were able to achieve a V O 2 steady-state when running 269 over-ground at ~21 km/h. The V O 2 profiles of a representative athlete from the group that 270 was able to achieve a steady-state (n = 7) and a representative athlete from the group that was 271 not able to achieve a steady-state (n = 9) are shown in Figure 4A and 4B, respectively. In the represented a slightly smaller fraction of V O 2 peak in the group that was able to reach a 279 steady-state (94 ± 3 %) compared to the group that could not reach a steady-state (97 ± 9 %). Group mean biomechanical characteristics are presented in Table 2. Four of the ten runners 289 presented ground reaction force-time histories that displayed an impact peak typical of a  However, when the sustainable V O 2 was assumed to be 96% LTP, as has been proposed 305 previously (34), the predicted marathon time was more realistic for the cohort (2:08:31 ±   (29). It is notable that the LT and LTP occurred at high 358 fractions of the athletes' V O 2 peak (83 ± 5 % and 92 ± 3 %, respectively). It is also notable 359 that the speed required to run a 2 hour marathon (21.1 km/h) exceeded the group mean LTP 360 speed, clearly indicating that not all of the elite athletes evaluated were capable of sustaining 361 the necessary speed without experiencing a progressive accumulation of lactate over time.

362
The LTP approximates the CS (60) and therefore delineates the heavy-intensity exercise 363 domain, within which steady-state physiological responses can be achieved, from the severe-364 intensity exercise domain (60). In the severe-intensity domain, a metabolic steady-state 365 cannot be achieved, fatigue develops more rapidly and exercise tolerance is limited to less 366 than approximately 30 minutes (60, 63). It appears that elite athletes run marathons at a mean 367 speed that resides in the heavy-intensity domain, that is, above LT but below CS (33,34). 368 Indeed, it has been calculated that elite distance runners are able to sustain a marathon race 369 speed at approximately 96 % of CS when the latter is estimated using personal best 370 performance times established over shorter race distances (5, 34). Therefore, for a 2 hour 371 marathon to be achievable, it is necessary for CS to occur at a minimum of 22 km/h. Because 372 CS occurs at approximately 90% V O 2 peak in elite endurance athletes (3, 4, personal 373 observations), this would indicate that these athletes might sustain a high fraction of V O 2 peak 374 (~86-90%) during a 2 hour marathon race. This coheres with estimates derived from 375 measurements made at altitude in elite Kenyan runners (62) and also with a recent report that 376 marathon race speed required 91% V O 2 peak in a masters' world marathon record holder (40).

377
Consistent with our previous analysis (34), when we calculated possible best marathon times 378 for the athletes in the present study, the most realistic estimate (i.e., the one closest to the 379 athletes' personal record times) was derived when the highest sustainable V O 2 was assumed 380 to occur at 96% of LTP (or approximately 88% V O 2 peak).

381
The type of training required to elicit a high CS and to enable a high fraction of V O 2 peak to 382 be sustained during a marathon is not entirely clear (30). However, it is known that critical 383 power (CP, which is analogous to CS) in cycling is related to a high proportion of highly-384 oxidative, fatigue resistant type I muscle fibers (63) and to muscle capillarity (44). In this 385 light, it may be pertinent to note that elite marathon runners complete a relatively high   Fast V O 2 kinetics, per se, might not be considered to be especially relevant to marathon 404 performance because the duration for which the athlete will be in an initial O 2 deficit is very 405 small relative to the event duration. However, it has been reported the phase II time constant 406 is significantly correlated with CP during cycle exercise (49), suggesting that the two 407 variables might be related through some common physiological mechanism such as skeletal 408 muscle oxidative capacity (64). In this light, it is intriguing that we observed a significant and O 2 cost that, in combination, permit a speed of ≥ 21.1 km/h to be sustained for the 456 marathon distance are even more rare.

458
There has been increasing interest in running with an anterior (non-rearfoot) foot strike in 459 recent years, despite minimal evidence to support the proposed benefits over running with a 460 rearfoot strike (20). The present study supports data from the 2017 IAAF World 461 Championships (21), which showed that on average 60% of men's marathon runners 462 displayed a rearfoot strike, including the top four finishers.

463
The mean ground contact time of 0.16 s, measured in the present study, is similar to values 464 previously reported in elite runners at running speeds of 19.5 km/h (39) and 20 km/h (58).

465
The mean ground contact time tends to be shorter in elite compared to sub-elite runners. For proposed that some of these characteristics may be related to running economy and running 479 performance (18,42,45). While there were no significant correlations between 480 anthropometric variables and running economy in the present study, it is important to note 481 that the athletes were relatively homogenous in their physical and physiological 482 characteristics and, therefore, the lack of correlation should not be interpreted to imply that 483 those variables are not important determinants of running economy.

485
There are, of course, many other factors that can influence marathon performance in addition 486 to athlete anthropometry and physiology. These include the psychological characteristics of 487 the athlete and sound biomechanics although this latter aspect may be captured, to a large 488 extent, in measurements of running economy (18,23). Due to the high absolute metabolic 489 rate that must be sustained and the related heat production, thermoregulation is another 490 important consideration and environmental factors such as ambient temperature, relative 491 humidity, radiant heat and wind speed can therefore significantly influence marathon 492 performance (43). It is also necessary to recognise that physiological variables, such as would not. To be sustainable for the requisite time, it is necessary for this metabolic rate to be 537 lower than the 'critical metabolic rate' associated with CS. Moreover, the higher the 538 V O 2 peak, the smaller the fraction of V O 2 peak that 4.0 L/min represents: for example, a 539 V O 2 peak of ~80 ml/kg/min in a 59 kg runner gives a fractional utilisation of 85% which 540 seems physiologically reasonable. It is essential to recognise that the traditional physiological 541 variables we measured in this study should be considered in combination rather than in 542 isolation (35). The absolute V O 2 that is sustainable for 2 hours is the critical metabolic factor, 543 with the O 2 cost of running at race pace, and its resilience to fatigue development over time 544 (7,11), being instrumental in translating the metabolic output into speed over the ground.

545
Given that these factors are likely to have been optimized by genetic predisposition and long-546 term training in today's elite athletes, it would appear that scientific innovations and/or 547 strategies which enable a higher mean oxidative metabolic rate to be sustained and/or 548 enhance running economy will play a significant role in future improvements in marathon 549 performance. This research was funded by a grant from Nike Inc. to AJ and AV (University of Exeter).