Dieting Down to ~10% Body Fat for Women: Contest Prep Study - Deficits, Muscles, Hormones and the Yo-Yo Effect

This would unquestionably be at the lower end of contest BF% levels (avg. 12%) the women in this study achieved,

The number of studies on fitness and bodybuilding competitions is limited. Probably you will remember my previous discussions of the paper/s by Rossow, et al. ("Natural bodybuilding competition preparation and recovery: a 12-month case study." | 2013) Kistler, et al. ("Case Study: Natural Bodybuilding Contest Preparation." International Journal of Sport Nutrition and Exercise Metabolism", 2014), and Robinson et al. ("A nutrition and conditioning intervention for natural bodybuilding contest preparation: case study" | 2015). And yes, three is not just the number of studies that I've covered at the SuppVersity, it's also the number of decently recent studies dabbling with dieting down for a contest in one of the "physique sports".
Speaking of fitness, ... you probably already guessed it: There's study #4 by Juha J. Hulmi, et al. (2017). The study that was conducted by researchers from Finland and Estonia has the telling title "The Effects of Intensive Weight Reduction on Body Composition and Serum Hormones in Female Fitness Competitors" and was, as its authors point out, designed at least partly in response to the "worries about the potential negative consequences of popular fat loss regimens for aesthetic purposes in normal weight females have been surfacing in the media" (Hulmi. 2017).

Since longitudinal studies investigating these kinds of diets are lacking, Hulmi et al. studied the effects of a 4-month fat-loss diet in 50 (initially) normal weight females competing in fitness sports. This means we are dealing with a large-scale observational study, not an experimental study in which parameters such as the daily or weekly caloric deficit, the protein intake or related variables were set by the researchers. Rather than that, the 27 females (27.2 ± 4.1 years | 15 newbies, 12 women who had competed before) who dieted for ~4 months, and their 23 (27.7 ± 3.7 years) peers, who were acting as weight-stable controls, managed their diet and exercise regimen as they saw fit (note: with only 3 dropouts from the diet group and 6 from the control group the adherence was pretty good).

Figure 1: The experimental design of the study. Two representative participants are shown from each group. The pre to mid time period lasted ∼20 weeks during which the participants decreased their energy intake and the amount of exercise, whereas the controls maintained their activity levels and nutrient intake. The mid to post period lasting ∼18 weeks was a recovery period with increased energy intake back toward the baseline levels in the diet participants, whereas the controls maintained their energy intake and exercise levels (Hulmi. 2017).

You can take a glimpse both the design and the results in Figure 1. The subjects' body composition was assessed on separate test with the DEXA test being done on empty (that makes the test more reliable, because it's not going to be tricked by overstocked glycogen levels). The exercise performance tests, on the other hand, took place after the subjects' had a standardized breakfasts, because the dieters would otherwise have underperformed due to their low energy intake in the later phase of the dieting period - a period during which the subjects maintained not identical, but similar resistance training and cardio regimen (quoting from the FT | my emphasis):

• Resistance training: Split routines were used for resistance training by all competitors in the diet group meaning that they focused on single muscle groups per session as is often the case also in bodybuilders (Hackett et al., 2013). The main muscle groups trained included thighs, hamstrings, buttocks, chest, shoulders, arms, upper and lower back, calves, and abdominals. Dividing training into separate body parts per session did not differ significantly throughout the training. At baseline the 3-, 4-, 5-, and 6-split training was used by 3, 10, 13, and 1 of the 27 participants, respectively, while the same numbers were during the diet on average 5, 8, 14, and 0 and during the recovery period 7, 8, 12, and 0. In addition, the competitors also practiced their posing routines. Training sessions lasted between 40 and 90min. 
• Cardio: Aerobic training for the participants was almost uniquely either high-intensity interval training (HIT) with a bicycle, cross trainer or other gym equipment or both HIT and steady-state low to medium intensity aerobics (usually walking/running or with cross trainer). During the competition week the participants did not report doing HIT, but instead lower intensity aerobics. Typical HIT-exercise was 10–25 min in total including high intensity 15–45s intervals with 30–60s of recovery between the sets. Steady state lower intensity aerobics was typically 30–60 min in duration. Part of the females completed their aerobic training mainly together with their resistance exercise workouts while most of the participants completed also separate aerobic workouts, especially during the diet.

As the authors point out, there was a tapering period (learn more) during which total training load is typically slightly decreased and carbohydrate and total energy intakes are increased toward the baseline levels, in the last week of the contest prep.
Speaking of the subjects' energy deficit. The latter was - as it is currently in vogue - achieved almost solely by a drastic reduction in carbohydrate intake while keeping protein intakes high and fat intakes moderate. You can see the result in form of a diet that lacked, during contest prep/dieting, ca. 20% of the habitual energy intake of the subjects in my plot in Figure 2 reduced (see Figure 6 for data on the energy balance, i.e. calories in vs. calories out from exercise = cardio + weights).

Figure 2: Energy intake expressed as protein, carbohydrates and fats rel. to body weight (g/kg | Hulmi. 2017).

With these dietary changes and the increase in aerobic exercises aka "cardio", the subjects achieved body fat reductions of ~12% and a ∼35–50% decrease in fat mass (DXA, bioimpedance, skinfolds, P < 0.001) during the dieting phase.

Table 1: Tabular overview of the exercise levels in the diet and control groups; learn more about METs (Hulmi. 2017).

As you can see in Table 1, the latter, i.e. the aerobic workload did indeed increase significantly, while the subject's maintained their resistance training volume fairly stable [ 4.7 ± 0.7 (diet group) and 3.9 ± 1.9 (controls) METs] - until the tapering week, which may now give you the false impression that volume and/or intensity were reduced during the dieting phase:

• lower body muscles were trained during the diet 1.4 ± 0.5 times per week, and 
• upper body muscle groups were trained 1.1 ± 0.3 times per week, 

as part of the previously hinted at split design. That allowed for 4.9 ± 2.9 extra cardio sessions per week (that's an increase of +27% | P < 0.05) that came mostly from an increased amount of steady state aerobics "in several subjects" (I interpret this as "the majority increased the steady-state, only a few the HIIT volume). During the recovery period, the subjects' cardio frequency dropped down to 2.3 ± 1.9 times per week - both, cardio and resistance training were yet never skipped completely.

Figure 3: Changes in body composition; for fat and lean mass I used the average of all three measuring methods, i.e. DXA, body impedance and skinfold; * p < 0.05 and *** p < 0.001 (Hulmi. 2017)

The fact that the women kept training after the contest is important, when it comes to the interpretation of the actual study results, which include the following observations:

• Figure 4: Changes in hormone levels; surprisingly, the cortisol levels (not shown) did not change - not even non-significantly at any time-point; *–*** is significant (p < 0.05–< 0.001) difference to Pre and #–### is significant (p < 0.05–< 0.001) difference between the groups in the change (Hulmi. 2017).

  a weight reduction of -12% of their initial weight and a loss of 35-50% body fat according to DXA, bioimpedance, and skinfold measurements -- the exact figures are -23.1 ± 5.6 to -12.7 ± 4.0% (DXA), from -19.7 ± 4.2 to -11.6 ± 3.9% (bioimpedance), and from -25.2 ± 3.0 to -18.3 ± 2.7% (skinfolds) -- in this context it may be worth mentioning that the women lost abdominal fat in particular, with the DXA scan indicating a -60% reduction in visceral fat 
• a small decrease in lean mass (bioimpedance and skinfolds) and in vastus lateralis muscle cross-sectional area (ultrasound) were observed in diet (P < 0.05) -- in that, it is certainly interesting that the scientists found that some competitors even managed to gain lean mass during the contest prep
• the total bone mass decreased by −1.3 ± 1.8% (DXA) in the competitors, 
• the subject's leg strength, measured as the isometric maximal strength and explosive strength of their leg extensors remained unchanged during dieting, their bench press performance, on the other hand, declined,
• a sign. changes in the hormonal system with decreased serum concentrations of leptin, triiodothyronine (T3), testosterone (P < 0.001), and estradiol (P < 0.01) coinciding with an increased incidence of menstrual irregularities (P < 0.05), 
• the surprisingly small (and non-significant) changes in mood-related parameters of which only the subjects' vigor showed a significant decline during the pre-contest phase,
• the normalization of all body weight/composition parameters and all hormones except T3 and testosterone which did not fully recover during the 3–4 month recovery period

Whether or not the small reduction of T3 is enough to produce what people refer to as "metabolic damage" is not clear. On the one hand, the small reduction in fT3 you can also see in Figure 5, would suggest that at least some of the subjects' (N=1...27) metabolisms were running slower than before.

Figure 5: Active thyroid hormone (fT3) concentration at baseline (pre), before the contest (mid) and after recovery (post). The number at the x-axis depicts participant numbers ordered based on the pre-value (Hulmi. 2017).

On the other hand, the subjects' body weight did not yo-yo back up to levels sign. above those they had before the contest, one may assume that this was not the case. After all, all women ate just as much as before the contest and did even less, not more cardio. This, in turn, translates into a 16% increase in their energy balance in the recovery (vs. baseline, see Figure 6) phase.
References:

• Dulloo, Abdul G., J. Jacquet, and Lucien Girardier. "Autoregulation of body composition during weight recovery in human: the Minnesota Experiment revisited." International journal of obesity and related metabolic disorders: journal of the International Association for the Study of Obesity 20.5 (1996): 393-405.
• Hulmi, Juha J., et al. "The effects of intensive weight reduction on body composition and serum hormones in female fitness competitors." Frontiers in Physiology 7 (2017): 689.
• Keys, Ancel, et al. "The biology of human starvation.(2 vols)." (1950).