There is accumulative evidence that lifestyle factors and physical exercise through their effects on neuroplasticity reduce the risk of dementia in older adults. However, a closer look at the study results reveals a heterogeneous picture. In sport science, it has been known since the 1980s that the response to physical exercise varies tremendously among individuals. Especially in endurance and strength training there is strong evidence for different individual physiological adaptations to an identical training parameter. Karavirta et al. (2011) reported that after a 21-week intervention cardiorespiratory fitness (VO2peak) varied from −8% to 42 % and strength (maximal isometric bilateral leg extension) from −12% to 87%. Based on these variations humans can be divided into 'responders' or 'non-responders' (Buford et al., 2013) although according to Bonafiglia et al. (2016) there are no 'global non-responders'. Instead individual patterns of response indicate that non-responders to endurance training can nevertheless improve physiologically by sprint interval training and vice versa. Results of the landmark study from 2010 indicate that these wide physiological responses to exercise are based, amongst other causes, on genetic factors (Timmons et al., 2010). This inter-individual variability indicates that there is an essential need for personalized training to achieve optimal results from physical exercise. Based on the hypothesis that cardiovascular factors are the central mediators of exercise-induced neuroplasticity (Sanchez and McGough, 2014), the different physiological adaptations of 'responders' and 'non-responders' have fundamental implications for the impact of exercise on neuroprotection. This suggests that training should be tailored to individual strengths and weaknesses to achieve maximal neuroprotection. We hypothesize that exercise training adapted to individual performance levels regarding both quality (i.e. type of training) and quantity (i.e. intensity) should be superior in inducing neuroplasticity and in preventing neurodegeneration than a 'one size fits all' program.

Neuromuscular junction (NMJ) degeneration and loss of motor units (MUs), markedly contribute to sarcopenia. NMJ damage in sarcopenic individuals can be assessed from serum measurements of c-terminal peptide agrin fragment (CAF) (Hettwer et al. Exp Gerontol 2013). Interestingly, aerobic exercise protects against NMJ degeneration (Valdez et al. Proc Natl Acad Sci 2010) in senile rats and in humans, no decline in MUs has been found in master runners (Power et al. Med Sci Sports Exerc 2010). The present study aimed to investigate whether ball-room dancing could have neuroprotective effects compared to gym training (endurance, strength-endurance and flexibility training). Thirty-seven older individuals (aged 71.6 ± 3.5 yr, 18 female and 19 male) were randomly assigned to a Dance Group (DG) or Gym Exercise Group (GEG). Both groups trained twice/week, 90 min/session, for 6 months. Blood serum CAF levels (Neurotune AG, Schlieren, Switzerland) were measured before and after the intervention. Data were compared to those of older sarcopenic and young controls. Values are means ± S.D., paired/unpaired Student's t-Test were used. Pre-training, CAF values of the DG and GEG groups were respectively 1.9 and 2.2-fold higher than of young controls. However, after the intervention, CAF levels decreased by 15% (P<0.001) in DG, while no changes were found in the GEG. These findings suggest a reduction of neuromuscular degeneration in older humans as a result of six-month recreational dancing but not by general fitness training. This neuroprotective effect of dancing may be due to improved neurotrophin levels, reduced oxidative stress and inflammation.