The best magazine
Discussion
Evidence for an increased risk of ALS with higher levels of leisure time physical activity is provided by the present population based case control study. Occupational physical activity and performing vigorous physical activities, however, do not appear to modify ALS susceptibility in this study. The discrepancy between leisure time and occupational physical activity strengthens the hypothesis that physical activity itself is not causative per se but that being athletic is a phenotypic expression of a genetic profile, mediated by exogenous factors, that increases the risk of ALS. Our observation that none of the physical activity measures was related to age at onset or survival further supports this hypothesis.
Two systematic reviews on the association between ALS and physical activity concluded that there is a consistent pattern of well designed studies showing no link between physical activity and sporadic ALS. The best evidence available at that time was provided by a single population based case control study that showed no association. After publication of these reviews, however, a small but well designed European population based pilot case control study identified an increased risk of ALS with higher levels of physical activity. In concordance with these conflicting results, a third and the most recent systematic review concluded that current evidence for physical activity as a risk factor in motor neuron disease is not of sufficient calibre to allow undisputed conclusions.
The conflicting results found in studies on the association between physical activity and ALS may partly be due to differences in methodological design. These differences concern: (1) the blinding of interviewers to disease status of respondents or the hypotheses being tested; (2) referral bias, which was common with cases often ascertained at specialist clinics; (3) adjustment for confounders, which was not carried out in all analyses; and (4) the method of assessing physical activity, which in most studies was susceptible to recall bias.Recall bias is due to differential recall of past exposures between patients and controls. As ALS patients actively search for an explanation of their disease or may have an assumption about the underlying cause, case control studies in ALS using questionnaires are prone to this bias. Our study was designed to minimise the risk of recall and referral bias. First, recall bias was reduced by using the Compendium of Physical Activities to quantify objectively physical activity based on type of occupation or type of leisure time activities, instead of directly asking participants how physically active they have been in their life or during the listed activities. As the questionnaire on leisure time and occupational activities was part of a more comprehensive questionnaire, participants were blinded to the study hypothesis, which further reduced the risk of recall bias. Interviewers, who called participants to complete returned questionnaires, were also unaware of the hypothesis being tested.
Referral bias may occur when patients are ascertained from tertiary care centres. It has been demonstrated that ALS patients attending these referral centres do not represent a random sample of all ALS patients. A difference in physical activity levels of these patients compared with non-referred patients will lead to biased results. The population based design using multiple sources to ensure complete case ascertainment minimised the risk of referral bias in the present study, which was strengthened by the observation that the demographics of the patients in our study resembled those of patients in other population based studies.
We acknowledge certain limitations of the present study: 18.6% of the participants had at least one missing value for the duration of, or the hours per week spent on, one of the listed activities, even after being called by an interviewer to complete the returned questionnaire. This is probably the result of the level of detail of the questionnaire concerning past events. The fact that this information was so elaborate, however, enabled us to precisely quantify lifetime energy expenditure during leisure time and occupational activities. Also, it is noteworthy that ALS patients had significantly less higher education (p<0.02), which is congruent with a previous observation that there is a preponderance among ALS patients of blue collar jobs for which a higher level academic education is often not required. Nevertheless, our controls may have been better educated as people with a higher education tend to participate in scientific surveys more readily. The effects, however, of this observation will have been minimal as we adjusted all analyses for education. Further, we acknowledge that the quantification of the lifetime energy expenditure is still an estimate of real energy expenditure. A study, however, in which these data are prospectively being collected will probably not be feasible in a low incidence disease such as ALS. Finally, although our study was designed to maximise blinding of the participants to the hypothesis of the study, we cannot exclude the fact that a proportion of the patients may have been aware of the theory of physical activity as a possible risk factor, which may have been a source of residual recall bias.
Our finding that an increased leisure time physical activity is related to an increased risk of ALS but occupation activity is not, raises doubts regarding the role of physical activity in causing ALS. Because of existing cellular and genetic evidence supporting the biological plausibility of the association, some have suggested that physical activity is indeed causative. Several genes associated with the response to exercise (ie, ciliary neurotrophic factor, leukaemia inhibitory factor and vascular endothelial growth factor 2) have been identified as possible modifiers of ALS susceptibility. Also, oxidative stress and glutamate excitotoxicity are considered candidate mechanisms to link ALS and physical activity. The biologically plausible link between physical activity and ALS has been carefully reviewed.
Biological plausibility alone, however, does not prove causation. Useful time tested criteria for determining whether an association is causal were designed by Bradford Hill. The Bradford Hill criteria include strength, consistency, specificity, temporality, dose–response relation, plausibility, coherence, experiment and analogy. The associations found in the present study do not meet most of these criteria. First, strength. If an association is weak, it is more plausible that underlying actual causative factors that go hand in hand with the studied factor are in fact responsible for the observed association. In our study, if physical activity were causative, an increase in physical activity of 10 000 MET, which can be provided by 50 years of 50 h cycling per week for example, would be associated with an increase in the odds of developing ALS of only 2.2. Further, if we had applied a more stringent threshold that also corrects for analyses on vigorous physical activities (threshold p=0.05/7=0.007), the association (p=0.008) would not even have been significant, further emphasising the weakness of the association. Second, consistency. A real causative factor is more likely to be repeatedly observed in different studies, using different methodologies and performed in different places, circumstances and times. Previous studies, as already emphasised, have shown large inconsistencies, and even within the present study there is an inconsistency between occupational and leisure time physical activity. Finally, the absence of a dose–response relation does not support the notion that causation is the most likely interpretation of the association between leisure time physical activity and ALS. Recent findings of a beneficial vascular risk profile in both patients and their relatives, a reduced frequency of coronary heart disease premorbidly in ALS and an increased risk of ALS with physical fitness but not muscle strength further indicate that a common factor underlies both physical/cardiovascular fitness and risk of ALS. A genetic profile, therefore, modified by exogenous factors, that both promotes physical fitness and increases ALS susceptibility, might be a more credible explanation for the associations between physical activity and ALS.
In conclusion, the present population based case control study strengthens this hypothesis. Identifying genetic, developmental and environmental factors that contribute to physical fitness may provide a worthwhile lead in unravelling the pathophysiological mechanisms in ALS.
Source: ...