Article Text
Abstract
Background Amyotrophic lateral sclerosis (ALS) is a rare neurodegenerative disorder mainly characterised by motor symptoms. Extensive physical activity has been implicated in the aetiology of ALS. Differences in anthropometrics, physical fitness and isometric strength measured at 18–19 years were assessed to determine if they are associated with subsequent death in ALS.
Method Data on body weight and height, physical fitness, resting heart rate and isometric strength measured at conscription were linked with data on death certificates in men born in 1951–1965 in Sweden (n=809 789). Physical fitness was assessed as a maximal test on an electrically braked bicycle ergometer. Muscle strength was measured as the maximal isometric strength in handgrip, elbow flexion and knee extension in standardised positions, using a dynamometer. Analyses were based on 684 459 (84.5%) men because of missing data. A matched case control study within this sample was performed. The population was followed until 31 December 2006, and 85 men died from ALS during this period.
Results Weight adjusted physical fitness (W/kg), but not physical fitness per se, was a risk factor for ALS (OR 1.98, 95% CI 1.32 to 2.97), whereas resting pulse rate, muscle strength and other variables were not.
Conclusions Physical fitness, but not muscle strength, is a risk factor for death at early age in ALS. This may indicate that a common factor underlies both fitness (W/kg) and risk of ALS.
Statistics from Altmetric.com
Introduction
Amyotrophic lateral sclerosis (ALS) is a rare neurodegenerative disorder characterised by degeneration of motor neurons in the central and peripheral nervous system, leading to progressive muscle weakness and death, usually within 3–5 years from onset. The aetiology of ALS is considered multifactorial with multiple genetic and environmental factors causing motor neuron degeneration.1 Many environmental and lifestyle risk factors have been suggested in ALS, including physical activity and premorbid low body weight.2–7 The role of sports in ALS has recently been questioned.8 9 However, a report of a sixfold increase in risk of ALS in Italian professional football players has added fuel to this discussion.10
We assess whether differences in physical fitness, muscle strength and anthropometrics in men measured at the age of 18–19 years are associated with subsequent death in ALS.
Methods
Population
Data were linked between the Population Register (PR), Military Service Conscription Register (MSCR) and the Swedish Cause of Death Register using unique personal identification numbers.
In the PR, we identified men born in 1951–1965 in Sweden and who lived in Sweden in the year of their 17th birthday (n=809 789, study population, 100%, including 122 ALS cases). Of these, 745 310 (92.0%, 105 ALS cases) had a record in the MSCR between 1969 and 1983, and 686 815 (84.8%, 85 ALS cases) had been tested for physical fitness and muscle strength and had complete datasets on all but three variables. Of 686 815 subjects, 2356 were excluded because they had a medical diagnosis possibly indicating early onset ALS (diagnosis of poliomyelitis, neurological disease except for epilepsy and headache (ICD 8 codes 040–046, 320–344, 347–358, 716, 733 and 781)). This was done to strengthen causality—that is, to avoid the fact that baseline disease would affect our findings. This left 684 459 (84.5%) men for the nested case control study (effective sample). Before the case control study was performed, imputations of missing values were done. The population was followed until 31 December 2006, and 85 men died from ALS during this period.
Administrative data
Data on date of birth, date of death and emigration of the men and whether they were alive on 31 December 2006 were obtained in the PR and residential area in the MSCR.
Conscription examinations
The Swedish military service conscription examination is required by law. Until recently, the only reasons accepted for non-participation were foreign citizenship, severe handicap or disease. At the age of 18 years, all men underwent highly standardised examinations on physical and intelligence status, including examination by a physician.11
Test results (physical fitness, muscle strength, IQ) were standardised on a stanine-like scale ranging from 1 (low) to 9 (high). Ranges for each stanine score are known.11
Physical fitness test
Physical working capacity was assessed using an electrically braked bicycle ergometer, using a pedal rate of 60 revolutions per minute.12 A maximal test is one in which the conscript is to work at high constant load until exhaustion.13 14 The maximum load that the conscript could sustain for 6 min was used as a measure of physical working capacity (WMAX6 min,12 or, if the time was shorter, a WMAX6 min value was estimated from a nomogram.11 This test has good reliability (correlation coefficient 0.90)15 and correlates well with other endurance tests.14 Standard starting load was 14000 Nm/min. Other starting loads could be chosen according to physical stature, history of physical activity and medical history (personal communication, Dr Björn Ahlborg, 7 May 2010).
In conscripts with a medical status not allowing a maximal test, a submaximal test (W170) was performed. There is good correlation between maximal and submaximal tests (correlation coefficient 0.74).15 In conscripts unable to participate in the physical fitness test because of current infectious disease or other causes, physical fitness was estimated, not measured, by the supervising personnel according to physical stature, history of physical activity and medical history (personal communication, Dr Björn Ahlborg, 7 May 2010).
Isometric muscle strength
Methods were developed by Tornvall.12 14 16 Measurements were performed by trained military personal, and apparatus were calibrated on a daily basis. The strength of each muscle group was tested three times.14
Handgrip strength of the strongest hand was measured in Newtons (N) in the standing position using a dynamometer devise with the upper arm being held vertically with the elbow flexed at 90°.
Strength of right elbow flexion (N) was determined in the sitting position in a chair with the elbow held at 90° and the forearm held vertically. The fasting strap of the dynanometer was placed at the level of the styloid process of the radius.
The strength of the right knee extension (N) was measured in the sitting position in a chair with the tested leg hanging vertically. The fasting strap of the dynamometer was placed at the level of the lateral malleolus.
Other measures
Body weight was measured to the nearest kilogramme, and body height to the nearest centimetre.17 Heart rate and blood pressure were measured in the supine position after 10 min of rest.18 19
Diagnoses
Cases were men who had a diagnosis of ALS on their death certificate (ICD 8 code 348.00, 348.10 and 348.20, ICD 9 code 335.2 and ICD 10 code G12.2).
We validated the ALS diagnosis in the MSDR by comparing data in the medical records of patients who died in ALS with El Escorial criteria.20 A total of 100 male patients were randomly selected from the MCDR, stratified by date of death (10 deaths in 1970–1986, 45 deaths in 1987–1996 and 45 deaths in 1997–2004).
Statistical analyses
Matching of controls
Controls were matched according to conscription office, county, year of conscription and date of birth. Controls with birth dates as close as possible to the cases birth date were selected. Ties in birth date were broken randomly.21 Controls were alive and had not emigrated when their corresponding case died. If there were fewer than 30 controls from the same conscription year as the case, controls were also sampled from the previous and the following year.22
Regression analysis
The impact of various variables on the occurrence of death from ALS was examined by conditional logistic regression.21 23 For each variable Y of interest, we performed the stratified analysis according to models of the form Death from ALS ∼ Y + strata (match group). We assessed the risk of ALS (dependent variable) by each variable in table 1. Independent variables in table 1 were treated as continuous variables and ALS as a nominal variable (yes/no). Weight related associations were also assessed taking body height into account. We examined whether there was an effect of strength of knee extension on the relation between ALS and weight adjusted physical fitness, and whether there was an interaction effect between weight adjusted physical fitness and strength of knee extension.
In sensitivity analyses, we assessed associations (a) by including only cases who had complete data, (b) by including only cases who had complete data and who died from ALS aged 40 years or older and their controls (c) by standardising physical fitness, isometric strength and weight adjusted fitness according to date of conscription and (d) in conscripts who had their fitness actually measured and not estimated by the supervising physician.
Standardisation was performed by fitting a linear regression line to the observed levels by time. The levels were then replaced by the global average level plus the residuals estimated from the regression model.
The results were summarised by ORs and their 95% CIs. A p value less than 0.05 was considered statistically significant.
Missing values
The effective sample was formed by means of case deletion—that is, to a level where imputations of missing data could be performed with a high degree of precision.
Data on fitness were recorded only as stanine scores in 1969 to mid 1972, and data on fitness measured in watts was missing in 130 394 (19.1%) conscripts. The relationship between fitness measured in watts and existing stanine scores are known. Date on conscription was not recorded by two test centres in 1969–1970 and were missing in 18 166 (2.7%) conscripts. There was a high correlation between year of birth and conscription in available data (r>0.99). These variables, five IQ measures (missing in 1435 (0.2%) conscript), systolic and diastolic blood pressure, type of bicycle test, date of birth and all variables related to table 1 were used for imputations.
We performed multiple imputation using additive regression, bootstrapping and predictive mean matching.24 25 Ten realisations of complete imputed datasets were created. All subsequent statistical analyses were performed on each of the 10 data sets, and the results were combined in terms of estimates and SEs, to incorporate imputation uncertainty in the final results.26
Standard protocol approvals, registrations and patient consent
The project was approved by the local ethics committee. All data were made anonymous after linkage and before we were given access to them. We were not able to obtain informed consent from the participants, and this was not required by the ethics committee. It is forbidden to identify persons in registers made anonymous, so the patients in the validation study represent the patients in the linkage study but they are not necessarily the same.
Results
The percentage of ALS deaths was 0.015 in the population of men born in 1951–1965 and 0.012 in the effective sample. Median age at death in the 85 cases with ALS was 44 years (10% quantile=34 years and 90% quantile=52 years, range 24–55 years). Median age at conscription was 18.4 years in cases and 18.5 in controls. The number of available controls differed slightly between the matched data sets. In 83 of 85 control sets there were 30 controls, and in two of 85 control sets the number of controls varied between five and 30 between the 10 imputed datasets.
Measures of anthropometrics, physical fitness and isometric strength in ALS cases and controls are shown in table 1, and so are risk estimates for ALS. Physical fitness was estimated, not measured, in 13.5% of conscripts. Weight adjusted physical fitness was the only statistically significant risk factor for ALS. The distribution of weight adjusted physical fitness according to disease status is shown in figure 1.
In sensitivity analyses we found a similar pattern as in the complete data set. Strength of handgrip was the only exception as it was a risk factor for ALS in men who died after the age of 39 years (OR 1.00 (95% CI 0.99 to 1.00)). OR for the association between weight adjusted physical fitness and ALS was: (a) 1.81 (95% CI 1.07 to 3.06) after inclusion of men with complete data only (see table e1, left hand side; table e1 is available online only); (b) 1.75 (95% CI 0.93 to 3.27) after inclusion of cases with complete data who died before the age of 40 years (see table e1, right hand side; table e1 is available online only); (c) 1.98 (95% CI 1.32 to 2.97) after standardisation of variables (see table e2; table e2 is available online only); and (d) 2.2 (95% CI 1.42 to 3.45) after exclusion of men who did not have their fitness actually measured but estimated by the supervising physician.
Accounting for body height in analyses of the effect of weight adjusted handgrip, elbow extension and knee extension on ALS affected the risk only to a minor extent (all 95% CIs included 1, data not shown). Taking body height and strength of knee extension into account, the OR for ALS by physical fitness per kg body weight was 1.93 (95% CI 1.28 to 2.92). There was no statistically significant interaction between weight adjusted physical fitness and strength in knee extension (p=0.95).
Of 100 male patients in the Swedish Cause of Death Register with a diagnosis of ALS, we were able to collect medical records from 81 (81%, mean age 69.5 years, SD 10.8). Of these, 67 (82.7%) had a diagnosis that fulfilled El Escorial criteria, eight (9.0%) had unclassified motor neuron disease and six (7.4%) were incorrectly diagnosed with ALS. No patient aged 55 years or younger was incorrectly diagnosed.
Discussion
This study indicates that men with greater physical fitness per kg body weight have a larger risk of subsequently dying at an early age in ALS compared with men with lower physical capacity. Physical fitness per se was not a risk factor. In other words, the higher the ability of young men to move their own body weight, the greater the risk of subsequent death in ALS.
Physical fitness measured as a maximal test correlates well with maximal oxygen uptake (MOA)15 which requires integration of respiratory, cardiovascular and neuromuscular function, and is considered a fundamental measure of physiological functional capacity for exercise.27 Body weight is an important determinant of MOA, and interindividual differences are compared as MOA per kg body weight.28 There is considerable genetic contribution.27
There are some indirect indications in this study that the observed difference in weight adjusted fitness cannot be attributed to vigorous training activity, such as in professional soccer players. Firstly, heart rate is a known surrogate for exercise activity29 which is in line with the observed negative correlation between fitness and resting pulse rate in this study (−0.27). Resting pulse rate was not a risk factor for ALS. Rather, resting heart rate was numerically higher in cases than in controls (means 77.0 vs 74.6 bpm, p=0.09). Secondly, we would expect to find isometric strength as a risk factor. Thirdly, fitness in cases (figure 1) was distributed over a wide range.
Without indications of differences in exercise activity between cases and controls, we suggest that the most plausible explanation for our findings is that a common subclinical phenotype underlies both fitness and ALS. Our hypothesis is that this at risk phenotype is characterised by relatively more type 1 muscle fibres (slow twitch fibres). Slow twitch fibres are the predominant muscle fibre type in endurance athletics.30 Interestingly, presymptomatic SOD-1 mediated ALS mice are more active runners (15–20 km/day) than control mice (7–9 km/day).31 Type 1 fibres are affected earlier and more severely on muscle biopsy in ALS.32 33 To our knowledge, there are no data on premorbid muscle fibre pathology in ALS. We do not have genetic information on cases and controls.
Low premorbid weight was associated with ALS in one study5 but not in another.34 In the present study, ALS cases had a lower mean weight than matched controls (66.3 compared with 68.5 kg) but this difference was not statistically significant.
This study has certain unique features. The main strengths lie in its coverage of the whole young male population of a country, with low dropout rates. The mean time elapsed between conscription and end of follow-up was long, approximately 25 years. Measures on anthropometrics and physical capacities were highly standardised and were not biased by self-reporting. Self-reported weight tends to be underestimated, and height overestimated,35 and the accuracy between measured and self-reported weight decreases with elapsing time.36 Furthermore, we were able to exclude patients with neurological disease at conscription.
Our main finding—an increase in the risk for ALS by weight adjusted physical fitness—was robust in all sensitivity analyses. This implies some protection against potential biases induced by data imputation, reverse causality, estimated physical fitness in some patients and measurement differences over time.
One limitation is failure to clinically diagnose ALS directly in the study population. Our finding of incorrect diagnoses (7%) in the validation set is in line with the report of a false positive diagnosis in 8% in an earlier study.37 All persons who died at age 55 years and younger were correctly diagnosed. Other motor neuron disorders that possibly could be mistaken for ALS in this age group are multifocal motor neuropathy, Kennedy's disease and spinal muscle atrophy. However, the clinical course of these disorders is different from that of ALS and, in general, they do not lead to death in early middle age. We do not believe that misdiagnosis of ALS affects the results of our study in any particular way.
Although studies of the fitness test have shown good reliability and validity in various groups,14 15 and there were central instructions and training of the military personnel, its performance at conscription over a long period of time has not been assessed. As a measure of validity, physical fitness in this population has been shown to be associated with intelligence and to predict educational achievements and occupational outcome later in life.38 Starting load of the bicycle test was based on physical stature and anamnesis. Heavy men started on average on higher loads than slimmer men did, which may have biased their results towards higher fitness results—that is, reduced their chance of obtaining low values. This may have underestimated the physical fitness in men who later develop ALS.
Certain factors that may affect the risk of ALS both before and after conscription (eg, genetic factors, smoking habits, exposure to toxins, heavy manual labour) have not been measured. However, this was not the aim of the study.
Although an increase in physical fitness by 1 W/kg was associated with doubled odds for ALS, the absolute risk for ALS in fit young men is still extremely low.
The findings of this study can be generalised to young men who died early from ALS. We need to know more about the relationship between physical activity, physical fitness, muscle pathology and ALS, and the possible connection with SOD mutations.
References
Supplementary materials
Web Only Data jnnp.2010.218982
Files in this Data Supplement:
Footnotes
Funding This study was sponsored by the Selanders Foundation, which was not involved in the completion of the study in any way.
Competing interests PM received research support from Selanders Foundation and Epilepsifonden. HA serves on the Advisory Board Neurology of H Lundbeck AB, Sweden, and on the Advisory Board Immunoglobulins of KL Behring, Sweden. He received research support from Selanders Foundation. IN received research support from an anonymous donor.
Ethics approval The study was approved by the local ethics committee.
Provenance and peer review Not commissioned; externally peer reviewed.
Linked Articles
- ALS and FTD Special Edition