Unravelling the nature of postexertional malaise in myalgic encephalomyelitis/chronic fatigue syndrome: the role of elastase, complement C4a and interleukin-1β

Authors

  • J. Nijs,

    1. From the Department of Human Physiology, Faculty of Physical Education & Physiotherapy, Vrije Universiteit Brussel, Brussels
    2. Division of Musculoskeletal Physiotherapy, Department of Health Care Sciences, University College Antwerp, Antwerp
    3. Department of Physical Medicine and Physiotherapy, University Hospital Brussels, Brussels
    Search for more papers by this author
  • J. Van Oosterwijck,

    1. From the Department of Human Physiology, Faculty of Physical Education & Physiotherapy, Vrije Universiteit Brussel, Brussels
    2. Division of Musculoskeletal Physiotherapy, Department of Health Care Sciences, University College Antwerp, Antwerp
    Search for more papers by this author
  • M. Meeus,

    1. From the Department of Human Physiology, Faculty of Physical Education & Physiotherapy, Vrije Universiteit Brussel, Brussels
    2. Division of Musculoskeletal Physiotherapy, Department of Health Care Sciences, University College Antwerp, Antwerp
    Search for more papers by this author
  • L. Lambrecht,

    1. Private Practice for Internal Medicine, Gent/Aalst
    Search for more papers by this author
  • K. Metzger,

    1. RED Laboratories N.V., Zellik, Belgium
    Search for more papers by this author
  • M. Frémont,

    1. RED Laboratories N.V., Zellik, Belgium
    Search for more papers by this author
  • L. Paul

    1. Nursing and Health Care, Faculty of Medicine, University of Glasgow, Glasgow, UK
    Search for more papers by this author

Dr Jo Nijs, Vrije Universiteit Brussel, Building L-Mfys, Pleinlaan 2, BE-1050 Brussels, Belgium.
(fax: +3226292876; e-mail: jo.nijs@vub.ac.be).

Abstract

Abstract.  Nijs J, Van Oosterwijck J, Meeus M, Lambrecht L, Metzger K, Frémont M, Paul L (Vrije Universiteit Brussel, Brussels; University College Antwerp, Antwerp; University Hospital Brussels, Brussels; Private Practice for Internal Medicine, Gent/Aalst; and RED Laboratories N.V., Zellik; Belgium, and University of Glasgow, Glasgow, UK). Unravelling the nature of postexertional malaise in myalgic encephalomyelitis/chronic fatigue syndrome: the role of elastase, complement C4a and interleukin-1β. J Intern Med 2010; 267: 418–435.

Objectives.  Too vigorous exercise or activity increase frequently triggers postexertional malaise in people with myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS), a primary characteristic evident in up to 95% of people with ME/CFS. The present study aimed at examining whether two different types of exercise results in changes in health status, circulating elastase activity, interleukin (IL)-1β and complement C4a levels.

Design.  Comparative experimental design.

Setting.  University.

Subjects.  Twenty-two women with ME/CFS and 22 healthy sedentary controls Interventions: participants were subjected to a submaximal exercise (day 8) and a self-paced, physiologically limited exercise (day 16). Each bout of exercise was preceded and followed by blood sampling, actigraphy and assessment of their health status.

Results.  Both submaximal exercise and self-paced, physiologically limited exercise resulted in postexertional malaise in people with ME/CFS. However, neither exercise bout altered elastase activity, IL-1β or complement C4a split product levels in people with ME/CFS or healthy sedentary control subjects (P > 0.05). Postexercise complement C4a level was identified as a clinically important biomarker for postexertional malaise in people with ME/CFS.

Conclusions.  Submaximal exercise as well as self-paced, physiologically limited exercise triggers postexertional malaise in people with ME/CFS, but neither types of exercise alter acute circulating levels of IL-1β, complement C4a split product or elastase activity. Further studying of immune alterations in relation to postexertional malaise in people with ME/CFS using multiple measurement points postexercise is required.

Introduction

There is evidence to support specific exercise therapies as a cornerstone in the comprehensive management of patients with myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) [1–4]. However, too vigorous exercise or activity [5–8] frequently triggers postexertional malaise in people with ME/CFS, a primary characteristic evident in up to 95% of people with ME/CFS [9]. Postexertional malaise is not present in other disorders where fatigue is a predominant symptom [10, 11] and it is one of the best predictors in the differential diagnosis of ME/CFS and major depressive disorder [12]. However, the aetiology of postexertional malaise has yet to be revealed.

People with ME/CFS respond to an exercise challenge with an enhanced complement activation [10], increased oxidative stress [5] and an exaggeration of resting differences in gene expression profile in peripheral blood mononuclear cells (PBMCs) [13]. Therefore, current strategies for applying ‘safe’ exercise interventions to patients with ME/CFS apply paced and physiologically limited types of exercise [2, 14, 15]. Yet it is unknown whether the immune changes in response to (sub)maximal exercise previously observed in ME/CFS patients (e.g. complement C4a split product [10, 16]) are equally present in response to self-paced, physiologically limited exercise. And if so, are immune changes related to symptom increases typically observed following exercise in people with ME/CFS?

In addition, debate regarding the appropriate method of pacing exercise for people with ME/CFS remains. Moreover, studies examining various types of exercise in people with ME/CFS are scarce. The ultimate goal of applying exercise limits is the prevention of postexertional malaise. Therefore, the current study compared a standardized submaximal type of exercise with a self-paced, physiologically limited exercise. Previous work revealed increased expression of the lectin pathway (C4 and mannan-binding lectin serine protease 2) in PBMCs of ME/CFS patients in response to submaximal exercise, resulting in significant increase of C4a spilt product [10, 16]. It was hypothesized that people with ME/CFS are able to perform a self-paced, physiologically limited exercise without symptom exacerbations and without triggering complement C4a split product increase.

In addition to complement C4a split product, elastase and interleukin-1β (IL-1β) may be of importance in relation to postexertional malaise in people with ME/CFS. Elastase is a proteolytic enzyme capable of initiating high molecular weight ribonuclease L (83 kDa) proteolysis in patients with ME/CFS, generating two major fragments with molecular masses of 37 and 30 kDa respectively [17–19]. In a previous study of people with ME/CFS, we reported that baseline elastase level was predictive of the respiratory exchange ratio and the oxygen uptake at the anaerobic threshold [20]. Based on these findings, it was hypothesized that submaximal exercise, as opposed to self-paced, physiologically limited exercise, triggers elastase activity in people with ME/CFS but not in healthy sedentary controls.

Chronic fatigue has been arbitrarily put forward as the primary symptom of ME/CFS, but pain and fatigue are equally disabling to ME/CFS sufferers [21]. IL-1β as well as elastase can be viewed as markers of the inflammatory response. IL-1β leads to widespread transcription of cyclooxygenase-2 in neurons, resulting in prostaglandin production, which in turn augments neuronal excitability in somatosensory pathways and triggers hyperalgesia [22]. In addition, IL-1 activates the hypothalamus pituitary adrenal-axis [23]. Cortisol and corticotropin-releasing-factor, two crucial players in the hypothalamus pituitary adrenal-axis, are both involved in pain sensitivity [24, 25]. Patients with ME/CFS have mild hypocortisolism, a blunted adrenocorticotropin response to stressors and an enhanced negative feedback, possibly explanatory for chronic widespread pain [reviewed in 26, 27]. A deficient hypothalamus pituitary adrenal-axis functioning might foster pathological immune activation with release of pro-inflammatory cytokines [28] like interleukin-1β, provoking a so-called ‘sickness response’ [29]. Too vigorous exercise might be one of the frequent stressors triggering IL-1β release and subsequent increased pain experience in patients with ME/CFS. Therefore, it was hypothesized that submaximal exercise as opposed to self-paced, physiologically limited exercise, triggers IL-1β release and related pain increases in people with ME/CFS but not in healthy sedentary controls.

In summary, the present study aims to examine the following research questions:

  • 1Does submaximal exercise triggers elastase activity, IL-1β and complement C4a level increases in people with ME/CFS and healthy controls?
  • 2Are people with ME/CFS able to perform self-paced, physiologically limited exercise without symptom exacerbations and without triggering elastase activity, IL-1β and complement C4a level increases?
  • 3Are changes in elastase activity, IL-1β and complement C4a levels associated with symptom exacerbations following exercise in people with ME/CFS?

Methods

Subjects

To be included in the study, subjects had to fulfil the Center for Disease Control and Prevention criteria for ME/CFS [30] determined through an extensive medical evaluation. Pain is considered an important aspect of postexertional malaise. Therefore, as well as suffering from ME/CFS, included patients had to present with chronic widespread pain as described in the 1990 criteria of the American College of Rheumatology [31]. Pooling of gender data has been identified as an important source of bias in studies addressing exercise physiology in ME/CFS patients [32]. Therefore, the study focused on women within the age range of 18 to 64 years. The post hoc power analysis from a previous and related study revealed that a sample of 22 people with ME/CFS was required [20].

Patients with ME/CFS were asked to bring a healthy, sedentary relative, friend or acquaintance to participate in the control group. Sedentary was defined as having a seated profession and performing a maximum of 1 h of sports per week [33]. Twenty-two women with ME/CFS and 22 healthy sedentary women were recruited. The mean age of the patients was 34.3 ± 8.8 years and their mean body mass index (BMI) was 24.1 ± 4.7. The mean age of the controls was 38.9 ± 15 years and their BMI was 24.5 ± 4.8. The independent samples t-test revealed no significant difference for age or BMI between patients and controls (P > 0.01). In the ME/CFS group, 12 patients used analgesics and 10 patients used anti-depressants. In the control group, 1 patient used anti-depressants. All participants were asked to refrain from taking their medication for at least 1 week before the start of the study and were instructed to stop medication use throughout the study period.

Procedure

All study participants provided informed consent and the study was approved by the Ethics Committee of the University Hospital Brussels/Vrije Universiteit Brussel. After collecting information addressing personal characteristics (age, usage of analgetics or anti-depressants, etc.), the participants’ height and weight were measured and they were provided by a tri-axial accelerometer for real time activity monitoring (Actical Mini Mitter, Bend, OR, USA). Participants were asked to wear the accelerometer continuously during study participation. Figure 1 displays the flow diagram of the study.

Figure 1.

 Flow diagram of the study. *Centre for Disease Control and Prevention; †Myalgic Encephalomyelitis/Chronic Fatigue Syndrome.

Participants were required to complete two exercise tests (Experiment 1 and Experiment 2) 1 week apart. At each visit, they were required to fill out the following questionnaires: the CFS Symptom List, the Medical Outcomes Short Form 36 Health Status Survey (SF-36) and the Checklist Individual Strength (CIS). After completing the questionnaires, four venous blood samples (in total 32.5 cc) were taken for determination of the immune variables. Next, study participants were subjected to one of two execise tests (described below). Afterwards, subjects were given the opportunity to take a 1-h break, during which they had the opportunity to drink water ad libidum and to take a shower. At 1 h postexercise, the patients were asked to provide four venous blood samples and to fill out the CFS Symptom List and SF-36 once again. For monitoring postexercise malaise up to 24 h postexercise, subjects were given two questionnaires (CFS Symptom List and SF-36) to fill out exactly 24 h postexercise. The subjects were asked to return the filled-out questionnaires to the university by postal mail (prestamped envelopes).

Self-reported measures

The CFS Symptom List is a self-reported measure for assessing symptom severity in people with ME/CFS. It encompasses the 19 most frequently reported symptoms in a sample of 1578 patients with ME/CFS [34]. Data in support of the internal consistency (Cronbach’s α = 0.88), test–retest reliability (ICC ≥ 0.97), content and concurrent validity of the CFS Symptom List have been reported [35, 36].

The SF-36 assesses functional status and well-being or quality of life [37]. The SF-36 has been documented to have reliability and validity in a wide variety of patient populations [37–39] and it appears to be the most frequently used measure in ME/CFS research [40].

The CIS quantifies subjective fatigue and related behavioural aspects [41]. The CIS consists of four fatigue aspects or subscales, i.e. fatigue severity, reduced motivation, reduced activity and reduced concentration. The CIS is well validated within the clinical setting and is able to discriminate between different patient populations, and between ME/CFS patients and healthy subjects [42].

Experiment 1: Submaximal exercise (Aerobic Power Index test)

The submaximal exercise protocol consisted of a bicycle test known as the Aerobic Power Index test [43], which has been shown to generate reliable data in sedentary people and patients with ME/CFS (intraclass correlation coefficient = 0.98 and 0.97 respectively) [43, 44]. In people with ME/CFS, the submaximal exercise data generated by the Aerobic Power Index correlate highly with peak exercise data [45]. According to the description of the Aerobic Power Index, the workload was increased with 25 W every minute, and the submaximal level was defined as 75% of the age-predicted target heart rate [46]. If subjects were unable to reach their individual target heart rate, then the kwatts per kilogram achieved during the last full minute of the exercise was recorded as the final power output.

Experiment 2: Self-paced and physiologically limited exercise

The self-paced and physiologically limited bicycle exercise was performed with three ‘safety breaks’ or exercise limits. Heart rate and workload limits were chosen to obtain an aerobic exercise, well below the anaerobic threshold [14]. The participants were monitored to make sure that the heart rate would not exceed 80% of the heart rate corresponding to the anaerobic threshold achieved during the Aerobic Power Index test. Where the anaerobic threshold was not achieved during the Aerobic Power Index test, 80% of the highest achieved heart rate was used. Where the heart rate exceeded the upper limit (80% of the heart rate corresponding to the anaerobic threshold) during the self-paced and physiologically limited exercise, the workload was lowered and subjects were instructed to lower their cycling frequency. The exercise duration was determined using the principles of pacing self-management as commonly used in people with ME/CFS to prevent postexertional malaise [15]. Therefore, the participants were asked to pace the exercise by estimating how long they thought they would be able to perform the bicycle exercise without exacerbating their symptoms. The activity duration estimated by the participants was reduced to account for typical overestimations. Seventy-five per cent of the estimated time was used when the participants reported having a ‘good’ day and 50% of the estimated time was used when the participants reported having a ‘bad’ day [15]. For the controls, the estimated time was decreased by 25%. Thus, during experiment 2, all subjects performed one bicycle exercise below all three safety breaks. Not aerobic, but anaerobic exercise has been shown to increase symptoms in people with ME/CFS [5, 6].

Both exercise tests were performed in a sitting position on an electrically braked cycle ergometer (Excalibur Lode, Groningen, the Netherlands) at a room temperature of 18 °C to 20 °C. After 3 to 5 min of adjustment to the position, baseline measurements were collected. The oxygen analyser was calibrated with known gas mixtures of 18% O2 and 5% CO2. The room air calibration was automatically run before each test to update the CO2 analyser baseline and the O2 analyser gain so that they coincided with atmospheric values. To collect pulmonary data during the test, an open circuit spirometer (Cortex Metamax I, Biophysik, GmbH, Duitsland) with automatic printout every 30 s was used. The Cortex Metamax I is a reliable instrument for exercise testing in medical routine and research [47]. A two-way breathing valve attached to a mask, which covered the patients’ nose and mouth, was used to collect the expired air. The air was analysed continuously for ventilatory and metabolic variables. Study participants were instructed to cycle with a pedalling rate of 60–70 rates per minute. The heart rate was recorded at the end of each minute of the exercise test using a Polar T61-Coded heart rate monitor (Polar Electro OY, Kempele, Finland). To measure the lactate concentration, blood samples (20 mL) were drawn from an arterialized ear lobe every 2 mins during the exercise test. Lactate concentrations were determined enzymatically (EKF; BIOSEN 5030, Magdeburg, Germany).

Determination of elastase activity

Elastase activity in PBMCs was measured using an enzymatic – colorimetric assay: EnzChek® Elastase Assay Kit E-12056 (Molecular Probes, Eugene, OR, USA). The EnzChek kit contains DQTM elastin – soluble bovine neck ligament elastin that has been labelled with BODIPY® FL dye such that the conjugate can be digested by elastase or other proteases to yield highly fluorescence fragments. The resulting increase in fluorescence was monitored with a fluorescence microplate reader. The assay procedure is reported in detail elsewhere [20]. According to the company supplying the assay, the elastase activity assay has been thoroughly tested before it was brought on the market, but reliability and validity data are proprietary and unpublished (Nijs J, personal communication). To ascertain a valid assessment of elastase activity in the present study, approximately 15 plasma samples were re-examined using the RD191021100 human PMN elastase ELISA (BioVendor GmbH, Heidelberg, Germany), an in vitro sandwich enzyme immunoassay for the quantitative measurement of the complex of human polymorphonuclear elastase and α1-proteinase inhibitor in plasma. Using the procedure as described in the manual, nearly identical results were found as using the EnzChek® Elastase Assay Kit E-12056.

Determination of complement split product C4a

The BD OptEIATM human C4a ELISA kit (BD Biosciences, San Jose, CA, USA) was used for the in vitro quantitative determination of human C4a and the human C4a-desArg in human EDTA plasma. It is a solid phase sandwich ELISA (enzyme-linked immunosorbent assay) utilizing a monoclonal antibody specific for human C4a/C4a-desArg coated on a 96-well plate. Standards (lyophilized human serum containing a defined amount of C4a-desArg) and samples were added to the wells and any C4a-desArg present bonded to the immobilized antibody. Next, the wells were washed and a mixture of biotinylated anti-human C4a polyclonal antibody (containing fetal bovine serum and ProClinTM-150 as a preservative) and streptavidin-horseradish peroxidase was added, producing an antibody–antigen–antibody ‘sandwich’. The wells were again washed and a substrate solution was added, which produced a blue colour in direct proportion to the amount of C4a-desArg present in the initial sample. The Stop Solution (13 mL of 1 mol L−1 phosphoric acid) then changed the colour from blue to yellow and the wells were read at 450 nm. For more details regarding the specimen collection, specimen handling, reagent preparation and assay procedure, the readers are referred to the manufacturers’ manual. The assay has a minimum detectable dose of 0.006 ng mL−1, has limited cross-reactivity and the intra-assay (% coefficient of variation ranges from 4.0 to 6.5) and inter-assay (% coefficient of variation ranges from 6.5 to 9.7) precision are adequate.

Determination of IL-1β

Quantitative in vitro detection of human IL-1β was performed using two different assays: the human Biotrak™ Easy ELISA (product code RPN5971; Amersham Biosciences Europe GmbH, Freiburg, Germany) and the Endogen® Human IL-1β ELISA kit (Pierce Biotechnology, Inc. Rockford, IL, USA). According to the manufacturer, the former assay has a detection range of 7.8 to 500 pg mL−1 and a sensitivity of 1.1 pg mL−1. The latter has a detection range of 10.24 to 400 pg mL−1, a sensitivity below 1 pg mL−1, a high reproducibility (intra-assay and inter-assay coefficient of variation below 10%) and it displays no cross-reactivity with other cytokines. For both assays, specimen collection and handling, reagent preparation and assay procedure were performed as described in the manufacturers’ manual. Very briefly, an anti-IL-1β monoclonal coating antibody was absorbed onto microwells. IL-1β present in the sample bonded to antibodies absorbed to the microwells. A biotin-conjugated monoclonal anti-IL-1β antibody bonded to IL-1β captured by the first antibody. Streptavidin-HRP bonded to the biotin conjugated anti-IL-1β. Following incubation, unbound biotin conjugated anti-IL-1β and Streptavidin-HRP was removed during a wash step and substrate solution reactive with HRP (tertamethyl-benzidine) was added to the wells. Next, a coloured product was formed in proportion to the amount of IL-1β present in the sample. The reaction was terminated by addition of phosphoric acid and absorbance was measured at 450 nm. A standard curve was prepared from seven IL-1β standard dilutions and IL-1β sample concentrations determined.

Real time activity monitoring throughout the study

The Actical (Mini Mitter) water resistant accelerometer was used for real-time monitoring of physical behaviour of all study participants. The Actical accelerometer has an omnidirectional sensor and is capable of measuring movement in one plane. The sensor functions via a cantilevered rectangular piezoelectric bimorph plate and seismic mass, and it is capable of detecting movements in the 0.5- to 3-Hz range. Voltage generated by the sensor is amplified and filtered via analogue circuitry. The amplified and filtered voltage is passed into an analogue to a digital converter and the process is repeated 32 times per second (32 Hz). The resulting 1-s value is divided by four and then added to an accumulated activity value (activity counts) for the epoch. Accelerometers are the gold standard for measuring physical behaviour during daily activities. The Actical accelerometer has been used in scientific research and has shown to be valid for the real-time assessment of physical behaviour [48]. For the present study, the monitors were initialized to save data in 1-min intervals (epochs).

Statistical analysis

All data were analysed using SPSS 16.0 for Windows (SPSS Inc., Chicago, IL, USA). Normality of the variables was tested with the Kolmogorov-Smirnov test and appropriate descriptive statistics were calculated. Baseline comparisons between ME/CFS patients and controls were performed using student’s t-testing. For each type of exercise (i.e. submaximal exercise and self-paced and physiologically limited exercise), possible changes in any of the outcome measures (including the immune variables) in response to both types of exercise were compared between the two groups using two-way repeated measures anova. Likewise, the evolution in immune variables in response to both types of exercise was compared using two-way repeated measures anova. The responses of the immune variables to the two types of exercise were reexamined using the average activity count as a covariate. This was done to examine the potential bias originating from the daily physical activity level. Normally distributed outcomes related to exercise performance were examined with the paired samples t-test. Nonparametric data were examined using a Wilcoxon signed ranks test. To examine the associations between immune variables and symptom occurrence, Pearson correlation analyses were used. The significance level was set at 0.05.

Results

Submaximal exercise versus self-paced and physiologically limited exercise

Control subjects had a lower peak ventilation (P = 0.010) and peak oxygen uptake (P = 0.002) during performance of the self-paced, physiologically limited exercise as compared with the submaximal exercise (Table 1). A similar pattern of results was seen in the ME/CFS group (peak ventilation P < 0.001 and peak oxygen uptake P = 0.009), where a significantly lower peak respiratory exchange ratio (P = 0.005) was registered as well. Exercise intensity was lower than during sub-maximal exercise. Although participants cycled longer during the self-paced, physiologically limited exercise, lactate levels were significantly lower than during the submaximal exercise (ME/CFS group P = 0.011, control group P = 0.001). All these results support our initial intention of studying two different exercise bouts in both groups.

Table 1.   Comparison of the submaximal exercise data with the data from the self-paced, physiologically limited exercise in the ME/CFS group (n = 22) and the sedentary control group (n = 22)
VariableSubmaximal exercise (mean ± standard deviation)Self-paced, physiologically limited exercise (mean ± standard deviation)P-value
  1. SD, standard deviation; RER, respiratory exchange ratio; VO2P, peak oxygen uptake; VEp, peak ventilation; statistically significant results are printed in bold.

  2. Comparisons between ME/CFS patients and controls were performed using student’s t-testing.

ME/CFS patients
 VO2P (mL min−1 kg−1)24.1 ± 13.917.9 ± 11.20.009
 VEp (L min−1)30.5 ± 8.925.5 ± 5.7<0.001
 RERP1.25 ± 0.981.77 ± 4.00.005
 Resting lactate (mmol L−1)0.89 ± 0.241.07 ± 0.350.065
 Lactate at exercise completion (mmol L−1)2.96 ± 1.51.91 ± 0.820.011
 Exercise duration (min)3.9 ± 1.34.71 ± 2.50.179
Control subjects
 VO2P (mL min−1 kg−1)27.6 ± 24.717.1 ± 3.10.002
 VEp (L min−1)30.5 ± 8.924.4 ± 4.80.010
 RERP0.92 ±  0.110.86 ±  0.250.299
 Resting lactate (mmol L−1)1.1 ±  0.461.3 ±  0.360.239
 Lactate at exercise completion (mmol L−1)2.6 ± 1.11.65 ±  0.660.001
 Exercise duration (min)4.2 ± 1.29.3 ± 5.2<0.001

Are people with ME/CFS able to perform self-paced, physiologically limited exercise without symptom exacerbations?

During experiment 1, a difference between patients and controls regarding the evolution (baseline vs. 1 h postexercise vs. 24 h postexercise) of SF 36 scores on subscale ‘physical functioning’ was observed (P = 0.029) (Table 2). Control subjects showed stable scores and ME/CFS patients showed a decrease in scores over time, indicating postexertional malaise in the patient group. Studying the course of the CFS Symptom List total scores from baseline to 1 h postexercise and 24 h postexercise, ME/CFS patients showed a worsening of symptoms, whilst controls experienced no changes. The difference between the two groups was found significant (P < 0.001). Thus, the submaximal exercise triggered postexertional malaise in the ME/CFS group that cannot be attributed to their sedentary life style.

Table 2.   Evolution of health status in response to submaximal exercise in people with ME/CFS (n = 22) and healthy sedentary control subjects (n = 22)
 ME/CFS patients (mean ± standard deviation)Control subjects (mean ± standard deviation)Between groups comparison (F-value; P-value)Within groups comparison (F-value; P-value)
  1. CIS, Checklist Individual Strength; SF 36, Medical Outcomes Short Form 36 Health Status Survey; statistically significant results are printed in bold.

  2. Between group and within group comparisons were performed using two-way repeated measures anova.

Pain (mm)
 Pre-exercise45.7 ± 24.89.5 ± 14.7  
 Postexercise64.8 ± 24.710.2 ± 16.667.0; <0.00110.1; 0.003
 24 h postexercise67.8 ± 27.1 4.9 ± 6.6 81.0; <0.00129.9; <0.001
Fatigue (mm)
 Pre-exercise68.6 ± 14.215.6 ± 18.1  
 Postexercise78.3 ± 21.114.2 ± 21.4127.9; <0.0015.4; 0.025
 24 h postexercise84.0 ± 18.4 12.8 ± 18.8197.6; <0.0018.8; 0.005
Concentration difficulties (mm)
 Pre-exercise54.8 ± 24.45.3 ± 7.1  
 Postexercise62.8 ± 29.84.4 ± 5.396.9; <0.0013.6; 0.064
 24 h postexercise68.0 ± 28.0 5.9 ± 8.8 105.1; <0.0018.2; 0.006
CFS Symptom List total score (mm)
 Pre-exercise53.2 ± 12.110.6 ± 8.4  
 Postexercise58.1 ± 14.98.4 ± 8.7193.1; <0.00118.5; <0.001
 24 h postexercise63.4 ± 16.9 6.1 ± 7.7 229.1; <0.00127.5; <0.001
CIS fatigue
 Pre-exercise50.9 ± 5.820.9 ± 7.7  
 Postexercise 52.3 ± 5.518.5 ± 7.4342.3; <0.0013.4; 0.072
CIS concentration difficulties
 Pre-exercise28.4 ± 6.210.5 ± 4.5  
 Postexercise27.6 ± 7.69.4 ± 4.3116.9; <0.0010.9; 0.765
CIS motivation
 Pre-exercise15.5 ± 5.18.6 ± 4.0  
 Postexercise17.0 ± 6.78.0 ± 3.032.4; <0.0012.6; 0.117
CIS physical activity
 Pre-exercise15.5 ± 5.16.3 ± 2.7  
 Postexercise16.1 ± 5.66.3 ± 3.255.4; <0.0011.1; 0.305
SF 36 bodily pain
 Pre-exercise41.4 ± 17.6 89.3 ± 11.6  
 Postexercise45.0 ± 15.289.8 ± 12.5125.8; <0.0011.2; 0.271
 24 h postexercise41.1 ± 15.689.9 ± 11.8141.8; <0.0010.038; 0.846
SF-36 physical functioning
 Pre-exercise40.0 ± 16.790.2 ± 12.9  
 Postexercise39.1 ± 19.193.4 ± 7.5168.1; <0.0011.2; 0.271
 24 h postexercise34.3 ± 22.093.6 ± 8.5158.6; <0.0015.1; 0.029
SF-36 role limitations due to physical functioning
 Pre-exercise5.6 ± 10.790.9 ± 26.2  
 Postexercise9.1 ± 22.693.2 ± 22.1309.8; <0.0010.018; 0.893
 24 h postexercise7.5 ± 23.189.1 ± 28.2285.5; <0.0010.172; 0.680
SF-36 role limitations due to emotional problems
 Pre-exercise66.7 ± 44.893.9 ± 22.1  
 Postexercise62.1 ± 46.495.5 ± 15.68.61; 0.0052.02; 0.162
 24 h postexercise63.3 ± 48.293.9 ± 22.16.28; 0.0161.91; 0.175
SF-36 social functioning
 Pre-exercise36.4 ± 21.195.4 ± 9.9  
 Postexercise39.2 ± 24.596.6 ± 9.6148.3; <0.0010.136; 0.714
 24 h postexercise34.4 ± 22.595.5 ± 9.1154.7; <0.0010.122; 0.729
SF-36 mental health
 Pre-exercise63.1 ± 17.478.7 ± 8.2  
 Postexercise64.7 ± 20.580.4 ± 10.312.8; 0.0010.000; 0.983
 24 h postexercise66.0 ± 21.082.0 ± 9.111.9; 0.0010.340; 0.563
SF-36 vitality
 Pre-exercise30.9 ± 11.772.5 ± 11.1  
 Postexercise30.9 ± 14.976.8 ± 12.0149.0; <0.0013.43; 0.071
 24 h postexercise31.8 ± 16.777.9 ± 9.9166.0; <0.0011.01; 0.322
SF-36 general health perception
 Pre-exercise16.6 ± 9.663.6 ± 12.4  
 Postexercise18.0 ± 13.264.8 ± 9.5 220.4; <0.0010.008; 0.928
 24 h postexercise17.5 ± 10.661.6 ± 12.7209.5; <0.0011.31; 0.260

When comparing the CIS scores between both groups during experiment 2 (self paced physiologically limited exercise), we notice a difference in the subscales fatigue (P = 0.002), motivation (P = 0.038) and physical activity (P = 0.006) (Table 3). In the patient group, these subscale scores increase after exercise, whilst in the control group, they slightly decrease. The change over time in SF 36 subscale ‘physical functioning’ (P = 0.006) and CFS Symptom List total score (P = 0.001) was different between the two groups. This indicates that even the paced exercise bout with application of three safety breaks increased symptoms in the ME/CFS group. Hence, the women with ME/CFS studied here were unable to perform self-paced and physiologically limited exercise without symptom exacerbations.

Table 3.   Evolution of health status in response to self-paced, physiologically limited exercise in people with ME/CFS (n = 22) and healthy sedentary control subjects (n = 22)
 ME/CFS patients (mean ± standard deviation)Control subjects (mean ± standard deviation)Between groups comparison (F-value; P-value)Within groups comparison (F-value; P-value)
  1. CIS, Checklist Individual Strength; SF 36, Medical Outcomes Short Form 36 Health Status Survey; statistically significant results are printed in bold.

  2. Between group and within group comparisons were performed using two-way repeated measures anova.

Pain (mm)
 Pre-exercise61.9 ± 29.85.5 ± 8.9  
 Postexercise71.0 ± 26.3 3.3 ± 7.6103.7; <0.00117.6; <0.001
 24 h postexercise73.3 ± 25.45.8 ± 10.4109.7; <0.0014.9; 0.043
Fatigue (mm)
 Pre-exercise76.6 ± 18.610.7 ± 9.5  
 Postexercise84.5 ± 13.25.9 ± 6.1408.0; <0.00120.2; <0.001
 24 h postexercise87.4 ± 19.36.5 ± 8.2312.6; <0.00112.9; 0.001
Concentration difficulties (mm)
 Pre-exercise62.6 ± 26.25.8 ± 8.7  
 Postexercise66.5 ± 25.63.3 ± 6.2115.8; <0.0015.0; 0.031
 24 h postexercise68.4 ± 28.53.4 ± 6.898.4; <0.0017.2; 0.011
CFS Symptom List total score (mm)
 Pre-exercise57.3 ± 16.76.4 ± 5.7  
 Postexercise60.5 ± 14.64.2 ± 4.7241.4; <0.00113.3; 0.001
 24 h postexercise63.7 ± 18.74.5 ± 5.1196.6; <0.00110.6; 0.002
CIS fatigue
 Pre-exercise52.2 ± 3.620.1 ± 7.6  
 Postexercise52.8 ± 3.718.2 ± 8.1334.2; <0.00111.0; 0.002
CIS concentration difficulties
 Pre-exercise28.4 ± 5.510.3 ± 5.3  
 Postexercise29.0 ± 5.69.8 ± 4.3144.6; <0.0012.9; 0.096
CIS motivation
 Pre-exercise16.6 ± 7.18.7 ± 4.6  
 Postexercise17.0 ± 7.07.7 ± 4.023.7; <0.0014.6; 0.038
CIS physical activity
 Pre-exercise16.2 ± 4.9 6.7 ± 3.4  
 Postexercise16.5 ± 4.76.0 ± 3.163.7; <0.0018.2; 0.006
SF 36 bodily pain
 Pre-exercise41.8 ± 18.288.9 ± 11.1  
 Postexercise41.2 ± 17.589.4 ± 13.2108.2; <0.0010.590; 0.447
 24 h postexercise40.1 ± 15.990.2 ± 12.5116.6; <0.0013.8; 0.060
SF-36 physical functioning
 Pre-exercise39.6 ± 19.091.8 ± 9.3  
 Postexercise36.6 ± 19.392.6 ± 9.8188.0; <0.00129.4; <0.001
 24 h postexercise31.8 ± 19.092.4 ± 10.0160.4; <0.0018.3; 0.006
SF-36 role limitations due to physical functioning
 Pre-exercise6.8 ± 22.195.5 ± 21.3  
 Postexercise6.8 ± 22.191.7 ± 22.8 167.1; <0.0013.5; 0.069
 24 h postexercise1.3 ± 5.692.9 ± 22.6257.7; <0.001.545; 0.465
SF-36 role limitations due to emotional problems
 Pre-exercise62.1 ± 45.295.5 ± 21.3  
 Postexercise60.6 ± 44.493.7 ± 22.79.36; 0.0040.001; 0.974
 24 h postexercise58.3 ± 45.793.7 ± 22.79.72; 0.0030.558; 0.459
SF-36 social functioning
 Pre-exercise38.6 ± 23.894.9 ± 12.0  
 Postexercise37.8 ± 21.595.2 ± 12.8107.2; <0.0010.309; 0.581
 24 h postexercise33.4 ± 18.695.8 ± 12.1145.8; <0.0012.31; 0.137
SF-36 mental health
 Pre-exercise64.1 ± 19.080.9 ± 11.6  
 Postexercise64.0 ± 17.683.2 ± 8.516.0; <0.0012.59; 0.116
 24 h postexercise65.6 ± 18.181.3 ± 11.110.4; 0.0030.479; 0.493
SF-36 vitality
 Pre-exercise31.5 ± 16.775.2 ± 12.4  
 Postexercise32.7 ± 16.276.2 ± 11.399.7; <0.0010.013; 0.909
 24 h postexercise28.8 ± 14.377.9 ± 9.9154.3; <0.0011.23; 0.275
SF-36 general health perception
 Pre-exercise15.7 ± 10.561.6 ± 9.3  
 Postexercise15.9 ± 10.461.0 ± 10.7222.1; <0.0010.214; 0.646
 24 h postexercise15.0 ± 7.661.7 ± 10.2289.7; <0.0010.451; 0.506

Does submaximal exercise triggers elastase activity, interleukin-1β and complement C4a level increase in people with ME/CFS and healthy controls?

At baseline, elastase activity (Fig. 2), but not complement C4a level, was significantly higher amongst the ME/CFS patients (P = 0.033). Neither immune variables changed in response to the submaximal exercise (P > 0.05) (Table 4). Neither ELISA kits were able to detect bioactive IL-1β levels in either groups prior to or following the submaximal exercise (all data below the detection limit).

Figure 2.

 Change in elastase activity in response to submaximal exercise in women with ME/CFS (n = 22) and sedentary women (n = 22).

Table 4.   Evolution of elastase and complement C4a in people with ME/CFS (n = 22) and healthy controls (n = 22) prior to and following submaximal exercise
 ME/CFS patients (mean ± standard deviation)Control subjects (mean ± standard deviation)Between groups comparison (F-value; P-value)Within groups comparison (F-value; P-value)
  1. Between group and within group comparisons were performed using two-way repeated measures anova.

Elastase activity pre-exercise (U mg−1)110.7 ± 123.579.4 ± 64.40.371; 0.5460.729; 0.398
Elastase activity postexercise (U mg−1) 84.2 ± 67.778.8 ± 90.9  
Complement C4a pre-exercise (pg mL−1)1790.4 ± 424.31901.8 ± 398.40.238; 0.6281.32; 0.257
Complement C4a postexercise (pg mL−1)1645.9 ± 503.61654.5 ± 399.9  

Are people with ME/CFS able to perform self-paced and physiologically limited exercise without triggering elastase activity, IL-1β and complement C4a level increase?

Baseline elastase activity, but not complement C4a level, was found to be higher in the ME/CFS group compared with the control group (P = 0.013). However, neither immune variables changed in response to the self-paced and physiologically limited exercise (P > 0.05) (Table 5). There was no group × time interaction for either immune variables (refer to Fig. 3 for complement C4a level changes). Again, neither ELISA kits were able to detect bioactive interleukin-1β levels in either groups prior to or following the self-paced and physiologically limited exercise (all data below detection limit).

Table 5.   Evolution of elastase and complement C4a in people with ME/CFS (n = 22) and healthy controls (n = 22) prior to and following self-paced, physiologically limited exercise
 ME/CFS patients (mean ± standard deviation)Control subjects (mean ± standard deviation)Between groups comparison (F-value; P-value)Within groups comparison (F-value; P-value)
  1. Between group and within group comparisons were performed using two-way repeated measures anova.

Elastase activity pre-exercise (U mg−1)95.6 ± 74.864.5 ± 39.96.72; 0.0130.523; 0.474
Elastase activity postexercise (U mg−1)95.2 ± 90.457.2 ± 62.6  
Complement C4a pre-exercise (pg mL−1)1786.5 ± 371.12019.5 ± 335.10.238; 0.6281.32; 0.257
Complement C4a postexercise (pg mL−1)1564.7 ± 514.81778.4 ± 512.7  
Figure 3.

 Change in complement C4a level in response to self-paced, physiologically limited exercise in women with ME/CFS (n = 22) and sedentary women (n = 22).

When comparing the change in elastase activity and complement C4a level between both types of exercise (submaximal versus self-paced and physiologically limited exercise), no differences were observed in the ME/CFS or control group. To examine the potential bias originating from the daily physical activity level in the week between the two exercise bouts, the responses of the immune variables to the two types of exercise were re-examined using the average activity count of the second week as a covariate. This did not change the outcome of the analyses presented above (data not shown). In addition, neither the average activity counts during week 1 nor the total activity count on the day prior to experiment 1 were correlated with any of the immune variables (P > 0.05). This refutes the possibility that the results were biased by the daily physical activity level during the study period.

Are changes in elastase activity, IL-1β and complement C4a level associated with symptom exacerbations following exercise in people with ME/CFS?

In the ME/CFS group, the change in elastase activity following submaximal exercise was associated with the concentration and physical activity scores of the CIS at 1 h postexercise (Table 6). Postexercise elastase levels were related to the CIS fatigue subscale scores at 1 h postexercise. The change in complement C4a level was associated with the CIS fatigue scores, both prior to and 1 h following exercise. Postexercise C4a levels were associated with seven out of eight SF-36 subscale scores obtained prior to and following (1 h and 24 h) the submaximal exercise (P < 0.05).

Table 6.   Submaximal exercise: statistically significant associations between immune variables and self-reported health status in people with ME/CFS (n = 22)
 Postexercise elastase activity r (P)Change in elastase activity r (P)Pre-exercise C4a level r (P)Postexercise C4a level r (P)Change in C4a level r (P)
  1. CIS, Checklist Individual Strength; SF-36, Medical Outcomes Short Form 36 Health Status Survey.

  2. Associations between immune variables and self-reported measures were computed using Pearson correlation analyses.

CIS fatigue
 Pre-exercise  −0.40 (0.008)  −0.31 (0.038)
 1 h postexercise0.52 (0.034) −0.43 (0.004) −0.32 (0.034)
CIS concentration
 Pre-exercise  −0.37 (0.013)  
 1 h postexercise −0.37 (0.012)−0.34 (0.024)  
CIS physical activity
 1 h postexercise −0.34 (0.023)   
SF-36 general health
 1 h postexercise   0.38 (0.012) 
 24 h postexercise   0.36 (0.017) 
SF-36 social functioning
 24 h postexercise   0.45 (0.002) 
SF-36 role limitations due to physical functioning
 Pre-exercise   0.42 (0.003) 
 1 h postexercise   0.46 (0.002) 
 24 h postexercise   0.31 (0.049) 
SF-36 bodily pain
 1 h postexercise   0.36 (0.016) 
 24 h postexercise   0.32 (0.040) 
SF-36 physical functioning
 1 h postexercise   0.36 (0.017) 
 24 h postexercise   0.031 (0.050) 
SF-36 vitality
 1 h postexercise   0.41 (0.005) 
 24 h postexercise   0.36 (0.019) 
SF-36 social functioning
 1 h postexercise   0.42 (0.004) 
 24 h postexercise   0.36 (0.019) 

For the self-paced, physiological limited exercise, the following correlations were observed. Baseline elastase levels were associated to SF-36 physical functioning scores 24 h postexercise (r = −0.54; P = 0.015). Likewise, baseline elastase levels were associated with the change in SF-36 emotional functioning scores from baseline to 24 h postexercise (r = −0.59; P = 0.006). The change in complement C4a level was strongly related to the increase in pain (r = 0.52; P = 0.014) and fatigue (r = 0.43; P = 0.044) 24 h postexercise. Postexercise C4a levels were related to the change in bodily pain scores (SF-36) following exercise (r = 0.32; P = 0.039). Postexercise elastase activity level (r = −0.53; P = 0.011) and the change in elastase activity level (r = −0.47; P = 0.028) were both inversely related to the fatigue increase 1 h postexercise. No other associations between immune variables and symptom exacerbations or health status changes following either types of exercise were observed (P > 0.05). Given the lack of detectable bioactive IL-1β, we were unable to examine the dataset for associations between IL-1β and changes in health status.

The dataset was analysed to search for associations between the immune variables in the ME/CFS group. For both types of exercise, the pre-exercise elastase activity levels were associated with the change in complement C4a levels (submaximal exercise: r = 0.65: P = 0.002; self-paced, physiological limited exercise: r = 0.45; P = 0.037). No other associations were observed (P > 0.05).

Discussion

Both types of exercise triggered postexertional malaise in people with ME/CFS. However, neither types of exercise altered elastase activity, IL-1β or complement C4a split product levels in people with ME/CFS or healthy sedentary control subjects. Furthermore, the change in complement C4a level was strongly related to the increase in pain and fatigue 24 h following the self-paced, physiologically limited exercise. Postexercise elastase activity level and the change in elastase activity level were inversely related to the fatigue increase 1 h following the self-paced, physiologically limited exercise. The level of complement C4a following submaximal exercise was identified as a clinically important biomarker of postexertional malaise in people with ME/CFS.

Are people with ME/CFS able to perform self-paced, physiologically limited exercise without symptom exacerbations?

The self-paced, physiologically limited exercise protocol applied strategies used to implement ‘safe’ exercise therapy and self-management for people with ME/CFS [2, 15]. In a previous study, we applied a similar but less stringent approach to limit postexertional malaise in people with ME/CFS: it was shown that the use of exercise limits (limiting both the intensity and duration of exercise) prevents important health status changes following a walking exercise in people with ME/CFS, but was unable to prevent short-term symptom increases [14]. Based on that experience, we made the exercise limits more stringent by decreasing the intensity to 80% of the heart rate corresponding to the anaerobic threshold of the first (submaximal) exercise. Nevertheless, it was unable to prevent postexertional malaise in the women with ME/CFS studied here. These results highlight the fact one should be cautious when using exercise in people with ME/CFS. More work is required to address the issue of prevention of postexertional malaise in people with ME/CFS, for instance by using exercise limits to maintain lower intensity of exercise as previously used in randomized controlled clinical trials of exercise therapy in people with ME/CFS (e.g. exercise intensity based on the heart rate value obtained midpoint during a submaximal exercise test [2] or the heart rate corresponding to 40% of peak oxygen consumption during a maximal exercise test [1, 49]).

Does exercise triggers elastase activity, IL-1β and complement C4a level increase in people with ME/CFS and healthy controls?

We were unable to find evidence in support of immune alterations in response to either type of exercise, in either group. This might indicate that the biological parameters measured are not involved in the pathogenesis of postexertional malaise. However, previous studies showed that people with ME/CFS respond to an exercise challenge with increased expression of the lectin pathway (C4 and mannan-binding lectin serine protease 2) in PBMC’s, resulting in significant increase of C4a spilt product [10, 16]. However, our results are not in contradiction with those earlier reports. First, the increase in complement C4a split product became apparent at 6 h after exercise [10]. In the present study, peripheral blood levels of C4a were measured only at 1 h after exercise, a time-point at which Sorensen et al. [10] were unable to find changes in circulating C4a levels either. Secondly, the inability to detect changes in IL-1β in response to exercise is consistent with their earlier report [10]. Other attempts to monitor the biology of postexertional malaise revealed that exercise in people with ME/CFS does not alter F2-isoprostane levels, the IL-6 signalling system or tumour necrosis factor α [50, 51]. However, vigorous exercise in people with ME/CFS results in accentuated oxidative stress, probably as a result of delayed and insufficient heat shock proteins production, and subsequent increased fatigue and musculoskeletal pain [5, 51].

Previous studies indicated that the inflammatory response of people with ME/CFS does not differ from that seen in healthy controls [51, 52]. This notion is underscored by our findings of undetectable IL-1β levels and unaltered elastase activity in response to submaximal or self-paced, physiologically limited exercise in ME/CFS patients. Elastase is a proteolytic enzyme produced by monocytes and neutrophils during inflammatory response. Ribonuclease L is one of the intracellular proteins activated by type I interferons and cardinal to the cellular defense mechanism [53]. Elastase has the capacity to initiate high molecular weight ribonuclease L (83 kDa) proteolysis [18, 19] and hence to cause (further) dysregulation of the 2′-5′ oligoadenylate synthetase/ribonuclease L pathway within PBMC’s of ME/CFS patients [54]. Therefore, identification of the triggers of elastase activiation would be important in understanding the pathophysiology of ME/CFS. However, elastase activity was not triggered in either groups, suggesting that exercise will not have delirious consequences on the ribonuclease L enzyme either, although this requires further empirical testing.

Are changes in elastase activity, IL-1β and complement C4a associated with symptom exacerbations following exercise in people with ME/CFS?

Our finding of a strong relation between the change in complement C4a level and the increase in pain and fatigue 24 h following the self-paced, physiologically limited exercise support the use of C4a as a marker for postexertional malaise in people with ME/CFS. This finding is in line with the previous findings from Sorensen et al. [10] who reported statistically significant correlations between the increase in C4a and total symptom score and individual symptoms like headache, joint problems and thinking difficulty. In general, a high number of associations between self-reported measures and both elastase activity and C4a levels were found, supporting the clinical importance of both immune markers for people with ME/CFS. These findings are in line with previous studies [10, 17, 20]. Given the high number of correlations observed with postexercise SF-36 subscale scores, the present study suggests that the C4a level at 1 h following the submaximal exercise represents an important marker of postexertional malaise in people with ME/CFS.

Study limitations and strengths

The results should be interpreted in relation to its limitations. To account for bias originating of pooling gender data [32], only women were studied. In addition, only ME/CFS patients experiencing chronic widespread pain were included. This limits the external validity of the results to women with ME/CFS experiencing chronic widespread pain. Still, this represents the majority of sufferers. Secondly, the immune variables were measured 1 h postexercise. It remains possible that changes in some of the immune variables studied here occur later in response to exercise. Thirdly, we were unable to randomly allocate the participants to the two phases of the study: the results obtained from the first exercise test were used to pace the second exercise bout. On the other hand, the study has several strengths. The patient group was sufficiently powered and selectively chosen to study an important and debilitating subgroup within the ME/CFS population. Using real-time physical activity monitoring allowed us to account for potential bias resulting from physical activities performed by the patients prior to the first exercise bout and between exercise sessions. A final strength worth mentioning is that the study strongly builds on previous work in the area of ME/CFS: the submaximal exercise protocol is the sole protocol known to be reliable and valid for testing people with ME/CFS [43, 45]; the self-paced, physiologically limited exercise protocol applied strategies used to implement ‘safe’ exercise therapy and self-management for people with ME/CFS [2, 15]; elastase and complement C4a were previously identified as correlates of exercise performance or postexertional malaise in people with ME/CFS [10, 20]; and postexertional malaise itself has been identified as a primary and disease-specific characteristic evident in up to 95% of people with ME/CFS [9–12].

Conclusion

It is concluded that submaximal exercise as well as self-paced, physiologically limited exercise triggers postexertional malaise in people with ME/CFS, but neither types of exercise alter circulating levels of IL-1β, complement C4a split product or elastase activity. Postexercise complement C4a level was identified as a biomarker for postexertional malaise in people with ME/CFS. Further studying of immune alterations in relation to postexertional malaise in people with ME/CFS using multiple measurement points postexercise is required.

Conflict of interest statement

No conflict of interest was declared.

Acknowledgements

The study was funded by ME Research UK (MERUK), a national charity funding biomedical research into Myalgic Encephalomyelitis/Chronic Fatigue Syndrome. Jessica Van Oosterwijck is financially supported by grant no. OZR1596 from the research council of the Vrije Universiteit Brussel, Brussels, Belgium. Mira Meeus is financially supported by the Research Foundation Flanders, Belgium. The authors are grateful to Lieve De Hauwere for taking the blood samples and assisting with the exercise testing.

Ancillary