Reply to: Should we be careful with exercise in post-exertional malaise after Long COVID?

replying to C. G. J. Saris et al. Nature Communications https://doi.org/10.1038/s41467-025-56427-3 (2025)

We thank Saris et al. for their interest in our recent paper¹, which focused on the skeletal muscle components of post-exertional malaise (PEM) in Long COVID patients. We did not address the efficacy of exercise as a treatment, but our findings do not support that deconditioning or bed rest explains the skeletal muscle abnormalities in Long COVID. PEM is a distinct symptom that is poorly understood and not analogous to other conditions. We caution against generalizing our results to all Long COVID patients or drawing causal conclusions beyond the study’s design, and we advocate adherence to established guidelines that recommend avoiding activities triggering PEM.

Whilst we recognize the exercise therapy experience of Saris et al. in patients with mitochondrial and neuromuscular diseases, we firmly reject the notion that our findings regarding post-exertional malaise (PEM) are not specific to Long COVID, nor that we address “how to treat this condition”. Saris et al. claim that “improvements in muscle fitness can act as therapy in Long COVID”, based on assumptions and experiences with other diseases. PEM is a specific symptom of many post-viral infection diseases (including Long COVID and ME/CFS) which is poorly understood, and poorly recognized by clinicians. As such, intense exercise above an unknown time- and patient-dependent PEM-inducing threshold is often avoided by patients, because it worsens their Long COVID symptoms. Therefore, it was the explicit goal of our study to examine the systemic and skeletal muscle component of PEM in Long COVID.

Importantly, our publication did not provide any statement about the merit or avoidance of exercise in treating patients with Long COVID, and we are therefore surprised that Saris and colleagues concluded this. We clearly understand that frequency, intensity and duration are important parameters of exercise training in improving aerobic fitness in healthy people². However, patients with Long COVID and PEM are unable to functionally recover completely after an exercise bout at a high intensity or duration, limiting their ability to frequently perform such exercise bouts. The interpretation of exercise-related results in Long COVID critically depends on which patients are included. Our patients could work ~5 h/week, and all experienced PEM, which is a key symptom in 87% of all Long COVID patients³. Hence, our results should not be extrapolated to all Long COVID patients (i.e., those not suffering from PEM or with less debilitating symptoms, such as in a recent study⁴). Further, our results cannot be extrapolated to applicability of exercise training for treatment, simply because we did not study this.

Deconditioning has been proposed as causative of skeletal muscle changes in Long COVID patients⁵. Our previous work on bed rest demonstrates that adaptations in skeletal muscle structure and function differ markedly between Long COVID patients and healthy persons exposed to severe deconditioning^6,7. Bed rest induces rapid muscle atrophy and loss of capillarization⁷, which was not observed in our Long COVID patients¹. We did not observe alterations in fiber type composition after 60 days of strict bed rest⁷, whereas our Long COVID patients exhibited a more glycolytic phenotype. Our Long COVID participants showed lower mitochondrial respiration, due to mitochondrial dysfunction (Fig. S5)¹ rather than mitochondrial loss, the latter of which is more typically seen after long-term bed rest^6,7,8. The absence of a correlation between ${\dot{V}O}_{2\max }$ and succinate dehydrogenase activity also suggests that factors other than deconditioning contribute.

Saris et al. criticize us for not providing an adequate deconditioned control group. We acknowledge that deconditioning confounds the conclusions related to the underlying cause of a lower exercise capacity, and that an age-, sex-, and activity-matched control group would be superior to age and sex alone. Althoff et al.⁹ showed that the average daily steps in the United States is 4774 per day, roughly similar to our Long COVID group, but PEM is not experienced in the general US population. As can be appreciated from Figure 1A, there was considerable variability in exercise capacity, as some patients still possessed fitness levels >50th percentile for their sex and age (n = 5), while some healthy controls had very low fitness levels (<25th percentile, n = 7). Yet, all patients exhibited PEM and healthy controls did not, irrespective of fitness. Additional analysis matching participants for step count (5181 vs. 4727 steps/day for patients and controls, respectively) still resulted in a significant group difference for ${\dot{V}O}_{2\max }$ (−24%, p = 0.004), peak power (−31%, p = 0.043) and gas exchange threshold (−20%, p = 0.046; effort-independent). These data strongly suggest deconditioning alone does not cause the skeletal muscle abnormalities.

Our results of impaired mitochondrial function and skeletal muscle myopathy in Long COVID patients are in line with other recent publications^4,5,10,11,12. The causal role of deconditioning in explaining skeletal muscle adaptations in Long COVID is impossible to ascertain from our study design, as we acknowledged in the paper. We do not know the underlying factors contributing to these changes, and as good scientific practice, we refrained from drawing causal relationships in our publication. We were therefore surprised to read that Saris et al. drew such conclusions based on work which was not set up to be causal in nature. Recently, we and others described many possible factors contributing to Long COVID symptomology, including viral persistence, immune cell and autonomous nervous system dysregulation, and auto-immunity^5,12,13,14.

Saris et al. question the muscle damage following exercise. Biopsies were scored for pathological findings by a group-blinded pathologist, where the tissue needed to contain more than a “normal” number of atrophied fibers (2–4% in healthy people), regenerative fibers, or infiltrating macrophages and T-cells. As such, scoring negative did not mean that there was no staining at all. Also, the size of the biopsy could not be controlled for. We are aware of exercise-induced muscle damage, and the occurrence of exertional rhabdomyolysis in patients with mitochondrial diseases^15,16. Saris et al. suggest that atrophied fibers cannot occur one-day following exercise, but the reference they provide¹⁷ proves the opposite. Hikida et al.¹⁷ show degeneration of myofibrils, and occurrence of atrophic fibers along with abnormal mitochondrial profiles one day following a marathon (which we would not refer to as “training and normal physiology”). Before PEM induction, patients show higher number of atrophied fibers, higher macrophage infiltration, higher T-cell infiltration and more cells with internalized nuclei compared to controls, similar to results by Hejbøl et al.¹¹. Since some of these were already different in the pre-PEM biopsies, we agree that this cannot be associated with our PEM induction, but we do not exclude remnants of previous PEM episodes, since the time-course of muscle regeneration can be slow.

In sum, we agree with the conclusion of Saris et al. that “research on the cause of decreased aerobic capacity in long COVID and similar post-viral conditions is needed”, and feel our work makes a valuable contribution to this need. However, we strongly encourage Saris and colleagues not to draw causal conclusions that were not evidenced or supported by our work. It is essential to realize the infancy of PEM research and that previous findings in other diseases may not apply to PEM in Long COVID. We also agree that “better therapies should become available”, as this is a clear wish for scientists, clinicians, and patients. The established guidelines that Saris et al. refer to, in fact, include specific statements about avoiding “exercise and activities that cause PEM and aim to conserve your energy”¹⁸. While we value and promote exercise below the PEM-inducing threshold for patients with Long COVID with and without PEM¹⁸, this was not the topic of our publication, and therefore we refrained from making conclusive statements. Because of the occurrence of PEM, intense exercise training above the PEM threshold is unlikely to be a curative treatment for Long COVID patients with PEM.