Review 2
Forty six studies with a total of 10921 participants were included in the second review. Studies included were published between June 2021 to October 2022 (17 from USA and 21 from Europe). See Table 3 for a full quality assessment of all publications included in Review 2. Of these, 19 and 15 were case controlled and cohort studies with control groups respectively. All studies included patients not hospitalised during their acute COVID-19 event with 21 studies clearly including patients hospitalised for acute COVID-19. Four studies recruited less than 25 patients 36–39; whilst another four included more than 1000 participants. 40–43
Overall, the quality of reporting across the studies included in this second review was moderate to poor. In particular, only four studies reported a sample size justification 43–46 and only two studies randomised the control group. 41,47
Most studies (n = 38) used a control group that was at least similar to the LC population; but nine studies did not clearly differentiate their control and intervention groups. 42,48–55 Most studies were able to determine that presenting symptoms developed after acute covid (ie were not existing symptoms prior to infection) but just eleven considered potential confounding variables in their analysis. 40,52,53,56–63 One study did not present a clear research question or objective, 59 with only seven studies analysed without the assessors knowing to which study group each participant belonged. 45,48,49,52,55,64,65
Symptoms strongly linked to identifiable pathophysiological pathway.
Immune system
The link between SARS-CoV2 infection and immune dysregulation has been established in a number of studies in the acute phase. 66 This relationship appears to be part of the development of LC; albeit the relationship between LC and the immune system is complex.
Symptoms of fatigue, POTS, and neurocognitive dysfunction have all been linked to immune dysregulation. Some studies 47,59,62,63,67–69 and systematic reviews 19,21,26–28,31,33–35,70 suggested relationship between Interleukins (IL) 6 and 8, IP 10 and TNF alpha with LC symptoms of fatigue and neurocognitive dysfunction.
In one study, participants with LC symptoms of fatigue, brain fog, memory difficulty, confusion, and POTS also displayed altered humoral responses to distinct herpes viruses including the Epstein-Barr Virus (EBV) minor viral capsid antigen gp23, the EBV fusion receptor component gp42, and the VZV glycoprotein E. 58
Viral persistence with reduced viral cleavage leading up to higher levels of monocytes,71 or maintained T cell response to SARS-CoV2 in LC patients has been reported in more recent papers;37 whilst a systematic review by Akbarialiabad in 2021 suggested that oxidative stress leading to weak immune response and incomplete viral eradication is associated with LC symptoms including neural loss associated with the finding of SARS-CoV2 RNA in the CNS.33 In a small study of 49 LC patients mainly suffering from systemic symptoms, chest pain or fatigue compared with 16 controls, the level of CD4 count was found to be lower in the LC population.64 Similar findings implicating the CD8 pathway were noted by a case-controlled study of 30 LC patients and 20 controlled with significant increase in level of CD8 + T cells for LC group relative to the recovered group. 72
Another study by Lee and colleagues associated symptoms of fatigue and dyspnoea to the involvement of interferon responsive genes in the pathophysiology of COVID-19 indicating a possible link to systemic autoimmune diseases. 52 Additionally, one systematic review including studies of hospitalised and non-hospitalised patients with COVID-19 infection identified complex pathways to LC. These pathways included Micro and Macro vascular inflammation associated with increased level of cytokines, circulating endothelial cells, coagulation activation, and microvascular retinal impairment. Some studies included in this review found biopsy evidence of lymphocytic or neutrophilic infiltrates, endotheliitis, microangiopathy, and microthrombosis. 21
Finally, one study using a metabolic phenotyping approach on stored sera from non-hospitalised LC patients compared healthy controls and sera from hospitalised acute COVID-19 patients. Results were indicative of ongoing inflammation, cellular damage, and immune activity in the LC patients (raised taurine, 3-indole acetic acid, glutamine/glutamate ratios). 51
Central nervous system
Neurocognitive symptoms are one of the most commonly reported findings in LC. Three systematic reviews23,25,30 and four recent papers73–76 reported links between neurocognitive symptoms and brain pathophysiology. Whilst there is evidence of brain microbleeds and structural damage in patients dying of severe COVID-19, evidence in the non-hospitalised population takes the form of positron Emission Tomographic (PET) imaging. One high quality-controlled PET MRI study included in a review of the relationships in LC sufferers between pain (including headache) 23and general cognitive function30 found persistent focal cerebral hypometabolism (frontal, temporal, brainstem and cerebellum) in this group. Two additional studies show similar findings although a large proportion of patients in these studies had been hospitalised.39,46 A case-control study of 26 patients with new onset neurocognitive and mood symptoms and 125 healthy retrospective controls used PET-MRI. Hypometabolism of the fronto-insular cortex along with hypermetabolism of the limbic system correlated with both markers of inflammation in the acute phase and the Mini-Mental State Examination.74
Small preliminary studies have also suggested changes in cerebral blood flow possibly due to endothelial dysfunction. Other studies included in the review found that small white matter lesions on imaging were associated with delayed recovery from mild COVID-19. 23 A study of cognitive impairment and high resolution MRI in mildly affected patients with COVID-19 compared with healthy volunteers found significant cortical thinning in areas of language and difficulties with verbal memory 73. Further study of brain post-mortem samples in patients who had died showed that SARS-CoV-2 was capable of infecting astrocytes, and that infected astrocytes impacted neuronal viability in vitro. Further evidence of structural changes in the brain comes from a case-control study comparing high resolution MRI and neuropsychiatric instruments scores in 30 LC subjects with neuropsychiatric symptoms and 20 healthy volunteer controls. 75 The study found significant associations between increased grey matter volume and depressive and cognitive symptoms. A review of depression symptoms after covid included two prospective cohort studies 25 that found associations between depressive symptoms and poorer performance on selective attention and processing speed, immediate recall, visual reaction times, executive function, and visuospatial abilities. The direction of causality in these studies cannot be determined.
Linking neurocognitive LC symptoms with abnormalities in the immune response, a case control study showed wide-ranging differences in the immune response compared with healthy controls.76 Broad activation of T helper cells to SARS-CoV-2 nucleocapsid proteins and impaired CD8 T cell memory were found and also found to be associated with the severity of cognitive impairment.
Respiratory system
Pulmonary involvement due to SARS-CoV2 infection with the resulting respiratory failure and acute respiratory distress syndrome (ARDS) are considered to be the leading cause of death from acute COVID-19. 77 Therefore, lung damage as a consequence, akin to SARS-1 infection, was a concern early in the pandemic. 78
In this review, LC patients with persisting dyspnoea, cough and chest pain following COVID-19 infection were found in a case-controlled study to have significantly lower FVC%, TLC% and DLCO% compared to controls. 79 Other studies also demonstrated similar findings. 38,54,65
Impaired gas transfer was also demonstrated by Xenon MRI scanning in LC patients many months after the acute infection in both hospitalised and non-hospitalised patients despite normal CT imaging of the lungs. 55 A number of systematic reviews also described this association between symptoms of LC (mainly breathlessness and cough) with lung damage demonstrated by either imaging or impaired lung function. 22,24,70 Although lung damage leading to impaired gas transfer has been demonstrated in these studies and systematic reviews as a pathophysiology underpinning symptoms of dyspnoea, cough, or fatigue, there is still little evidence describing the causality of this lung damage, albeit micro-clots and endothelial damage have been proposed by some. 65
Other direct organ system damage
Our review identified two systematic reviews 24,32 and three new primary studies 42,80,81 linking non-respiratory organ damage to disease pathophysiology.
A systematic review of LC effects on multiple systems found a number of imaging studies in LC patients, although most were uncontrolled cohorts, and the majority were probably in hospitalised patients.24 This review pointed to myocarditis and peri-myocarditis being the primary areas of concern outside the lungs. A high quality systematic review specifically looking at this issue found 35 studies (22 solely in hospitalised patients) reporting on a variety of cardiac assessments including MRI, echocardiography, NT ProBNP, ECG endomyocardial biopsy and angiography.32 Cardiac MRI (cMRI) evidence of myocarditis was found in 0–37% of patients in studies whereas ECG and echocardiography were usually normal. It was unclear from the review what the impact of hospitalisation was on pathophysiology. Four controlled individual studies have examined the impact of LC on the heart,42,45,80,82,83 whilst another study focused on the heart and other organs.81 Mostly cMRI with perfusion and gadolinium has been used. A study of military personnel with LC symptoms compared to healthy controls found that 4/50 of cases met the criteria for myocarditis at 1–4 months post infection.80 Follow up scans showed resolution in ¾ of cases by 9 months. One case of Takotsubo cardiomyopathy was found (ischaemic damage without any evidence of coronary artery disease). A prospective cohort study in volunteers at the point of COVID-19 infection found reduced exercise capacity on cardiopulmonary exercise testing in those developing LC symptoms.80 The study found this reduced capacity was associated with increased perceived effort and chronotropic incompetence. There was no evidence of abnormalities on cMRI (stress-perfusion MRI) or rhythm monitoring, reduced exercise capacity was associated with raised CRP, IL-6 and TNF alpha, linking back to a non-cardiac cause. Two smaller studies45,82 also found no significant evidence on cMRI. A much larger study involving whole body MRI in 536 individuals with LC and 92 healthy controls found evidence that mild structural organ damage was relatively common (myocarditis 9% pancreas 9% kidney 15% Liver 11%) and two thirds had at least one organ inflamed at the initial scan although these changes were not associated with symptomology statistically. 81
Symptoms with reasonable link to identifiable pathophysiological pathway
Endothelial and blood clotting disturbances
Three systematic reviews21,26,29 and three studies39,41,84 have established a relationship between endothelial dysfunction and LC. One additional high quality study has shown abnormalities in blood clotting mechanisms.50 A systematic review of LC identified a number of studies where macro- and microvascular thrombosis played a part in pulmonary and cardiovascular complications, but it is unclear to what extent these occurred in hospitalised patients alone.29 A review of erectile dysfunction after COVID-19 noted that endothelial function is necessary for erection, but the papers within this review were not quality scored nor were they explicitly looking at erectile function.26
In a more recent study, PET was added to whole body imaging in a case-controlled study of a small sample of 13 patients in a LC clinic compared with non-contemporaneous controls (cancer patients). Mild/moderate increased uptake was seen in vessels, bone marrow and joints in LC patients, but no correlations were found with symptoms though this analysis would likely lack power. 39 The effects of SARS-CoV-2 on the vascular endothelium (measured by flow mediated skin fluorescence, FMSF) and exercise capacity has been investigated in a 49 patients from a LC clinic who were compared with healthy controls and a group of amateur runners. 84 FSMF measures Nitric Oxide mediated vessel dilation in response to reactive hyperaemia and small vessel oscillations. High intensity exercise and LC had similar effects on blood vessel function, except that the effect of high intensity exercise resolved after several hours. Although no direct correlations with symptoms were made, the researchers noted that the vascular changes after exercise are known to be associated with post-exercise fatigue. In the only RCT in our review, volunteer patients with at least 4 weeks of symptoms post-COVID were randomised to multivitamins or L-arginine and high dose Vitamin C, proposed as being able to reduce oxidative stress and improve endothelial function. There were significant improvements after 30 days in a range of LC related symptoms and in Borg Scale perceived effort. 41Micro-thrombi in the pulmonary vasculature are well recognised in patients who have died of acute COVID-19. A case control study in patients attending a LC clinic was conducted to determine the role of endothelial damage using biomarkers for endothelial cell damage (von Willebrand Factor Antigen (VWF:Wg), VWF propeptide (VWFpp) and soluble thrombomodulin (sTM)). The study included 13/50 patients who had not been hospitalised. Levels of all these markers and Factor VIIIC were found to be significantly raised in LC patients. Abnormalities were associated with but not restricted to hospital admission in the acute phase. An attempted correlation with symptoms was underpowered.
Mast cells
Mast cells may become activated as a side-effect of T cell activation and many LC patients report new allergic responses, wheeze, gastrointestinal symptoms and urticaria. There is no accepted international definition of mast cell activation, with some authors proposing a symptom-based definition, others restricting the term to the identification of raised serum tryptophan (itself transient and not widely available as an assay). One paper using the clinical definition found that LC patients had similar Mast Cell Mediator Release Scores to clinically confirmed Mast Cell patients. 85