Lactate dehydrogenase contribution to symptom persistence in long COVID: A pooled analysis

There's critical need for risk predictors in long COVID. This meta‐analysis evaluates the evidence for an association between plasma lactate dehydrogenase (LDH) and long COVID and explores the contribution of LDH to symptoms persistent across the distinct post‐acute sequelae of COVID‐19 (PASC) domains. PubMed, EMBASE, Web of Science, and Google Scholar were searched for articles published up to 20 March 2023 for studies that reported data on LDH levels in COVID‐19 survivors with and without PASC. Random‐effect meta‐analysis was employed to estimate the standardized mean difference (SMD) with corresponding 95% confidence interval of each outcome. There were a total of 8289 study participants (3338 PASC vs. 4951 controls) from 46 studies. Our meta‐analysis compared to the controls showed a significant association between LDH elevation and Resp‐PASC [SMD = 1.07, 95%CI = 0.72, 1.41, p = 0.01] but not Cardio‐PASC [SMD = 1.79, 95%CI = −0.02, 3.61, p = 0.05], Neuro‐PASC [SMD = 0.19, 95%CI = −0.24, 0.61, p = 0.40], and Gastrointestinal‐PASC [SMD = 0.45, 95%CI = −1.08, 1.98, p = 0.56]. This meta‐analysis suggests elevated LDH can be used for predicting Resp‐PASC, but not Cardio‐PASC, Neuro‐PASC or gastrointestinal‐PASC. Thus, elevated plasma LDH following COVID infection may be considered as a disease biomarker.


| INTRODUCTION
More than three years into the COVID-19 pandemic, and at least 65 million people are believed to be living with long COVID (also known as post-acute sequelae of COVID-19 (PASC) as at January 2023. 1 According to the Centre for Disease Control, long COVID is defined as a post-COVID-19 condition in which there is new, evolving or persistent health issues occurring ≥4 weeks after SARS-CoV-2 infection 2 Emerging evidence reports that fatigue (27.4%), anosmia (25%), impaired lung diffusion capacity (24.6%), alopecia (10.2%), and dyspnoea (10.1%) are the most frequent symptoms and findings in COVID-19 survivors 2 years after SARS-CoV-2 infection. 3ctate dehydrogenase (LDH) is a predominantly cytoplasmic (oxidoreductase) enzyme of cytosol of glycolytic cells and tissues; specifically, LDH catalyses the conversion of pyruvate to lactate using NADH (reduced nicotinamide adenine dinucleotide). 4 LDH also appears in the mitochondrial reticulum where it catalyses the oxidation of lactate to pyruvate for entry into the TCA Cycle.6][7] Elevated lactate level is a known marker of inflammation and is believed to generate a proinflammatory microenvironment. 8Long COVID symptoms have been demonstrated to have a crippling effect with significant impact on patient's quality of life. 1 Globally, labour force has struggled owing to a good number of long COVID patients that are yet to return to their original work and presently, there are no known effective pharmacological or non-pharmacological interventions. 1,3,9Though myriads of pathophysiological factors have been implicated including dysregulated immune response that results in some degree of tissue damage, 10 the exact cause of residual symptoms remains unclear. 11evious studies had shown that raised plasma LDH is associated with the risk of having severe.COVID-19. 12,134][15][16] However, as the included studies were carried out during acute SARS-CoV-2 infection, these reviews did not contribute to our understanding of the role of LDH elevation regarding symptom persistence during recovery in COVID-19 survivors.Presently, various biomarkers are under investigation for their role in predicting who will develop PASC following acute COVID-19.Despite paucity of data on LDH association with long COVID, the existing evidence on the predictive role of LDH for pulmonary fibrosis makes it an essential area of research focus. 17Therefore, we conducted an evidence synthesis to investigate LDH contribution to symptom persistence in long COVID, especially its differential contribution to distinct system domains of PASC.

| METHODS
The present systematic review and meta-analysis followed methodological guidelines from the Cochrane Handbook for Systematic Reviews and reported according to the Preferred Reporting Items for Systematic Review and Meta-Analyses statement 2020 (Supplementary Figure 1). 18This systematic review was registered with the International Prospective Register of Systematic Reviews (PROS-PERO ID: CRD42023431873).

| Search strategy
Two researchers (RU and XD) electronically searched three databases including PubMed/MEDLINE, Web of Science, and Google Scholar applying the keywords "LDH" OR "LDH" AND "long COVID" OR "PASC" OR "PASC" up to 20 March 2023 and updated afterwards through periodic search (Supplementary Table 1).The title, abstract and full text of all the articles retrieved were analysed and for the purpose of our meta-analysis, only those that reported data on LDH values of adult COVID-19 convalescents with and without PASC.
Moreover, to find all eligible articles, we searched all reference lists of included studies.Additionally, no language restrictions were applied in our systematic search.

| Eligibility criteria
The present systematic review and meta-analysis followed inclusion criteria according to the PICOs criteria. 18Eligibility criteria include confirmed case of SARS-CoV-2 infection, group of patients that reported persistent or new symptoms at least 1 month after discharge, compared with another group that did not report any new or persistent symptom and has information on the plasma LDH level for both groups.It could be a prospective, retrospective or crosssectional study.Studies that evaluated review articles, commentaries, case reports and case series were excluded (although their reference lists were explored for potentially eligible studies).

| Data extraction and quality assessment
The following data were extracted from the eligible studies: name of first author, date of publication, study location, study design, study size, demography (age and gender), data timepoint, plasma LDH level in different groups.The variable of interest in our pooled analysis was plasma LDH level in mean � SD or median (interquartile range), respiratory-PASC, neurological-PASC, cardiovascular-PASC, and gastrointestinal-PASC from PASC and non-PASC groups.Quality assessments of all included studies were performed using Newcastle-Ottawa Scale (NOS) for prospective and retrospective studies, 19 and Joanne Briggs Institute critical appraisal tool for cross-sectional studies. 20Studies will be considered of good, moderate or poor quality if they scored >7, 5-7 or <5 respectively.Two reviewers (RU and XD) independently extracted data using a standardized dataextraction form and assessed the quality of included studies.Any discrepancies were settled by a third researcher (MM or TK) before meta-analysis.

| Statistical analysis
The meta-analysis was conducted using STATA version 17 (Stata Corp LP, College Station, TX, USA) to estimate the standardized mean difference (SMD) and 95% confidence interval (95% CI) of the LDH values in COVID-19 survivors with and without PASC.Standardized mean difference cut-off points of 0.2, 0.5 or 0.8 were considered to be of small, moderate or large effect sizes respectively. 21The degree of between-study heterogeneity that could not be ascribed to sampling error was explored using Cochran's Q statistics and I 2 statistics.A random effects model was performed to assess if heterogeneity (I 2 statistics) was substantial (>50%) in the meta-analysis.For studies that reported LDH level as median, we estimated the mean and Standard deviation from sample size, median, interquartile range, minimum, maximum values. 22We performed three sets of subgroup analyses where possible.We conducted subgroup analysis by different study design (cohorts vs. non-cohorts), different follow-up duration (<3, 3-6, and 6-12 months), gender in terms of female proportion (<50% vs. >50%), different age categories (<50 years vs. >50 years) and sample size (<100 vs. >100).Sensitivity analysis was conducted by leaving a study out each time via influence analysis while observing the stability of our pooled estimate.Funnel plot asymmetry and Egger's regression test 23 were used to assess publication bias.p value less than 5% was considered to be statistically significant.

| Study identification and characteristics of studies
Initial literature search yielded 2031 records, and 581 studies were later excluded as they were found to be duplicates.Further 1364 studies were excluded after review of the titles and abstracts and 86 studies were left.Following full-text review, 40 studies were excluded as they either had no data on LDH, were review papers or did not provide extractable data.Finally, 46 studies with 8289 participants (3338 PASC and 4951 controls) were included in our pooled analysis (Table 1).  Thesincluded studies were all peerreviewed, published between 2020 and 2023; had data from 28 studies in Resp-PASC, 10 studies in Neuro-PASC, 5 studies in Cardio-PASC, 2 studies in gastrointestinal-PASC and as only 1 study was in Reno-PASC.Thirty-four studies were of cohort design, 11 were case control and 1 cross-sectional.Overall, 15 studies were conducted in China, Spain (8), USA (4), Italy (4), UK (3), Germany (3), Turkey (2) and 1 study each in Austria, Russia, Romania, Colombia, Republic of Korea, Iran and India [Table 1].The duration of follow-up varied among the studies from 2 to 12 months.Majority of the studies reported ICU admission of 4%-100% except a few studies that had mostly mildly COVID-19 cases in their cohorts during acute infection. 28,41,56,68Twenty-four included studies were rated 'good' while 22 studies were rated to be of 'moderate score' based on the criteria outlined in NOS and Joanna Briggs Institute Critical Appraisal checklist [Supplementary Table 2A and 2B].A total of 17% of included studies reported the LDH analytical tool employed in their study (Table 1).Immunocompromised status was reported as matched in 37% of included studies or excluded from the cohort in 9% of the studies (Table 1).1).The overall pooled effect estimate showed a large effect size, with a significantly higher serum LDH concentration in COVID-19 convalescents with Resp-PASC relative to the controls (SMD = 1.07, 95%CI = 0.72, 1.41, p = 0.01) (Figure 1).Substantial heterogeneity was found in the comparison (I 2 = 96.0%,p = 0.01).Subgroup analysis by study design demonstrated the pooled estimates in cohorts compared to other types as SMD 1.11 (95%CI = 0.71, 1.51, p = 0.001; I 2 = 97%, p = 0.01) and SMD 0.85 (95%CI = 0.24, 1.47, p = 0.01; I 2 = 88%, p = 0.01) respectively (Supplementary Figure 3A).

PASC patients
Non        3).This substantial association between LDH and cardio-PASC is not statistically significant as the 95%CI crossed 0 and the p-value is not less than 0.05 (Figure 3).
Significant heterogeneity was observed among the included studies (I 2 = 99.7%,p = 0.01).The number of studies was too small to allow stratified analysis and meta-regression.4).Substantial heterogeneity was also observed (I 2 = 93.8%,p = 0.01).The number of studies was too small to allow stratified analysis and meta-regression.

| DISCUSSION
In this systematic review and meta-analysis of LDH contribution to symptom persistence in long COVID, we identified and critically assessed 46 studies that reported LDH data on distinct PASC patterns.Risk of bias assessment showed that all included studies were peer-reviewed; and while 24 were of good quality, 22 studies were of moderate quality.Our pooled analysis for these PASC subtypes revealed a substantial association between elevated LDH and Resp-PASC and Cardio-PASC, but not with Neuro-PASC and GI-PASC.
Our data could only permit a robust subgroup analysis and Further findings from our pooled analysis revealed that older age and female gender have a great impact on the relationship between LDH activity level and pulmonary sequelae.Earlier evidence had reported female gender and old age to be associated with increased risk of long COVID, 24 and this could explain the higher LDH level in these groups.However, larger future studies will be needed to establish the gender and age relationship with long COVID due to conflicting evidence of no correlation. 73Our data further suggest that LDH samples collected earlier before the 12 th week show higher statistical significance regarding the LDH activity and pulmonary sequelae correlation compared to data from the follow-up.Emerging evidence from a pooled analysis of 1, 289, 044 study participants from 12 studies reports a 24.6% prevalence for impaired lung diffusion capacity among COVID-19 survivors up to 2 years after recovery from SARS-CoV-2 infection. 3Similar evidence from a pooled analysis of 28 studies up to 6 months after recovery from SARS and MERS revealed impaired lung diffusion capacity at a prevalence of 27%. 74However, a recent study by Peixun Zhang and colleagues followed up 71 healthcare workers for up to 15 years after recovery from SARS infection and reported that most recovery from pulmonary impairment occurred in the first 2 years (first year in 2003 = 9%; second year in 2004 = 3.2%) but persisted thereafter up to 15 years (15 th year in 2018 = 4.6%). 75Interestingly, future research on longer term follow up of COVID-19 survivors remain an area of great interest as this will facilitate our understanding of the full spectrum of Resp-PASC, and even more interesting will be the data of the association between LDH activity and pulmonary sequelae.
Some level of association was also observed between raised LDH activity and cardio-PASC.This is consistent with the age-long traditional use of LDH as a marker of cardiac damage, though abnormally raised levels show multiple organ injury. 5Again, the elevation in plasma LDH may represent a biomarker of the body's strain response to the stress of COVID infection.Recent evidence synthesis on myocardial dysfunction following COVID-19 recovery showed that significant cardiac impairment (including reduced left/right ventricular ejection fraction, left ventricular end-diastolic volume and others) remains unresolved up to 12 months after SARS-CoV-2 infection in 2394 recovered COVID-19 survivors from 21 studies. 76madan and colleagues had earlier reported other patterns of cardiac sequelae including myocarditis, pericarditis, arrhythmia, myocardial infarction and heart failure across 35 studies in 52,609 COVID-19 survivors after recovery. 77However, the LDH/cardio-PASC association in our data did not reach statistical significance.
Our data also showed a smaller association between LDH activity and GI-PASC but almost no association with Neuro-PASC.Both were not statistically significant.

| Mechanistic insight and clinical significance of lactate dehydrogenase/post-acute sequelae of COVID-19 association
Lactate dehydrogenase is a marker of multiple tissue damage and has previously been shown to be elevated in severe COVID-19. 78An earlier pooled analysis of nine studies involving 1532 patients during acute SARS-CoV-2 infection had shown that raised LDH has a 6-fold increase in odds of predicting severe COVID-19 and a 16-fold increase in odds of predicting death in COVID-19 patients. 5LDH often serves as a general predictor of acute and chronic disease.Our data suggests that the tissue damage during acute illness, to a certain extent, persists into the post-COVID-19 era, and possibly partly accounts for some of the persistent symptoms widely reported.Raised serum LDH activity follows isoenzyme pattern often peculiar to various diseases, with huge clinical implication as a diagnostic marker. 4LDH has been demonstrated to be a more effective predictor (compared to C-reactive protein (CRP), D-dimer and others) of lung disease progression in COVID-19 survivors after recovery. 79A conflicting evidence by Jia-Ni Zou and colleagues demonstrated that IL-6 and albumin, rather than LDH were independent predictors of lung fibrosis at 3 months follow-up. 80 Increasing evidence reveals an essential role for lactate in signal transduction, 82 with evidence of potential feed-forward loop involving lactic acid, TGF-b, HIF1a, and LDH. 83Inflammatory immune cells have been shown to share the same metabolic rewiring previously described in tumour cells, which is characterised by a phenomenal switch from oxidative phosphorylation (via kreb's cycle) to aerobic glycolysis. 84This is called the Warburg effect.Warburg first observed that the tumour cells used huge amount of glucose and fermented glucose to generate lactate even when oxygen was present. 85Recent evidence shows that impaired lung diffusion capacity persists in COVID-19 survivors 24 months after recovery from SARS-CoV-2 infection, which suggests that the prevailing hypoxic state could be responsible for persistent lung inflammation. 86[89][90] Recent evidence reported significantly higher levels of lactate among PASC patients compared to the controls, which is suggestive of mitochondrial dysfunction in PASC. 91Guntus and colleagues in their recent study supports the theory of mitochondrial dysfunction among PASC patients, however they reported a lower level of lactate in the PASC group relative to the controls. 92A previous study had demonstrated that LDH enhances lactate generation which reinforces the immune-suppressive cells (such as macrophages and dendritic cells) and represses cytolytic cells (like natural killer cells and cytotoxic T-lymphocytes), and also stated that lactate increase is a predictor of poor outcome in tumours. 93Recent studies on long COVID have reported myriads of dysregulated immune response including reduced level of neutralising antibodies, anti-spike antibodies and raised level of positive Anti-nuclear antibody titres [94][95][96] from possible protracted germinal centre dysfunction 95,97 ; reduced IgG3 at 6 months of follow-up 98 ; role of anti-idiotype antibodies 99 ; raised levels of activated CD8 þ T cells at 8 months in long COVID patients 100 which correlates with autoimmunity 101 and drives chronic inflammation. 102,103 Mounting evidence of dysregulated metabolism in PASC 91,105 should drive well-powered future studies on the generic crosstalk between oxygen, lactate kinetics and mitochondrial function.Should such studies establish a strong evidence of mitochondrial dysfunction, then intervention strategies aimed at boosting mitochondrial function will become a priority area of research focus.Interestingly, such well-powered studies can sufficiently ascertain the impact of common confounders (like age, gender, pre-existing illness and disease severity at acute phase of SARS-CoV-2 infection) on this metabolic crosstalk.This is very important owing to the conflicting reports in two different studies by de Boer and colleagues with findings with findings of raised lactate 91 and lower lactate levels. 92wever, unlike the first study where samples were collected after graded exercise, 91 the second study had significant limitation as the samples were collected at rest, 92 rather than after small exercise intensity as recommended by the mitochondrial medicine society. 106

| Study limitations
To our knowledge, this is the first comprehensive systematic review and meta-analysis devoted solely to LDH that explored its association with distinct PASC domains.Additional strength is that the majority (74%) of our included studies were cohort studies and only 26% are either of retrospective or cross-sectional design.Admittedly, our pooled analysis had some limitations.First, most of the studies were focused on short-and medium-term follow-up data on LDH and long COVID sequelae after recovery.Our highest follow-up length is 12 months.Thus, further research studies should focus on longer-term follow-up which will delineate the full natural history of this association.Second, LDH isozyme functions were not assessed due to lack of data on LDH isoenzyme assays.Also, studies included in our metaanalysis mainly represented the respiratory sequelae, hence more data on all key body systems will improve our understanding on how LDH contributes to symptom persistence in long COVID.Third, due to paucity of data, the robustness of our analysis in terms of metaregression and subgroup analysis was constrained.Fourth, substantial heterogeneity was reported throughout the analysis, which is suggestive of inherent variability in LDH levels across studies as well as varied description of sequelae type.Sensitivity analysis was unable to completely eliminate it.Finally, there were no separate data for severe and non-severe COVID-19 patients in included studies and we could not analyse the risk of PASC based on COVID-19 disease severity.Fifth, in the current meta-analysis based on Centre for Disease Control and Prevention's recommendation, we defined long COVID as new, returning, or ongoing health problems occurring ≥4 weeks after being infected with SARS-CoV-2 infection.However, persistent symptoms lasting for 4 weeks or more may be ongoing COVID and the patient may still be in the acute phase of COVID-19.
Sixth, the use of multiple time points run the risk of introducing precision error in our pooled estimates.We identified two systematic reviews that reported on 113 107  studies with LDH data and did not carry out a meta-analysis, but rather described IL-6, CRP and TNF-α as potential markers of long COVID. 107In contrast, Yong et al. did a pooled analysis and reported a significantly higher levels of CRP (10 studies), D-dimer (11 studies), LDH (8 studies) and leucocytes (7 studies) in PASC. 108However, neither of them reported on LDH association with specific PASC domains, most probably due to limited LDH data.

| CONCLUSION
It is critical that a substantial association exist between raised LDH and Cardio-PASC, Resp-PASC in COVID-19 survivors after recovery.
Furthermore, based on the findings from our pooled analysis, we demonstrate that older age and female gender have a significant impact on the association between raised LDH activity and Resp-PASC.In contrast, our data suggests that there is no relationship between raised LDH and Neuro-PASC.Interestingly, admission LDH could be considered for inclusion in future risk stratification models for PASC.Future research works will confirm these findings.The presence of elevated plasma LDH level in PASC domains warrants further investigation; is the association a response to the stress of disease or is the correlation indicative of physiological strain against the infection.For the present, elevated plasma LDH following COVID infection may be considered as a disease biomarker.Therefore, we suggest that LDH and lactate are essential molecular mediators relevant to PASC and represent a potential therapeutic target that requires further research studies to explore their efficacy.A recent UDEH ET AL.

a
Follow-up data is described in months: 3-month follow-up data (3F), 6-month follow-up (6F) etc. b IC status: Immunocompromised status either excluded from the cohort [E] or matched [Y/N = Yes/No].Immunocompromised patients includes those patients with cancer, organ transplant recipients and those with rheumatological disorders.c long COVID incidence and risk factors.

F I G U R E 2 4 . 1 |
Forest plot showing Standardized mean difference (SMD) and 95% confidence interval (95% CI) of lactate dehydrogenase (LDH) in COVID-19 convalescents with or without Neuro-PASC.meta-regression for Resp-PASC and Neuro-PASC, and we report that there is no difference across the various follow-up durations; while older age, and female gender have huge impact on the LDH/Resp-PASC association.Lactate dehydrogenase elevation associates more with respiratory/cardio-PASC Concerning our finding of the association between Resp-PASC and raised LDH (which is statistically significant), one possibility is that LDH elevation may be linked to underlying lung tissue damage.This is supported by earlier evidence of higher predilection for the lungs during the acute phase of COVID-19, as the virus has been shown to bind to human angiotensin converting enzyme 2 in the lungs. 70This interpretation would be consistent with previous evidence of LDH as a predictor of lung disease among COVID-19 survivors 71 and predictor of lung fibrosis among Middle East Respiratory Syndrome (MERS) survivors 72 after recovery.Alternatively, the elevation in plasma LDH may be an alarm reaction biomarker for the presence of long COVID and the body's rallying against the infection.

F I G U R E 4
Forest plot showing Standardized mean difference (SMD) and 95% confidence interval (95% CI) of lactate dehydrogenase (LDH) in COVID-19 convalescents with or without GI-PASC.F I G U R E 3 Forest plot showing Standardized mean difference (SMD) and 95% confidence interval (95% CI) of lactate dehydrogenase (LDH) in COVID-19 convalescents with or without Cardio-PASC.

-PASC patients Outcome [IC status b ]
Further contrasting evidence by Mengqi Liu and colleagues reported that age, steroid treatment, and 18, 108 potential biomarkers of long COVID from 28 to 23 included studies with last searches in August and November 2022 respectively. 107,108Lai et al. had only 4