Abstract
Background
People with HIV might be at an increased risk of long COVID (LC) because of their immune dysfunction and chronic inflammation and alterations in immunological responses against severe acute respiratory syndrome coronavirus‐2 (SARS‐CoV‐2; coronavirus disease 2019 [COVID‐19]). This systematic review aimed to evaluate the association between HIV infection and LC and the prevalence and characteristics of and risk factors for LC among people with HIV.
Methods
Multiple databases, including Embase, PubMed, PsycINFO, Web of Science, and Sociological Abstracts, were searched to identify articles published before June 2023. Published articles were included if they presented at least one LC outcome measure among people with HIV and used quantitative or mixed‐methods study designs. For effects reported in three or more studies, meta‐analyses using random‐effects models were performed using R software.
Results
We pooled 39 405 people with HIV and COVID‐19 in 17 eligible studies out of 6158 publications in all the databases. It was estimated that 52% of people with HIV with SARS‐CoV‐2 infection developed at least one LC symptom. Results from the random‐effects model showed that HIV infection was associated with an increased risk of LC (odds ratio 2.20; 95% confidence interval 1.25–3.86). The most common LC symptoms among people with HIV were cough, fatigue, and asthenia. Risk factors associated with LC among people with HIV included a history of moderate–severe COVID‐19 illness, increased interferon‐gamma‐induced protein 10 or tumour necrosis factor‐α, and decreased interferon‐β, among others.
Conclusions
The COVID‐19 pandemic continues to exacerbate health inequities among people with HIV because of their higher risk of developing LC. Our review is informative for public health and clinical communities to develop tailored strategies to prevent aggravated LC among people with HIV.
Keywords: HIV/AIDS, long COVID, meta‐analysis, systematic review
INTRODUCTION
People with HIV may be at an increased risk for long COVID (LC), the long‐term clinical consequences of infection with severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2; coronavirus disease 2019 [COVID‐19]) [1, 2]. For example, a recent study revealed that the prevalence of LC in people with HIV was four‐fold higher than in a well‐matched comparison group of people without HIV [1]. However, the small sample size of this study (n = 82) meant it was underpowered to estimate the population‐level prevalence of LC and make comparisons within people with HIV (different immunity levels). Despite several LC‐related systematic reviews conducted among the general COVID‐19 population, the synergistic findings of LC in people with HIV have rarely been reported because data so far are extremely limited. To our knowledge, no recent large‐scale meta‐analysis has been conducted among people with HIV. The only literature review article (rather than a systematic review article) we identified presented a number of reasons why people with HIV might be vulnerable to developing LC, including sociodemographic factors, medical comorbidities, and dysregulation of immunological and/or physiological systems that could be exacerbated by acute and post‐acute SARS‐CoV‐2 infection [3]. Given the lack of a systematic assessment, the application of meta‐analysis was particularly useful when the results of primary investigations were either inconsistent or confusing. Therefore, the aim of this review was to systematically synthesize the global evidence base on the prevalence, symptomology, risk factors, immune responses, and inflammatory cytokines of LC in people with HIV. A greater understanding of LC in people with HIV will provide important information to inform the identification, management, and treatment of this condition.
METHODS
Data retrieval strategies
This review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) guideline (Figure S1 in Data S1) and was pre‐registered at https://www.crd.york.ac.uk/prospero/ as CRD42023445493. We conducted a comprehensive literature search of multiple databases, including Embase, PubMed, PsycINFO, Social Sciences, and Web of Science, for relevant papers published up to 28 May 2023. The pre‐determined search terms and search strategy are provided in Table S1 in Data S1. Search terms were only entered in English. Two authors (FS and HZ) independently reviewed the titles and abstracts of each article. The 14‐item Quality Assessment Tool for Observational Cohort and Cross‐Sectional Studies by the National Heart, Lung, and Blood Institute was used for the 17 quantitative studies. Two independent reviewers (FS and HZ) assessed the quality of the included studies, and discrepancies were resolved through discussion by involving a third reviewer (XY). An overall quality rating of good, fair, or poor was assigned for each study. The specific eligibility of included studies, data extraction process, and quality assessment of eligible studies are described in the eMethods in the Supplementary materials.
Statistical analysis
We used random‐effects models to estimate the pooled prevalence of LC manifestation reported in two or more studies and presented it with 95% confidence intervals (CIs). Meta‐analytic procedures were performed for the association between HIV infection status (HIV positive vs HIV negative) and the existence of LC, as well as the association between severity of acute SARS‐CoV‐2 infection and LC occurrence. A DerSimonian–Laird random‐effects model was employed to weight and pool the individual estimates because of the heterogeneous study populations and settings across studies. Pooled estimates of adjusted odds ratios (aORs) and the corresponding 95% CIs were calculated. We estimated the heterogeneity among studies using the Cochran Q score (reported as I 2, representing the percent value of the heterogeneity) with the corresponding p‐value. We used R software to perform the meta‐analyses.
RESULTS
Characteristics of each study
The process used to select the eligible studies is shown in Figure S1 in Data S1. The 17 evaluated studies included three cross‐sectional studies, eight cohort studies, one case–control study, and five observational or longitudinal studies. Seven studies were conducted in the USA (including one in both the USA and Peru), three in China, two in South Africa, and one each in Sudan, Italy, Zambia, India, and Spain. We pooled 39 405 participants in our analysis, including 3334 hospitalized participants, 1351 non‐hospitalized participants, and 34 720 participants whose hospitalization status was unknown. Time to follow‐up ranged from 30 days to 2 years. In quality assessment, 14 quantitative studies were classified as having good quality with a score of ≥8, and three studies were categorized as fair quality with a score of 7 (Table S3 in Data S1).
Prevalence of LC symptoms
The average study follow‐up time was 182 days, and we identified 36 LC symptoms in the literature reviewed (Table 1). The pooled prevalence of COVID‐19 survivors experiencing at least one unresolved symptom was 52% (95% CI 34%–71%) (Figure 1). Ranked symptom prevalence is presented in Table 1. The most reported symptoms were fatigue (30%; 95% CI 10%–49%; two studies), asthenia (21%; 95% CI 13%–32%; one study), vision problems (21%; 95% CI 9%–36%; one study), sleep disturbance (18%; 95% CI 0%–46%; two studies), neuropathy (18%; 95% CI 8%–34%; one study), and rash (18%; 95% CI 8–34; one study).
TABLE 1.
Manifestations of long COVID among people with HIV.
| Manifestations | Studies | Sample size | Cases | Prevalence, % (95% CI) |
|---|---|---|---|---|
| Clinical manifestations | ||||
| One or more symptom | 5 | 320 | 160 | 52 (34–71) |
| Shortness of breath | 3 | 205 | 30 | 15 (6–24) |
| Headache | 3 | 205 | 24 | 16 (0–38) |
| Fatigue | 2 | 130 | 35 | 30 (10–49) |
| Cough | 2 | 130 | 25 | 16 (0–32) |
| Sleep disturbance | 2 | 130 | 17 | 18 (0–46) |
| Myalgia | 2 | 130 | 15 | 14 (0–30) |
| Joint pain | 2 | 130 | 12 | 11 (0–27) |
| Anosmia | 2 | 130 | 10 | 1 (0–3) |
| Abdominal pain | 2 | 130 | 6 | 6 (0–17) |
| Dizziness | 2 | 130 | 6 | 6 (0–17) |
| Nausea | 2 | 130 | 6 | 6 (0–17) |
| Fever | 2 | 130 | 4 | 3 (0–6) |
| Chills | 2 | 130 | 3 | 4 (1–7) |
| Constipation | 2 | 130 | 2 | 1 (0–4) |
| Sore throat | 2 | 130 | 1 | 1 (0–3) |
| Asthenia | 1 | 75 | 16 | 21 (13–32) |
| Vision problems | 1 | 39 | 8 | 21 (9–36) |
| Neuropathy | 1 | 39 | 7 | 18 (8–34) |
| Rash | 1 | 39 | 7 | 18 (8–34) |
| Balance problems | 1 | 39 | 6 | 15 (6–31) |
| Back pain | 1 | 39 | 5 | 13 (4–27) |
| Diarrhoea | 1 | 39 | 5 | 13 (4–27) |
| Rhinorrhoea | 1 | 39 | 5 | 13 (4–27) |
| Palpitations | 1 | 39 | 4 | 10 (3–24) |
| Chest pain | 1 | 39 | 3 | 8 (2–21) |
| Loss of appetite | 1 | 39 | 3 | 8 (2–21) |
| Sinus congestion | 1 | 91 | 5 | 5 (2–12) |
| Bodyache | 1 | 91 | 3 | 3 (1–9) |
| Depression | 1 | 91 | 3 | 3 (1–9) |
| Anxiety | 1 | 91 | 2 | 2 (0–8) |
| Anorexia | 1 | 91 | 1 | 1 (0–6) |
| Hair loss | 1 | 91 | 1 | 1 (0–6) |
| Memory impairment | 1 | 91 | 1 | 4 (1–7) |
| Pancreatitis | 1 | 91 | 1 | 4 (1–7) |
| Seizure | 1 | 91 | 1 | 4 (1–7) |
| Stroke | 1 | 91 | 1 | 4 (1–7) |
| Laboratory tests and other examinations | ||||
| Viral shedding | 2 | 37 | 19 | 30 (0–72) |
| Positive SARS‐CoV‐2‐specific‐mBc | 1 | 11 | 11 | 100 (72–100) a |
| Positive SARS‐CoV‐2 spike‐specific‐IgG | 1 | 11 | 8 | 73 (40–94) |
Note: Random‐effects models were used for effects reported in two or more studies using the R software to estimate the pooled prevalence using an inverse transformation.
Abbreviations: CI, confidence interval; IgG, immunoglobulin G; mBc, memory B cell; SARS‐CoV‐2, severe acute respiratory syndrome coronavirus‐2.
One‐sided 97.5% CI.
FIGURE 1.
Prevalence of long COVID among people with HIV. CI, confidence interval.
HIV status and LC: Are people with HIV more vulnerable?
All four quantitative studies that investigated the relationship between HIV infection and LC provided data quantifying their associations with either cross‐sectional or cohort design [1, 2, 4, 5]. Figure 2 displays forest plots illustrating the association between HIV infection and LC. Overall, the pooled effect results from the random‐effects model showed that HIV infection was statistically associated with an increased risk of LC (OR 2.20; 95% CI 1.25–3.86). A relatively small heterogeneity was detected (I 2 = 22%, p = 0.28) between studies, suggesting no significant inconsistency across included studies. No additional subgroup analyses were performed because of the small number of studies.
FIGURE 2.
Forest plot presenting a pooled estimate of the association between HIV infection and the development of long COVID. OR, odds ratio.
Risk factors for LC among people with HIV
Regarding the risk factors, the severity of COVID‐19 in the acute phase was significantly associated with an increased risk of LC in people with HIV in two studies. Mazzitelli et al. reported an effect size of 6.61 (95% CI 1.74–25.2) [6]. Pujari et al. reported an effect size of 4.70 (95% CI 1.40–17.90) when investigating the relationship with moderate–severe COVID‐19 illness [7]. Overall, the pooled effect results from the random‐effects model indicated that the severity of COVID‐19 in the acute phase was statistically associated with an increased risk of LC (OR 8.27; 95% CI 2.94–23.26) (Figure S2 in Data S1). A relatively small heterogeneity was detected (I 2 = 0%, p = 0.34) between studies, suggesting that there was no significant inconsistency across included studies. Lower CD4 counts were associated with longer SARS‐CoV‐2 viral shedding (adjusted hazard ratio 0.14; 95% CI 0.07–0.28) [8] but not necessarily related to LC development after adjusting for relevant covariates [7] (Table 2).
TABLE 2.
Clinical manifestations and outcomes of long COVID among people with HIV.
| Outcomes | Studies | Risk factors | Results |
|---|---|---|---|
| LC | 7. Mazzitelli et al. | Severity of COVID‐19 disease | The severity of the COVID‐19 episode (not requiring vs. requiring hospitalization) (OR 6.61; 95% CI1.74–25.20; p = 0.006) was associated with LC (covariates include psychiatric disorders, polypharmacy, previous AIDS episodes, and the severity of the COVID‐19 episode) |
| LC | 11. Pujari et al. | Severity of COVID‐19 illness; median CD4 counts closest to screening, and receipt of steroids | Univariate analysis: history of moderate–severe COVID‐19 illness (OR 6.8; 95% CI 2.3–18.0; p = 0.0004), median CD4 counts closest to screening (p = 0.009), and receipt of steroids (OR 4.5; 95% CI 1.4–12.0; p = 0.008) during acute illness were significantly associated with LC. Logistic regression analysis: only moderate/severe COVID‐19 illness was significantly associated with LC (OR 4.7; 95% CI 1.4–17.9; p = 0.016) |
| LC | 15. Peluso et al. | IP‐10 levels and TNFα levels | Among people with HIV, there were increased odds of LC with each 10% increase in IP‐10 (aOR 1.06; 95% CI 1.00–1.11; p = 0.05) and a trend for increased LC with higher TNFα (aOR 1.20, per 10% increase; 95% CI 0.97–1.49; p = 0.09) but not IL‐6 (p = 0.64) |
| Viral shedding (SARS‐CoV‐2 viral loads) | 16. Meiring et al. | CD4 cell count and viral load | People with HIV with a CD4 cell count <200 cells/μL shed at high SARS‐CoV‐2 viral loads for longer (median 27 days [IQR 8–43]; aHR 0.14; 95% CI 0.07–0.28; p < 0.001) |
| Mortality | 10. Krishnan et al. | NAFLD | The mortality between NAFLD and non‐NAFLD cases had no significant difference at 30 (2.0% vs 2.1%), 60 (3.4% vs. 1.7%), and 90 days (3.5% vs.2.0%) after COVID‐19 infection |
Abbreviations: aHR, adjusted hazard ratio; aOR, adjusted OR; CI, confidence interval; IL, interleukin; IP‐10, interferon‐gamma‐induced protein 10; IQR, interquartile range; LC, long COVID; NAFLD, non‐alcoholic fatty liver disease; OR, odds ratio; SARS‐CoV‐2, severe acute respiratory syndrome coronavirus 2; TNF, tumour necrosis factor. The numbers preceding each author's name correspond to the articles listed in eTable 2.
Inflammatory biomarkers of LC in people with HIV
Regarding the inflammatory biomarkers, increased levels of interferon‐gamma‐induced protein 10 (IP‐10; aOR 1.06; 95% CI 1.00–1.11) [1, 5] and tumour necrosis factor (TNF)‐α (aOR 1.20, per 10% increase; 95% CI 0.97–1.49) [1] were more likely in people with HIV with LC than in those without HIV, whereas the level of interferon (IFN)‐β (p = 0.01) was significantly lower in people with HIV with LC than in those without LC [4]. Other cytokines, such as interleukin (IL)‐1β (p = 0.29), IL‐6 (p = 0.84), and IL‐10 (p = 0.28), were not significantly different between people with HIV and with or without LC [4] (Tables 2 and 3). Mean IL‐6 levels were 1.55‐fold higher (95% CI 1.06–2.26; p = 0.02), mean IP‐10 levels were 1.31‐fold higher (95% CI 1.06–1.62; p = 0.01), and TNFα levels were 1.26‐fold higher (95% CI 1.08–1.47; p = 0.003) in people with than without HIV [1].
TABLE 3.
Inflammatory markers, SARS‐CoV‐2 antibody response among people with and without HIV: values and influencing factors.
| Biomarkers measured | Studies | Values | Influencing factors |
|---|---|---|---|
| IFN‐β, IL‐1β, and IL‐10 | 9. Kamanzi, et al. | In HIV+ group, IFN‐β (p = 0.01) was significantly higher in people without than in those with LC symptoms; IL‐1β (p = 0.29), IL‐6 (p = 0.84), and IL‐10 (p = 0.28) were not significantly different between people with HIV with and without LC | NA |
| IL‐6, IP‐10, and TNFα | 15. Peluso, et al. | Mean IL‐6 was 1.55‐fold higher (95% CI 1.06–2.26; p = 0.02), mean IP‐10 was 1.31‐fold higher (95% CI 1.06–1.62; p = 0.01), and TNFα was 1.26‐fold higher (95% CI 1.08–1.47; p = 0.003) in people with HIV and LC than in those without HIV with LC | NA |
| IgM and IgG | 2. Alcaide, et al. | People with HIV had levels of IgM and IgG (p > 0.05) similar to people without HIV at all time points: baseline (T0), day 14 (T1), 1 mo (T2), 3 mo (T3), and 6 mo (T4) |
Age and IgG: significant correlation at T0 (r = 0.44), T1 (r = 0.50), T3 (r = 0.50), and T4 (r = 0.78) in those without HIV, no significant correlation at T2 in those without HIV; no significant correlation in people with HIV at all timepoints. Age and IgM: significant correlation at T4 (r = 0.68) in those without HIV, no significant correlation at T0, T1, T2, and T3 in those without HIV; no significant correlation in people with HIV at all timepoints |
| IgG, total lymphocyte, and CD4+ T‐lymphocyte counts | 4. Yang, et al. |
Levels of SARS‐CoV‐2 IgG in people with and without HIV: 5.11 ± 32.33 vs 37.45 ± 15.48 AU/mL, p = 0.042. Total lymphocyte counts in patients with COVID‐19 and HIV were lower than in patients with COVID‐19 without HIV: 2–4 weeks of illness, 0.91 vs 1.57 × 109/L; p = 0.040 versus 4–6 weeks of illness, 0.94 vs 1.81 × 109/L, p = 0.007. CD4+ T‐lymphocyte counts in patients with COVID‐19 and HIV were persistently lower than in patients with COVID‐19 without HIV: 2–4 weeks of illness, 337 vs 705/uL, p = 0.012 versus 4–6 weeks of illness, 370 vs 768/uL, p = 0.006 |
NA |
| IgM and IgG | 6. Huang et al. | NA | HIV VL, IgG, and IgM: both IgM and IgG were lower in patients with high VL than in those with low VL (≥20 vs <20 copies/mL, respectively); median S/CO for IgM was 0.03 vs 0.11, respectively (p < 0.001); median S/CO for IgG was 10.16 vs 17.04, respectively (p = 0.069) |
| Functional adaptive immunity, SARS‐CoV‐2‐specific humoral memory, SARS‐CoV‐2‐specific T‐cell memory | 12. Donadeu et al. |
Functional adaptive immunity: SARS‐CoV‐2‐specific humoral (both serological and cellular) and T‐cell memory were detected at 3 and 6 mo equally between people with and without HIV. SARS‐CoV‐2‐specific humoral memory: (1) T‐cell memory: although not significant, evident hierarchical T‐cell immune response for all cytokine‐producing T cells was observed regarding the three main immunogenic SARS‐CoV‐2 antigens (spike > membrane > nucleocapsid). (2) B‐cell memory: 32 of 36 (88.9%) displayed detectable RBD‐specific IgG‐producing mBc at T1, frequencies of IgG‐producing mBc were not different (p > 0.05) between both patient groups; at T2, for 11% of those with and 19.35% of those without HIV, RBD‐specific IgG‐producing mBc were not detectable; numerically higher RBD‐specific IgG‐producing mBc frequencies (p = 0.026) were also observed among the severe COVID‐19 group. Progression of SARS‐CoV‐2 immune memory in people with and without HIV over time: (1) SARS‐CoV‐2‐specific IgG titres: significantly dropped in both groups (p = 0.002 and p = 0.007 in those with and without HIV, respectively), especially in patients recovering from mild disease. (2) RBD‐specific SARS‐CoV‐2 mBc frequencies: did not change (p = 0.359 and p = 0.651 in people with and without HIV, respectively). (3) all SARS‐CoV‐2‐specific cytokine‐producing T‐cell responses (except IL‐21) dropped (IFN‐γ: p = 0.001 and p = 0.441 with and without HIV, respectively), (IL‐2: p = 0.001 and p = 0.003 with and without HIV, respectively), (IFN‐γ‐IL‐2: p = 0.001 and p = 0.614 with and without HIV, respectively), (IL‐21: p = 0.365 and p = 0.546 with and without HIV, respectively), particularly more evident within the severe clinical groups in people with HIV |
COVID‐19 severity and functional adaptive immunity: Both convalescent patients with and without HIV after a severe COVID‐19 infection displayed the highest detectable adaptive immune response at all immune compartments at 3 and 6 mo after COVID‐19 infection. COVID‐19 severity and SARS‐CoV‐2‐specific humoral memory: (1) people with and without HIV with severe COVID‐19 showed similar seroconversion rates and IgG titres (2) people with HIV and mild COVID‐19 had lower seroconversion rates and IgG titres than those without HIV and mild COVID‐19. COVID‐19 severity and SARS‐CoV‐2‐specific T‐cell memory: patients with severe COVID‐19 had higher T‐cell responses than those with mild infection, especially for IL‐2 and IFN‐g/IL‐2‐producing T cells. Patients with mild COVID‐19 had significantly higher cytokine‐producing T‐cell frequencies than people without HIV and mild COVID‐19 at 3 mo after COVID‐19 infection. Association between serological and cellular SARS‐CoV‐2‐specific immunity: (1) A strong correlation between the different immune responses was observed among people without but not in people with HIV correlation between IgG‐spike titres and RBD‐specific IgG‐producing mBc frequencies was R = 0.398, p = 0.003 without HIV and R = 0.113, p = 0.636 with HIV; correlation between IgG‐spike titres and different cytokine‐producing T‐cell frequencies was R = 0.372, p = 0.002 for IFN‐g; R = 0.456, p < 0.001 for IL‐2; and R = 0.431, p < 0.001 for INF‐g/IL‐2 without HIV and R = 0.191, p = 0.394 for IFN‐g; R = 0.262, p = 0.265 for IL‐2; and R = 0.346, p = 0.15 for IFN‐g/IL‐2 with HIV. (2) RBD‐specific IgG‐producing mBc frequencies and the different cytokine‐producing T‐cell frequencies: no statistical correlation (p > 0.05) |
Abbreviations: CI, confidence interval; IFN, interferon; IgG, immunoglobulin G; IgM, immunoglobulin M; IL, interleukin; IP‐10, interferon‐gamma‐induced protein 10; LC, long COVID; mBc, memory B‐cell; mo, month(s); NA, not applicable; SARS‐CoV‐2, severe acute respiratory syndrome coronavirus 2; RBD, receptor‐binding domain; S/CO, signal value divided by cutoff value; TNF, tumour necrosis factor; VL, viral load. The numbers preceding each author's name correspond to the articles listed in eTable 2.
SARS‐CoV‐2 antibody response
When comparing people with and without HIV, one study observed similar levels of SARS‐CoV‐2 immunoglobulin (Ig)‐G and IgM (p < 0.05) at all time points (baseline [T0], day 14 [T1], 1 month [T2], 3 months [T3], and 6 months [T4]) [9], whereas the other study observed higher IgG in people with than without HIV (p = 0.04) [10]. When we stratified the immune response by different viral loads of people with HIV, both IgM and IgG levels were lower in people with high HIV viral loads (≥20 copies/mL) than in those with low HIV viral loads (<20 copies/mL) [11]. When we stratified the immune response by the severity of initial SARS‐CoV‐2 infection, people with and without HIV showed similar seroconversion rates and IgG titres if they presented with severe outcomes in the initial SARS‐CoV‐2 infection; among individuals presenting mild symptoms, people with HIV had lower seroconversion rates and IgG titres than those without HIV [12].
In terms of the progression of SARS‐CoV‐2 immune memory in people with and without HIV over time, SARS‐CoV‐2‐specific IgG titres significantly dropped in both groups (p = 0.002 with HIV, p = 0.007 without HIV). All SARS‐CoV‐2‐specific cytokine‐producing T‐cell responses (except IL‐21) also dropped (IFN‐γ: p = 0.001 with HIV, p = 0.441 without HIV; IL‐2: p = 0.001 with HIV, p = 0.003 without HIV; IFN‐γ‐IL‐2: p = 0.001 with HIV, p = 0.614 without HIV; IL‐21: p = 0.365 with HIV, p = 0.546 without HIV). This was particularly more evident within the clinical groups in people with HIV [12]. However, receptor‐binding‐domain‐specific IgG‐producing memory B‐cell frequencies did not change over time (p > 0.05) [12] (Table 3).
DISCUSSION
Clinical manifestations
In this systematic review, we synthesized the findings of 17 eligible studies focusing on the relationship between HIV infection and prevalence of LC, as well as the clinical manifestations, risk factors, immune responses, and mechanisms of LC in HIV populations after COVID‐19 infection. Most of the studies were conducted in the USA, followed by Asian countries such as China and India, and the rest were conducted in Europe or South Africa. Our results indicated a pooled prevalence of 47% of people with HIV experiencing at least one unresolved symptom, which is similar to the 45% reported by another systematic review conducted in the general population [13].
As in the general population, LC is common among people with HIV recovering from acute SARS‐CoV‐2 infection. LC symptoms in people with HIV are also heterogeneous and involve different organ systems, such as respiratory symptoms and neurological symptoms. Fatigue, asthenia, sleep disturbance, neuropathy, and rash are the most reported symptoms in people with HIV, ranging from 18% to 30% in our study, which is comparable with the 15%–34% reported in the general population [3]. However, the most common LC symptoms in people with HIV were neurological symptoms such as fatigue and sleep disturbance, rather than respiratory symptoms. The variability in the estimated prevalence of different LC symptoms between studies remained high. The heterogeneity between studies may be explained not only by differences in study designs, follow‐up measurement tools, and the wide range of follow‐up durations but also the lack of standardized data collection tools. This emphasizes the need for standardized tools to harmonize data collection, such as the Symptom Burden Questionnaire™, which comprehensively assesses a wide range of symptoms highlighted by existing literature and with input from patients, researchers, and clinicians [14]. Future research should aim to map symptom classifications onto a core outcome set [15] for LC to help synthesize findings.
Prevalence of LC in people with versus without HIV
People with HIV might be at an elevated risk for experiencing LC because of the substantial overlap between the demographics and clinical comorbidities of people with HIV and the clinical risk factors of LC [3, 16]. First, the literature suggests that people with HIV are more likely to develop severe symptoms during the acute SARS‐CoV‐2 infection, and severity of acute infection has been identified as a risk factor for LC [16]. Second, some common chronic comorbid conditions (e.g., diabetes, obesity, cardiovascular disease, and mental health conditions), lower socioeconomic status, and substance use among people with HIV are also predictors for LC. Furthermore, other potential contributors to LC or exacerbating factors such as immune dysfunction, increased inflammation, and alteration in the immunological response against COVID‐19 are also common among people with HIV [16]. Our meta‐analysis demonstrated the significant pooled effect of HIV infection on LC development (OR 2.98). The four quantitative studies included in our meta‐analysis had small sample sizes, with participants primarily middle‐aged men, and most were on antiretroviral therapy or had good immune reconstitution. Nevertheless, the synthesized effect from our meta‐analysis has provided relatively robust evidence of an elevated risk of LC in people with HIV. However, the small number of studies meant we were unable to calculate the regional differences regarding LC in people with HIV. Large‐scale population‐based studies with more representative samples, such as the inclusion of people with more pronounced immunodeficiency, are warranted to confirm the relationship between HIV infection and the development of LC.
Immune responses of LC in people with HIV
The concentration and duration of antibodies to SARS‐CoV‐2 infection predicts the severity of the disease and the clinical outcomes, including LC outcomes. Development of IgM and IgG antibodies against the spike protein could likely indicate protection against reinfections and potential protection after vaccination. It was hypothesized that an impaired immune response would result in lower and shorter‐term antibody responses. The impaired immune systems of people with HIV might mean they are unable to mount robust immune responses after viral infection or vaccination. The findings of our systematic review are mixed. Three studies [1, 9, 12] indicated that short‐ and long‐term natural functional cellular (T‐cell memory) and humoral immune responses (i.e., antibody responses to SARS‐CoV‐2 infection) were comparable in people with and without HIV (although lower levels of SARS‐CoV‐2‐specific memory CD8+ T cells and higher levels programmed death (PD‐1)+ SARS‐CoV‐2‐specific CD4+ T cells were observed in Peluso et al. [1]), which suggests that people with HIV are capable of developing and maintaining a robust T‐ and B‐cell memory immune response, similar to people without HIV. Yet, the fourth study [10] observed that people co‐infected with HIV and SARS‐CoV‐2 had persistently lower levels of IgG and other immune responses, such as total lymphocyte and CD4+ T‐lymphocyte counts. The evidence from these studies could shed light on the comparison of immune responses between people with and without HIV; however, a range of limitations, such as small sample size and different study participant characteristics, still prevent us from drawing concrete conclusions. Given these knowledge gaps, larger and more diverse studies are needed to understand the natural functional cellular (T‐cell memory) and host immune response and duration of antibody levels after SARS‐CoV‐2 in the HIV setting and to draw conclusions.
Within the population of people with HIV, absolute CD4 counts did not correlate with IgM and IgG responses [9], yet both IgM and IgG levels were lower in people with high HIV viral loads (≥20 copies/mL) than in those with low HIV viral loads (<20 copies/mL) [11]. It was hypothesized that low levels of CD4+ T‐lymphocyte counts in those with both HIV and COVID‐19 might be one of the reasons for the relatively weak ability to produce antibodies. However, findings from Alcaide et al. did not support such a relationship, which might in turn explain the comparable immune responses to SARS‐CoV‐2 between people with and without HIV. It is worthwhile to note that people with HIV with lower CD4 T‐cell counts may have elevated IFN levels. It could be argued that these patients might be protected from the severe manifestations of COVID‐19, given the higher IFN [17]. Furthermore, it has also been suggested that the lower but active immune status of people with HIV can confer a protective immune status, especially from a virus‐induced cytokine storm [18]. Future studies should account for the impact of CD4+ T‐lymphocyte counts on SARS‐CoV‐2 antibody screening results when conducting epidemiological investigations of COVID‐19 herd immunity.
Inflammatory markers associated with LC in people with HIV
A growing body of literature has identified immune dysregulation and inflammation, specifically pathways involving IL‐6, TNFα, and – in some cases – IL‐1B, among others, in individuals with LC in comparison with people experiencing full recovery [19, 20, 21, 22]. However, evidence of immunomodulatory therapy in LC is limited, so whether these markers drive LC symptoms or are simply a result of other pathophysiological processes is unknown. It has been hypothesized that chronic immune activation and immune dysregulation in chronic HIV infection could increase the risk of LC. Several studies included in our systematic review compared the levels of the pro‐ and anti‐inflammatory cytokines, including IL‐1β, IL‐6, IL‐10, and IFN‐β between people with HIV with and without LC. IP‐10 and TNFα levels were high in people with HIV with LC, whereas IFN‐β was protective from LC in people with HIV. IFN‐β is a type 1 IFN that has both antiviral and immunoregulatory functions, including both pro‐ and anti‐inflammatory effects [23, 24]. The mechanisms for the anti‐inflammatory effects of IFN‐β include the downregulation of class 2 human leukocyte antigen, inhibition of T‐cell migration, and upregulation of IL‐10 secretion [25]. The relationship between IL‐6 and persistent symptoms of LC was mixed: one study found no relationship between the two variables, whereas another observed a higher mean level of IL‐6 in people with HIV than in those without. As a pleiotropic cytokine, IL‐6 has both pro‐ and anti‐inflammatory effects. IL‐6 is known to increase T‐helper 17 activity and hence to reduce regulatory T‐cell activity, and this imbalance is associated with a systemic inflammatory state that can result in LC symptoms [26, 27]. That might explain why, in some studies, IL‐6 has been persistently elevated in the plasma of patients with LC [28]. However, given the mixed findings, the role of IL‐6 in the development of LC requires further studies with larger sample sizes to confirm such a relationship. In summary, it may be difficult to distinguish whether the different levels of these cytokines are caused by LC or by HIV infection themselves, as long‐term HIV infection may lead to a proinflammatory immune‐dysregulated state due to the chronic immune activation associated with HIV infection [29, 30].
Limitations
Although this is the most comprehensive and contemporary review to date, systematically synthesizing the global evidence on the prevalence, symptomology, risk factors, and immune response of LC in people with HIV and comparing the risk of LC in people with and without HIV, several limitations need to be highlighted. First, we could not perform meta‐analyses for risk factors for LC in people with HIV other than severity of initial SARS‐CoV‐2 infection because of the small number of eligible studies with applicable outcomes. Second, the large variability in both outcome measures (no consensus on LC definition) [31] and study design across the included studies makes it difficult to compare effect sizes and draw definitive conclusions from this modest body of research. Lastly, most included studies were conducted in the USA (n = 8 studies, 47%) and Asian countries. This geographical heterogeneity highlights that more work is needed to address the burden and long‐term effects of COVID‐19 in different geographic settings, especially low‐ and middle‐income countries, such as Africa, to ensure resources are appropriately targeted and utilized and culturally appropriate intervention strategies are developed.
CONCLUSION
This systematic review provides the most contemporary and comprehensive estimates of the prevalence and long‐term health effects of LC among people with HIV. The meta‐analysis suggested an elevated risk of LC in people with HIV compared with those without HIV. Unlike in the general population, the most common LC symptoms in people with HIV are neurological symptoms such as fatigue and sleep disturbance instead of respiratory symptoms. Findings for immune responses to SARS‐CoV‐2 infection were mixed, with one study demonstrating higher IgG and IgM in people with HIV and another finding no such relationship. Regarding the inflammatory markers, increased IP‐10 and TNFα and decreased IFN‐β were observed in people with HIV with LC compared with people with HIV without LC, yet no significant differences in other cytokines were observed. These results, albeit limited because of the small number of available studies, are informative, provide a better understanding of the risks of LC in people with HIV, and help elucidate the mechanisms of LC in people with HIV. As the mechanisms of LC in the general population are further delineated, we should pay attention to the contributors that might be unique in driving LC in people with HIV, such as the persistent state of immune activation and inflammation. Large population‐based studies are needed to generate more robust evidence of the role of HIV infection in LC development.
AUTHOR CONTRIBUTIONS
XY conceptualized the study design. XY and FS prepared the first draft of this manuscript. FS, WG, and AK performed the literature screening. FS and HZ performed data extraction, quality assessment, and data analysis. HC provided clinical input, particularly the biomarkers for the manuscript. XY and XL guided the review structure and revised the manuscript critically. All authors read and approved the final manuscript.
FUNDING INFORMATION
This work was made possible with funding to the University of South Carolina through R21AI170159‐01A1 from the National Institutes of Health, with support from the NIH Office of AIDS Research. The content is solely the responsibility of the authors and does not necessarily reflect the official views of the NIH.
CONFLICT OF INTEREST STATEMENT
The authors have no conflicts of interest.
Supporting information
Data S1. Supporting Information.
Yang X, Shi F, Zhang H, et al. Long COVID among people with HIV: A systematic review and meta‐analysis. HIV Med. 2025;26(1):6‐16. doi: 10.1111/hiv.13708
Xueying Yang and Fanghui Shi contributed equally to this work.
The abstract of this paper has been accepted for presentation in AIDS 2024, the 25th International AIDS Conference.
REFERENCES
- 1. Peluso MJ, Spinelli MA, Deveau T‐M, et al. Postacute sequelae and adaptive immune responses in people with HIV recovering from SARS‐COV‐2 infection. Aids. 2022;36(12):F7‐F16. doi: 10.1097/qad.0000000000003338 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Kingery JR, Safford MM, Martin P, et al. Health status, persistent symptoms, and effort intolerance one year after acute COVID‐19 infection. J Gen Intern Med. 2022;37(5):1218‐1225. doi: 10.1007/s11606-021-07379-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Peluso MJ, Antar AAR. Long COVID in people living with HIV. Curr Opin HIV AIDS. 2023;18(3):126‐134. doi: 10.1097/coh.0000000000000789 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Kamanzi P, Mulundu G, Mutale K, Mumba C, Ngalamika O. HIV and inflammatory markers are associated with persistent COVID‐19 symptoms. Immun Inflamm Dis. 2023;11(5):e859. doi: 10.1002/iid3.859 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Peluso MJ, Deveau TM, Munter SE, et al. Chronic viral coinfections differentially affect the likelihood of developing long COVID. J Clin Invest. 2023;133(3):e163669. doi: 10.1172/JCI163669 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Mazzitelli M, Trunfio M, Sasset L, et al. Factors associated with severe COVID‐19 and post‐acute COVID‐19 syndrome in a cohort of people living with HIV on antiretroviral treatment and with undetectable HIV RNA. Viruses. 2022;14(3):493. doi: 10.3390/v14030493 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Pujari S, Gaikwad S, Chitalikar A, Dabhade D, Joshi K, Bele V. Long‐coronavirus disease among people living with HIV in western India: an observational study. Immun Inflamm Dis. 2021;9(3):1037‐1043. doi: 10.1002/iid3.467 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Meiring S, Tempia S, Bhiman JN, et al. Prolonged shedding of severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) at high viral loads among hospitalized immunocompromised persons living with human immunodeficiency virus (HIV), South Africa. Clin Infect Dis. 2022;75(1):e144‐e156. doi: 10.1093/cid/ciac077 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Alcaide ML, Nogueira NF, Salazar AS, et al. A longitudinal analysis of SARS‐CoV‐2 antibody responses among people with HIV. Front Med (Lausanne). 2022;9:768138. doi: 10.3389/fmed.2022.768138 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Yang RR, Gui X, Zhang YX, Xiong Y, Gao SC, Ke HN. Clinical characteristics of COVID‐19 patients with HIV coinfection in Wuhan, China. Expert Rev Respir Med. 2021;15(3):403‐409. doi: 10.1080/17476348.2021.1836965 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Huang J, Xie N, Hu X, et al. Epidemiological, Virological and serological features of coronavirus disease 2019 (COVID‐19) cases in people living with human immunodeficiency virus in Wuhan: a population‐based cohort study. Clin Infect Dis. 2021;73(7):e2086‐e2094. doi: 10.1093/cid/ciaa1186 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Donadeu L, Tiraboschi JM, Scévola S, et al. Long‐lasting adaptive immune memory specific to SARS‐CoV‐2 in convalescent coronavirus disease 2019 stable people with HIV. Aids. 2022;36(10):1373‐1382. doi: 10.1097/qad.0000000000003276 [DOI] [PubMed] [Google Scholar]
- 13. O'Mahoney LL, Routen A, Gillies C, et al. The prevalence and long‐term health effects of long Covid among hospitalised and non‐hospitalised populations: a systematic review and meta‐analysis. EClinicalMedicine. 2023;55:101762. doi: 10.1016/j.eclinm.2022.101762 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Hughes SE, Haroon S, Subramanian A, et al. Development and validation of the symptom burden questionnaire for long covid (SBQ‐LC): Rasch analysis. BMJ. 2022;377:e070230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Munblit D, Nicholson T, Akrami A, et al. A core outcome set for post‐COVID‐19 condition in adults for use in clinical practice and research: an international Delphi consensus study. Lancet Respir Med. 2022;10(7):715‐724. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Yang X, Sun J, Patel RC, et al. Associations between HIV infection and clinical spectrum of COVID‐19: a population level analysis based on US national COVID cohort collaborative (N3C) data. Lancet HIV. 2021;8(11):e690‐e700. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Roff SR, Noon‐Song EN, Yamamoto JK. The significance of interferon‐γ in HIV‐1 pathogenesis, therapy, and prophylaxis. Front Immunol. 2014;4:498. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Laurence J. Why aren't people living with HIV at higher risk for developing severe coronavirus disease 2019 (COVID‐19)? AIDS Patient Care STDS. 2020;34(6):247‐248. [DOI] [PubMed] [Google Scholar]
- 19. Peluso MJ, Lu S, Tang AF, et al. Markers of immune activation and inflammation in individuals with postacute sequelae of severe acute respiratory syndrome coronavirus 2 infection. J Infect Dis. 2021;224(11):1839‐1848. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Peluso MJ, Sans HM, Forman CA, et al. Plasma markers of neurologic injury and inflammation in people with self‐reported neurologic postacute sequelae of SARS‐CoV‐2 infection. Neurol Neuroimmunol Neuroinflamm. 2022;9(5):e200003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Schultheiß C, Willscher E, Paschold L, et al. The IL‐1β, IL‐6, and TNF cytokine triad is associated with post‐acute sequelae of COVID‐19. Cell Rep Med. 2022;3(6):100663. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Su Y, Yuan D, Chen DG, et al. Multiple early factors anticipate post‐acute COVID‐19 sequelae. Cell. 2022;185(5):881‐895.e20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Bolívar S, Anfossi R, Humeres C, et al. IFN‐β plays both pro‐and anti‐inflammatory roles in the rat cardiac fibroblast through differential STAT protein activation. Front Pharmacol. 2018;9:1368. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Mariani MK, Dasmeh P, Fortin A, et al. The combination of IFN β and TNF induces an antiviral and immunoregulatory program via non‐canonical pathways involving STAT2 and IRF9. Cells. 2019;8(8):919. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Jacobs LD, Cookfair DL, Rudick RA, et al. Intramuscular interferon beta‐1a for disease progression in relapsing multiple sclerosis. Ann Neurol. 1996;39(3):285‐294. [DOI] [PubMed] [Google Scholar]
- 26. Sadeghi A, Tahmasebi S, Mahmood A, et al. Th17 and Treg cells function in SARS‐CoV2 patients compared with healthy controls. J Cell Physiol. 2021;236(4):2829‐2839. [DOI] [PubMed] [Google Scholar]
- 27. Meckiff BJ, Ramírez‐Suástegui C, Fajardo V, et al. Imbalance of regulatory and cytotoxic SARS‐CoV‐2‐reactive CD4+ T cells in COVID‐19. Cell. 2020;183(5):1340‐1353.e16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Evans RA, Leavy OC, Richardson M, et al. Clinical characteristics with inflammation profiling of long COVID and association with 1‐year recovery following hospitalisation in the UK: a prospective observational study. Lancet Respir Med. 2022;10(8):761‐775. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Nasi M, De Biasi S, Gibellini L, et al. Ageing and inflammation in patients with HIV infection. Clin Exp Immunol. 2017;187(1):44‐52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Williams ME, Ipser JC, Stein DJ, Joska JA, Naudé PJ. Peripheral immune dysregulation in the ART era of HIV‐associated neurocognitive impairments: a systematic review. Psychoneuroendocrinology. 2020;118:104689. [DOI] [PubMed] [Google Scholar]
- 31. Groff D, Sun A, Ssentongo AE, et al. Short‐term and long‐term rates of postacute sequelae of SARS‐CoV‐2 infection: a systematic review. JAMA Netw Open. 2021;4(10):e2128568. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1. Supporting Information.