key: cord-0775384-vd01nthu
authors: Foret, Thomas; Dufrost, Virginie; du Mont, Lucie Salomon; Costa, Patricia; Lefevre, Benjamin; Lacolley, Patrick; Regnault, Veronique; Zuily, Stephane; Wahl, Denis
title: Systematic Review of Antiphospholipid Antibodies in COVID-19 Patients: Culprits or Bystanders?
date: 2021-07-03
journal: Curr Rheumatol Rep
DOI: 10.1007/s11926-021-01029-3
sha: 62dec26171045e39b36e05af0c5a186908514998
doc_id: 775384
cord_uid: vd01nthu

PURPOSE OF REVIEW: COVID-19 patients have a procoagulant state with a high prevalence of thrombotic events. The hypothesis of an involvement of antiphospholipid antibodies (aPL) has been suggested by several reports. Here, we reviewed 48 studies investigating aPL in COVID-19 patients. RECENT FINDINGS: Prevalence of Lupus Anticoagulant (LA) ranged from 35% to 92% in ICU patients. Anti-cardiolipin (aCL) IgG and IgM were found in up to 52% and up to 40% of patients respectively. Anti-β(2)-glycoprotein I (aβ(2)-GPI) IgG and IgM were found in up to 39% and up to 34% of patients respectively. Between 1% and 12% of patients had a triple positive aPL profile. There was a high prevalence of aβ(2)-GPI and aCL IgA isotype. Two cohort studies found few persistent LA but more persistent solid phase assay aPL over time. SUMMARY: aPL determination and their potential role is a real challenge for the treatment of this disease.

Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) is at the origin of coronavirus disease 2019 (COVID- 19) , which has immersed the world in a new global pandemic since early 2020. In the first descriptions in China, COVID-19 clinical manifestations are dominated by respiratory symptoms with pneumonia and inflammatory state [1, 2] . With the progress of the pandemic, a significant number of thrombotic events were identified. Indeed, the incidence of both arterial and venous thromboembolism is high in COVID-19 patients [3] , sometimes in spite of preventive anticoagulant treatment [4, 5] . In some cases, this viral infection may be associated with modifications in coagulation parameters revealing a procoagulant state in COVID-19 patients associated with poor clinical outcome [6•, 7, 8] . Zhang et al. first suggested a possible correlation between antiphospholipid antibodies (aPL) and thrombosis by reporting three cases of COVID-19 patients with multiple thrombosis and anticardiolipin (aCL), immunoglobulin (Ig) A, and anti-β 2 -This article is part of the Topical Collection on Antiphospholipid Syndrome glycoprotein I (aβ 2 -GPI) IgA and IgG positivity [9] . Many case series and cross-sectional studies have been published in order to further investigate the role of these aPL during COVID-19 infection. Thus, the aim of this systematic review was firstly to analyze the frequency of aPL in COVID-19 patients in different settings and to evaluate their persistence over time and secondly, to analyze the role of aPL during the infection in particular their participation in thrombotic events.

We conducted a systematic review of all articles about aPL in COVID-19 patients. We performed this search for international English articles in Medline database with the following keywords: "(antibody, antiphospholipid[MeSH Terms] OR antibody syndrome, antiphospholipid[MeSH Terms] OR lupus anticoagulant[MeSH Terms] OR lupus anticoagulant OR anticardiolipin OR anti-beta2 glycoprotein I OR antiphospholipid antibod* OR antiphospholipid antibody syndrome) AND (coronavirus, sars[MeSH Terms] OR COVID OR coronavirus disease 2019)". Each article published was analyzed and only studies evaluating the prevalence of aPL in case series of at least two COVID-19 patients over 18 years old were included. Percentages were calculated from studies of more than 10 patients. Single patient case reports and studies of children were excluded.

We identified a total of 190 publications (last search on May 4, 2021) after excluding duplicates and non-English papers. Of the 190 references selected, 142 were excluded as indicated in the Flowchart (Fig. 1) . Overall, 48 studies were eligible for a complete analysis of their results [6•, 9, 10, 11, 12, 13••, 14-55] . Only two reports were cohort studies with repeated assays for aPL after one month for the first [13••] and between 3 and 6 months for the second [52] . Eight publications were case reports from two to six patients [9, 17, 30, 33, 34, 36, 38, 51] . Other publications were cross-sectional studies. Tables 1, 2, and 3 display the main results for studies evaluating aPL in intensive care units (ICU, Table 1 ), medical ward (MW) or without specific information (Table 2) , and both ICU and MW patients (Table 3) . According to the type of antibodies, there was a high prevalence of lupus anticoagulant (LA), from about 35% up to 90% in ICU patients with one exception: a study found LA in 5% of patients [15] . In studies combining ICU and MW patients, the prevalence of LA was between 20% and 66% except for one study who found LA in 2% of patients [45] . In MW patients, two studies have performed LA assays and found a prevalence of 39% and 46% [32, 37] . In studies without information on patients setting, prevalence was between 22% and 91%. Of note, Bauer et al. did not find more LA in COVID-19 patients on admission to their emergency department compared to patients without COVID-19 [47] . A total of 91% of these COVID-19 patients were subsequently hospitalized.

The strict application of the three-step LA testing recommended by the International Society on Thrombosis and Haemostasis (ISTH) [56] was explicitly described by 18 among 23 studies performing LA assays. Inflammation parameters were reported in 17 among 21studies. Mean fibrinogen and C-reactive protein (CRP) were higher than normal values in all these studies. CRP and fibrinogen values varied between 36 and 286 mg/L and 4.2 and 7.6 g/L respectively. Several studies found a statistical association between the presence of LA and the levels of CRP or fibrinogen [28, 32, 55] . The two studies with the lowest prevalence of LA (2% and 5%) had the lowest level of fibrinogen (4.5 and 4.4 g/L respectively).

The prevalence of aCL IgM varied between 0% and 40% and the prevalence of aCL IgG varied between 0% and 59%. The prevalence of aβ 2 -GPI was also variable in most studies: between 0% and 39% of patients had aβ 2 -GPI IgG and between 0% and 34% of patients had aβ 2 -GPI IgM. The proportion of triple positivity (combined positivity for LA, aCL and aβ 2 -GPI antibodies) was from 1% to 12% across studies. Assays for aPL and the cut-off used were explicitly described in 21 among 32 studies.

Many studies have also investigated less conventional antibodies (i.e., that are not classification criteria nor assayed in routine clinical practice as opposed to LA, IgG and IgM aCL and aβ 2 -GPI). Thus anti-phosphatidylserine/prothrombin (aPS/PT) were found in 0% to 24% of patients, and antiannexin V (aAV) in 3% to 19% of patients. One study performed anti-phosphatidylinositol (aPI) IgG and IgM only in ICU patients. No aPI were found. IgA aCL were found more frequently, from 20% to more than 90%, except for four studies that described a low prevalence between 0 and 4% [27, 31, 41•, 55] . IgA aβ 2 -GPI was present from 0% to 86% of patients.

Almost all publications studied hospitalized patients only, while Gatto et al. studied both hospitalized and COVID-19 outpatients [27] . They did not show any association between the presence of aPL and thrombotic events or with the necessity to hospitalize patients [27] . The prevalence of LA was 30% and 1% to 8% for the other aPL in COVID-19 outpatients.

Two studies followed-up aPL persistence over time. The first study was conducted in ICU patients [13••] and investigated the persistence of aPL at 1 month. Initially 23 out 31 patients had at least one aPL (mostly LA, in 67% of patients). At 1 month, 10 patients were tested again and only one had persistent aPL. Thus, aPL were confirmed at 1 month for only 1 among 10 positive LA, 0 among 4 aCL and 1 among 2 aβ 2 -GPI IgG. Persistent LA and aβ2-GPI were present in the same patient.

A second study performed aPL assays between 3 and 6 months after a first positive LA test [52] . A total of 42 patients among 79 patients initially tested positive for LA were tested again. LA was found negative in all these patients. In these 42 patients, 7 were positive for aCL, 1 for aβ 2 -GPI and 5 for unconventional antibodies. Authors did not indicate if these antibodies were similar to the initial samples.

Some studies have found a high prevalence of aPL [6•, 18, 23, 37, 43, 57] while others found a low prevalence and this could be linked to disease severity [19, 24, 26, 35•, 42, 45, 55] . Xiao et al. found aPL in 31 out of 66 patients requiring ICU admission but not in patients with noncritical conditions [45] . Several studies suggested that aCL IgG or IgM were highly and independently associated with COVID-19 severity [40, 52, 58] . However, others studies did not confirm these results. Ferrari et al. found a similar prevalence for LA, aβ 2 -GPI and aCL in severe and non-severe COVID-19 patients [43] , and other authors did not find more aPL (aCL or aβ 2 -GPI) between patients with COVID-19 related acute respiratory disease syndrome and patients with pneumonia-associated acute respiratory disease syndrome in ICU [19, 53] . One study did not find more LA in COVID-19 non-survivors than in survivors [32] , likewise other studies did not find any association between overall aPL positivity and in-hospital mortality [50, 55] .

Regarding the risk of thrombosis several studies have found a statistical association between the presence of aPL and thrombotic events [6•, 28, 37] , or between their presence and the inflammatory state of the patients [12, 55] . Indeed, Le Joncour et al. found more aPL (aCL IgG and IgM and aβ 2 -GPI IgA) in patients with thrombotic events in MW. These patients had also higher neutrophils counts and higher D-Dimers and CRP levels. However, this was not in line with other authors who did not find an association between the presence of aPL and the thrombotic complications [18, 55] .

Specific studies analyzed the prevalence of aPL in COVID-19 patients with stroke or myocardial infarction. In these retrospective studies, between 78% and 83% of stroke had aPL [11, 29] , and 36% of myocardial infarction [25] . They highlighted that the presence of multiple aPL with moderate serum titers of at least one type of aPL was found to be statistically associated with a higher incidence of cerebral infarction [11, 45] .

It was not possible to extract data from the primary studies to determine an overall association between aPL positivity and thromboses. A meta-analysis of individual patients' data would be timely to draw definitive conclusions.

Overall results reported are conflicting. Two studies have studied coagulation in COVID-19 patients with or without LA. Patients with LA had a higher level of inflammation markers (CRP and fibrinogen) but the same level of D-Dimers [32, 55] . Zuo et al. showed a positive association with the presence of Neutrophil Extracellular Traps (NETs), platelet count and neutrophil activation (by calprotectin assay) [41• ]. They did not find a statistical association with levels of D-Dimers. Likewise, one study showed that levels of D-Dimers, ferritin and CRP were higher in COVID-19 patients with aPL [12] while another comparison between patients with or without autoantibodies (including aPL and antinuclear antibodies) [39] and did not find any significant difference in blood parameters. Several studies did not show any differences between COVID-19 patients with aPL or not [28, 43, 45, 46] . Finally, Bauer et al. did not find any difference on activated protein-C resistance between patients with or without COVID-19 [47] .

There was a great discrepancy in aPL prevalence in studies, from 0% to 90% according to aPL type and isotype. A high proportion of LA were identified in ICU patients. There was a high prevalence of IgA isotypes during COVID-19 infection. Several studies suggested an association between aPL and a high incidence of thrombotic events. However other studies question this association between aPL and thrombotic events and some questions remain unsolved.

Zhang et al. were the first to suggest a pathogenic role of aPL. They found aCL and aβ 2 -GPI IgA positivity in stroke patients. Although IgA is one of the unconventional aPL, it has been described as a potential source of thrombosis and pregnancy morbidity [59] . Furthermore Hasan Ali et al. confirmed in their study that IgA were highly and independently associated with COVID-19 [60] . Similar data were later reported by other studies linking thrombosis to other isotypes of aPL, and suggested a pathogenic role, partly because they are more prevalent in severe patients in ICU. Pathological mechanisms 

It is widely known that aPL can appear during a viral infection. During other viral infections, aPL prevalence varies from 2% to 63% depending on the aPL studied, they are classical known to be transient and non-pathogenic [64, 65] . Yet during COVID-19, some authors have suggested a pathological role to aPL to explain high number of thrombotic events. However some authors did not show any relationship of aPL and thromboses [18, 54, 55] . Differences of aPL prevalence could be observed in all types of aPL studied. The main reason is probably linked to aPL tests and the interpretation of the results. Assays may be affected by several analytical factors, including methodological issues due to the heterogeneity of aPL, different tests from one laboratory to another, and preanalytical factors due to the clinical condition of the patient in whom the assay is performed [57] . In particular inflammation may cause false positive determination of LA [66] [67] [68] . The latest recommendations of the ISTH suggest not to test for LA in the acute phase of inflammation when possible [69••] . The presence of anticoagulant treatments may also interfere with LA tests [56, 70] , and finally a higher prevalence of aPL is usually found in elderly people with chronic diseases (up to 18%), who are at high risk for severe COVID-19 [71] [72] [73] , and in severe patients in ICU without COVID-19 [74, 75] .

Presence or absence of aPL is not sufficient to determine the patient's thrombotic profile: high aPL titers and the simultaneous presence of several aPL increase thrombotic risk [76, 77] . Isolated LA is an independent risk factor for myocardial infarction and ischemic stroke [78, 79] , but interpretation of positivity may be difficult in critical care patients.

Many studies do not clearly report titers, associations of several aPL and their isotypes. Finally, the severity of the clinical condition could explain in part the presence of aPL.

The persistence over time has been studied only twice [13, 52] . Results with the low persistence of aPL at one month must be contrasted by the large number (more than 50%) of those lost for follow-up in the first study. Indeed, the followup in this situation is difficult, especially in ICU patients, with many deaths. The second cohort study did not find any LA in patients tested again. The other aPL seem to be more persistent, suggesting that positive LA can be frequent in COVID-19 patients at their admission in relation to the acute inflammatory phase.

It has been reported that the majority of aPL tested in ICU patients were identified within 10 days of admission [53] . A study of conventional and unconventional aPL at different time points of COVID-19 infection [35•] . Suggested that during the course of the infection, prevalence of different aPL varied over time, possibly linked to the inflammatory phase of the disease. The types of aPL may also vary over time [45] . Unfortunately, their long term persistence overtime has not been studied in most instances.

Based on these data, routine screening of aPL in COVID-19 patients may be questioned. There are no specific recommendations about aPL and their determination in COVID-19 patients, but the American Society of Hematology (ASH) stated that "there are only very limited data on aPL antibodies in COVID-19 and it is unclear if they represent an epiphenomenon or are actually involved in any haemostatic abnormalities seen in COVID-19 disease" [80] .

However, their pathogenic role remains possible. While a systematic screening does not seem indicated, we suggest that aPL testing should be performed in COVID-19 patients with thrombotic events. In addition as indicated in the general recommendations, [69, 80] patients with, thrombotic storms, venous thrombosis at unusual sites or despite preventive anticoagulation or arterial thrombosis in younger patients (<50 years) as well as suggestive obstetrical history or underlying systemic autoimmune diseases should lead to an aPL assessment.

In the same recommendations, patients with systemic lupus erythematosus and COVID-19 should be tested for LA and other aPL in order to assess their thrombotic risk. Indeed, the presence of this antibodies, and even more so their association, would change their management.

When aPL assay is indicated, only LA, IgG/IgM aCL, and IgG/IgM aβ 2 -GPI should be performed routinely. Indeed, the impact and the role in clinical practice of unconventional aPL (IgA isotype especially), are still debated [59, 81, 82] . Thus, their determination is recommended in well-designed research protocols [76, 83] .

In all cases the interpretation of the presence of LA in ICU patients must be done with care due to the inflammatory state of the patients. Titers and combination of aPL should be taken into account for anticoagulant treatment decisions in case of thrombosis. Finally, all identified aPL should be systematically confirmed at 3 months whenever possible.

Simple descriptive data are not sufficient to clearly determine aPL involvement in COVID-19 infection. Further follow-up studies to research the persistence of these antibodies over time are needed. More studies directly investigating the pathogenic role of aPL are important. The issue will be to determine if they participate directly in thrombosis, or if their presence is only an additional feature of the major infectious pro-inflammatory state of the disease. Future multicenter studies must also standardize with aPL assessment to harmonize the timing of tests, preanalytical and analytical variables and results and their interpretations in this specific context and use a core laboratory if necessary. The determination of the role of unconventional aPL should also be explored in future studies.

COVID-19 is a new viral disease causing frequent thrombotic events. The designation of the "perfect culprits", aPL, has been discussed since the initial findings. However, aPL are frequently found in infected patients. COVID-19 patients experience many thrombotic complications, particularly in ICU, for which aPL could be responsible and that may require specific anticoagulant strategies. aPL screening should currently be reserved for COVID-19 patients with thrombosis or in specific situations such as underlying auto-immune diseases. Finally, more studies investigating the pathogenic role of aPL are important, as well as further follow-up studies to research the persistence of these antibodies over time are needed.

Conflict of Interest The authors declare that they have no competing interest.

Human and Animal Rights and Informed Consent This article does not contain any studies with human or animal subjects performed by any of the authors.

Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study

Clinical features of patients infected with 2019 novel coronavirus in Wuhan

Prevalence of venous thromboembolism in patients with severe novel coronavirus pneumonia

Incidence of thrombotic complications in critically ill ICU patients with COVID-19

High incidence of venous thromboembolic events in anticoagulated severe COVID-19 patients

High risk of thrombosis in patients with severe SARS-CoV-2 infection: a multicenter prospective cohort study

Abnormal coagulation parameters are associated with poor prognosis in patients with novel coronavirus pneumonia

Anticoagulant treatment is associated with decreased mortality in severe coronavirus disease 2019 patients with coagulopathy

Coagulopathy and antiphospholipid antibodies in patients with Covid-19

High frequency of antiphospholipid antibodies in critically ill COVID-19 patients: a link with hypercoagulability?

Neurological manifestations in critically Ill patients with COVID-19: a retrospective study

Presence of antiphospholipid antibodies in COVID-19: case series study

Antiphospholipid antibodies in patients with COVID-19: a relevant observation?

Anti-phospholipid Antibodies in COVID-19 are different from those detectable in the anti-phospholipid syndrome

Profile of natural anticoagulant, coagulant factor and anti-phospholipid antibody in critically ill COVID-19 patients

COVID-19 associated coagulopathy in critically ill patients: a hypercoagulable state demonstrated by parameters of haemostasis and clot waveform analysis

Life-threatening COVID-19 presenting as stroke with antiphospholipid antibodies and low ADAMTS-13 activity, and the role of therapeutic plasma exchange: A case series

Are antiphospholipid antibodies associated with thrombotic complications in critically ill COVID-19 patients?

Antiphospholipid antibodies and thrombosis in patients with COVID-19

Autoantibodies related to systemic autoimmune rheumatic diseases in severely ill patients with COVID-19

Lupus anticoagulant is frequent in patients with Covid-19

Lupus Anticoagulant and Abnormal Coagulation Tests in Patients with Covid-19

High prevalence of antinuclear antibodies and lupus anticoagulant in patients hospitalized for SARS-CoV2 pneumonia

Changes in characteristics and management among patients with ST-elevation myocardial infarction due to COVID-19 infection

Antiphospholipid antibodies are not elevated in patients with severe COVID-19 pneumonia and venous thromboembolism

Frequency and clinical correlates of antiphospholipid antibodies arising in patients with SARS-CoV-2 infection: findings from a multicentre study on 122 cases

Assessment of Lupus Anticoagulant Positivity in Patients With Coronavirus Disease

Acute Cerebrovascular Events in Hospitalized COVID-19 Patients

Clinically significant anticardiolipin antibodies associated with COVID-19

The pathogenesis of thromboembolic disease in covid-19 patients: Could be a catastrophic antiphospholipid syndrome?

Lupus anticoagulant and mortality in patients hospitalized for COVID-19

Delayed pulmonary embolism after COVID-19 pneumonia: a case report

Clinical spectrum of ischaemic arterial diseases associated with COVID-19: a series of four illustrative cases

Anti-phospholipids antibodies and immune complexes in COVID-19 patients: a putative role in disease course for anti-annexin-V antibodies

Transient Anti-Phospholipid Antibodies in Two Patients With COVID-19

Antiphospholipid antibodies and thrombotic events in COVID-19 patients hospitalized in medicine ward

Characteristics of ischaemic stroke associated with COVID-19

COVID-19 and immunological dysregulation: can autoantibodies be useful?

Anticardiolipin IgG Autoantibody Level Is an Independent Risk Factor for COVID-19 Severity

Prothrombotic autoantibodies in serum from patients hospitalized with COVID-19

Prevalence of autoantibody responses in acute coronavirus disease 2019 (COVID-19)

High Prevalence of Acquired Thrombophilia Without Prognosis Value in Patients With Coronavirus Disease

Thrombosis and antiphospholipid antibodies in patients with SARS-COV-2 infection (COVID-19)

Antiphospholipid Antibodies in Critically Ill Patients With COVID-19

Lupus anticoagulant in patients with COVID-19

A Matter of Caution: Coagulation Parameters in COVID-19 Do Not Differ from Patients with Ruled-Out SARS-CoV-2 Infection in the Emergency Department

Antiphospholipid antibodies in COVID-19-associated pneumonia patients in intensive care unit

Latent rheumatic, thyroid and phospholipid autoimmunity in hospitalized patients with COVID-19

Beta-2-Glycoprotein-I Deficiency Could Precipitate an Antiphospholipid Syndrome-like Prothrombotic Situation in Patients With Coronavirus Disease

COVID-19 Disease and hypercoagulability leading to acute ischemic stroke

Follow-up of COVID-19 patients: LA is transient but other aPLs are persistent

Anticardiolipin and other antiphospholipid antibodies in critically ill COVID-19 positive and negative patients

Prevalence and clinical significance of antiphospholipid antibodies in patients with coronavirus disease 2019 admitted to intensive care units: a prospective observational study

Lupus anticoagulant single positivity at acute phase is not associated with venous thromboembolism or inhospital mortality in COVID-19

Update of the guidelines for lupus anticoagulant detection

Testing for antiphospholipid antibodies: advances and best practices

Is the association between IgG anti-cardiolipin autoantibodies and COVID-19 severity related to the lung injury or to the SARS-CoV-2 infection?

The clinical relevance of IgA anticardiolipin and IgA anti-β2 glycoprotein I antiphospholipid antibodies: a systematic review

Severe COVID-19 is associated with elevated serum IgA and antiphospholipid IgA-antibodies

Endothelial cell-activating antibodies in COVID-19. medRxiv

Release of neutrophil extracellular traps by neutrophils stimulated with antiphospholipid antibodies: a newly identified mechanism of thrombosis in the antiphospholipid syndrome

Antiphospholipid antibody-activated NETs exacerbate trophoblast and endothelial cell injury in obstetric antiphospholipid syndrome

Risk of developing antiphospholipid antibodies following viral infection: a systematic review and meta-analysis

Viral infections and antiphospholipid antibodies

Coagulopathy of COVID-19 and antiphospholipid antibodies

Lupus anticoagulant is significantly associated with inflammatory reactions in patients with suspected deep vein thrombosis

Lupus Anticoagulant (LAC) testing in patients with inflammatory status: does C-reactive protein interfere with LAC test results?

Guidance from the Scientific and Standardization Committee for lupus anticoagulant/ antiphospholipid antibodies of the International Society on Thrombosis and Haemostasis: Update of the guidelines for lupus anticoagulant detection and interpretation

The Effects of Indirect-and Direct-Acting Anticoagulants on Lupus Anticoagulant Assays: A Large, Retrospective Study at a Coagulation Reference Laboratory

High detection rates of antithrombin deficiency and antiphospholipid syndrome in outpatients aged over 50 years using the standardized protocol for thrombophilia screening

Lupus anticoagulant identifies two distinct groups of patients with different antibody patterns

Anti-phospholipid antibody prevalence and association with subclinical atherosclerosis and atherothrombosis in the general population

Antiphospholipid antibodies and infections

Frequent development of lupus anticoagulants in critically ill patients treated under intensive care conditions

International consensus statement on an update of the classification criteria for definite antiphospholipid syndrome (APS)

Laboratory Diagnostics of Antiphospholipid Syndrome

Antiphospholipid antibodies and risk of myocardial infarction and ischaemic stroke in young women in the RATIO study: a case-control study

Clinical and laboratory characteristics of isolated lupus anticoagulants

COVID-19 and aPL Ab -Hematology.org

Current Promising Biomarkers and Methods in the Diagnostics of Antiphospholipid Syndrome: A Review. B i o m e d i c i n e s . 2 0 2 1 ; 9

The clinical relevance of noncriteria antiphospholipid antibodies

14th International Congress on Antiphospholipid Antibodies Task Force. Report on antiphospholipid syndrome laboratory diagnostics and trends

Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations