Abstract
Nebulised unfractionated heparin (UFH) might reduce time to ventilator separation in patients with COVID-19 by reducing virus infectivity, pulmonary coagulopathy, and inflammation, but clinical trial data are limited. Between 1 July 2020 and 23 March 2022, we conducted, at two hospitals in Victoria, Australia, a randomised, parallel-group, open-label, controlled trial of nebulised UFH. Eligible patients were aged 18 years or more, intubated, under intensive care unit management, had a PaO2 to FIO2 ratio of 300 or less, had acute opacities affecting at least one lung quadrant and attributed to COVID-19, and were polymerase chain reaction-positive for SARS-CoV-2 or had further testing planned. The target sample size was 270, however, the trial was stopped due to slow recruitment. There were 50 enrolments, all of whom were analysed. The median age was 55 (interquartile range (IQR) 46–64) years, 28 (56%) were males, and 46 (92%) had acute respiratory distress syndrome. Twenty-seven (54%) were randomised to nebulised heparin and 23 (46%) to standard care. Nebulised UFH was administered to the heparin group on 6 (IQR 4–10) days; median daily dose of 83 (IQR 75–88) kIU. The primary outcome, time to separation from invasive ventilation to day 28 adjusted for the competing risk of death, was not significantly different between groups but took numerically longer in the nebulised heparin group (12.0, standard deviation (SD) 10.4 days versus 7.4, SD 6.9 days; hazard ratio (HR) 0.56, 95% confidence interval (CI) 0.31 to 1.01, P = 0.052). One patient died by day 28 in each group, fewer than expected. Time to separation from invasive ventilation among survivors to day 28 occurred more quickly than expected in the standard care group and was, without correction for multiple comparisons, significantly slower in the heparin group (11.3, SD 10.0 days, n = 26 versus 6.4, SD 5.2 days, n = 22; HR 0.52, 95% CI 0.30 to 0.92, P = 0.024). Nebulised heparin did not reduce time to ventilator separation in intubated adult patients with COVID-19. The study is limited by the small sample size and potential for sampling bias. Further study is required.
Introduction
Unfractionated heparin (UFH) has antiviral, anticoagulant and anti-inflammatory properties.1 –9 When administered by nebulised inhalation, UFH might limit epithelial invasion by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) and ameliorate the pulmonary coagulopathy and inflammation associated with coronavirus disease 2019 (COVID-19) leading to improved clinical outcomes, but data from clinical trials is limited. The scientific rationale for nebulised UFH as a treatment for COVID-19 has previously been described by our group and others.10 –12
UFH has been shown in preclinical studies to induce conformational change in the SARS-CoV-2 spike (S1) protein receptor-binding domain and inhibit invasion of Vero cells at concentrations that could be achieved by nebulisation.1 –3 The pulmonary histopathological features of fatal COVID-19 include progressive diffuse alveolar damage, which is initially characterised by widespread pulmonary oedema, fibrin deposition and hyaline membrane formation, as well as damage to the alveolar-capillary barrier, and excessive thrombosis and impaired clot fibrinolysis.13 –15 UFH inhibits coagulation activation through a range of mechanisms, including catalysing the action of antithrombin, promoting tissue factor pathway inhibitor expression, reducing tissue factor expression, increasing endothelial expression of heparan sulphate and through release of tissue-plasminogen activator by the endothelium.4 –7 The anti-inflammatory effects of UFH include inhibiting chemokines, inhibiting leucocyte adhesion to endothelial cells, inhibiting neutrophil elastase activity, and moderating complement activity.8,9
Early-phase trials of nebulised UFH in intubated patients with acute lung injury and related conditions found nebulised UFH was safe and reduced pulmonary dead space, pulmonary coagulation activation and the Murray Lung Injury Score (Murray Score), and increased time free of ventilatory support.16 –19 A pre-pandemic, multicentre, randomised, double-blind, placebo-controlled phase 3 trial of nebulised UFH enrolled invasively ventilated intensive care unit (ICU) patients with impaired oxygenation who were expected to require invasive ventilation beyond the next calendar day, 47% of whom had acute respiratory distress syndrome (ARDS). 20 Nebulised UFH did not significantly improve the primary endpoint, the SF-36 Physical Function score of survivors at day 60. However, secondary outcomes showed less deterioration in the Murray Score, fewer cases of progression to ARDS and earlier return home, and, in a sub-group analysis defined by the median baseline Murray Score, a relative advantage for the effect of nebulised UFH was observed on time to ventilator separation among those with higher Murray Scores. We hypothesised that nebulised UFH could reduce the duration of mechanical ventilation in patients requiring invasive ventilation for COVID-19.
Methods
This was an investigator-initiated, randomised, parallel-group, open-label, controlled trial conducted at two hospitals in Victoria, Australia. The trial was prospectively registered (Australian New Zealand Clinical Trials Registry, ACTRN1262000517976; CHARTER study) and the trial protocol prospectively published. 21 St Vincent’s Hospital (Melbourne, Australia) Human Research Ethics Committee (HREC) granted ethics committee approval for the two participating centres (Reference HREC 086/20; Project ID 64197). Funds provided by the Department of Critical Care Medicine, St Vincent’s Hospital, Melbourne, Australia, supported the study.
Eligible patients were aged 18 years or older and under ICU management, had an endotracheal tube in place and had been intubated on the day of randomisation or the preceding day, had impaired oxygenation defined by PaO2 to FIO2 ratio of 300 or less while intubated, had acute opacities on chest X-ray or computed tomography affecting at least one lung quadrant and attributed to COVID-19, and had returned a polymerase chain reaction (PCR)-positive sample for SARS-CoV-2 or had further testing planned. These characteristics indicated the presence of acute lung injury according to Murray Score criteria, 22 and oxygenation impairment consistent with ARDS, 23 due primarily to COVID-19.
Exclusions on safety grounds were heparin allergy, heparin-induced thrombocytopenia, activated partial thromboplastin time (APTT) greater than 120 s that was not due to anticoagulant therapy, platelet count less than 20 × 109/L, uncontrolled bleeding, and pulmonary bleeding. Use of extracorporeal membrane oxygenation (ECMO) was an exclusion, as this could impede delivery of the intervention. Myopathy or nerve injury or disease with likely prolonged incapacity to breathe, acute brain injury, home oxygen, dependency due to physical or cognitive decline, and imminent or inevitable death were exclusions due to their potential to confound outcome assessments.
Patients lacked capacity to consider participation at the time of eligibility due to their critical illness and intubated status, and their substitute decision makers were unable to attend the hospital due to pandemic-associated visitor restrictions. Therefore, in accordance with the HREC-approved procedure, verbal (telephone) informed consent was obtained from the substitute decision maker prior to enrolment, and written study information, including the option to withdraw, was provided by email or post.
Patients were randomised in a one-to-one ratio, stratified by site, to usual care without nebulised heparin (standard care) or to usual care with nebulised heparin (nebulised heparin). Randomisation was performed using a web-based, automated system hosted at the University of Sydney (Sydney, Australia). To ensure pre-enrolment allocation concealment, the random allocation sequence used blocks of variable size and a random seed, and the sequence was developed by a researcher (GSD) working independently of the investigators and clinicians. Upon randomisation, clinicians and investigators were notified of the participant’s randomised treatment (standard care or nebulised heparin). Knowledge of the randomised treatment was not withheld from participants or their representatives.
For participants assigned to nebulised heparin, the nebulised study medication was UFH sodium 25,000 IU in 5 mL (Pfizer Australia, Sydney, Australia). 24 It was administered 6-hourly using a vibrating mesh nebuliser (Aeroneb Solo, Aerogen, Galway, Ireland) but only while receiving invasive mechanical ventilation and only to day 10. Heated ventilator circuit humidification was employed and an expiratory filter (Servo Duo Guard, Maquet Critical Care AB, Solna, Sweden) was used to prevent the nebulised drug impairing ventilator function. The dose and method of nebulisation were developed in pre-pandemic studies by our group in patients with ARDS and similar conditions.16 –20 All other aspects of participant management were at the discretion of the treating physicians, including investigations, use of immune therapies and antimicrobials, and use of intravenous and subcutaneous UFH and low molecular weight heparin (LMWH) at prophylactic and therapeutic doses.
The primary outcome was time to separation from invasive ventilation to day 28, where non-survivors were treated as not separated from ventilation. Invasive ventilation was defined as positive pressure via an endotracheal or tracheostomy tube. If separation occurred more than once, the final separation was used to calculate the outcome.
Pre-specified secondary efficacy endpoints were: deterioration in driving pressure, oxygenation index, and ventilatory ratio, calculated by subtracting the baseline value from the highest of the values measured each day, if invasively ventilated, on day 2 and day 4; deterioration in white cell count, C-reactive protein, D-dimer and international normalised ratio (INR), calculated by subtracting the baseline value from the highest value measured whilst in the ICU up to day 10; post-randomisation initiation of neuromuscular blockade, prone positioning, and ECMO by day 10; tracheostomy by day 28; time to separation from invasive ventilation to day 28 among survivors; time to ICU separation to day 28, where non-survivors were treated as though not separated and, if a participant achieved separation more than once, the final separation was used to calculate the outcome; time to ICU separation to day 28 among survivors; survival to day 28; number residing at home at day 60; number at home at day 60 among survivors; and survival to day 60.
Safety outcomes were red cell transfusion, red cell transfusion volume, maximum absolute fall in haemoglobin from baseline, and maximum percentage fall in platelet count from baseline, assessed for each outcome whilst in ICU up to day 10. Pre-defined reportable serious adverse events were pulmonary bleeding, major bleeding, 25 and heparin-induced thrombocytopenia.
Predicted body weight was calculated using the ARDSNet methodology. 26 ARDS was determined by the Berlin criteria. 23 The Murray Score was determined with respiratory compliance defined as tidal volume divided by the difference of peak inspiratory and expiratory pressures. 27 Driving pressure was determined as the difference of peak inspiratory and expiratory pressures.28,29 Oxygenation index was calculated as mean airway pressure multiplied by percentage FIO2 divided by PaO2, where mean airway pressure was the sum of the peak inspiratory and expiratory pressures divided by two. 30 Ventilatory ratio was calculated as the product of minute ventilation multiplied by PaCO2 divided by the product of predicted body weight multiplied by 100 multiplied by 37.5. 31
Chest imaging was assessed during screening by a medical site investigator using a protocol-defined methodology. Site research personnel collected data from participants’ health records and, to ascertain day 60 vital status and place of residence, directly contacted the participants or their representatives. If a participant was transferred to the ICU of a non-participating hospital, data were obtained from the receiving hospital to inform day 28 and day 60 outcomes, including the primary outcome, but not day 4 or day 10 outcomes. Treatment with Janus kinase inhibitors was collected from September 2021 after their use became recommended in guidelines.32,33 Data were entered into a password-protected REDCap® electronic data capture platform 34 hosted by St Vincent’s Hospital (Sydney, Australia). Day 0 was defined as the period from randomisation to midnight on the day of enrolment, day 1 the first calendar day after the day of enrolment, and so forth.
To demonstrate a 50% improvement in the hazard of separation from invasive ventilation with survival to day 28, a sample of 270 was required. This assumed 20% overall mortality, 35 5% overall who survived but failed to achieve ventilator separation, 20 5% overall withdrawn as might have occurred due to consent withdrawal, one-to-one allocation ratio, power 80%, and a two-sided significance level of 0.05.
The statistical analysis plan was agreed by the investigators a priori. Analyses followed the intention-to-treat principle and were performed using Stata statistical software (Version 17, StataCorp LLC, College Station, Texas, USA). Hypothesis tests were two-sided. P-values of secondary outcomes were not adjusted for multiple comparisons and should not be used to infer definite differences between groups. For one participant, although the date of ventilator separation and date of ICU separation were known, the time of day of each was not known. For this participant, the missing time-of-day data were imputed using the median values of participants from the same enrolment centre and study group. There was no other imputation of missing data.
On time to separation from invasive ventilation to day 28 (the primary outcome) and time to separation from ICU to day 28, the groups were compared by hazard ratio (HR) calculated using the Fine–Gray methodology to account for the competing risk of death. 36 Data for the primary outcome are reported as mean (standard deviation (SD)) and are presented using Kaplan–Meier curves, with the deceased treated as though ventilated at the end of day 28. Survival was assessed using the HR calculated by Cox regression. Normally distributed continuous data are displayed as mean (SD), and differences between groups assessed using the Student’s t-test. Skewed continuous data are displayed as median and interquartile range (IQR), or as a range if the number of observations is small, and differences assessed using the Wilcoxon rank sum test. Discrete data are displayed as number and percentage of total number non-missing. Binary outcomes were assessed by logistic regression, or by exact logistic regression if there was a count of zero. There were no sub-group, adjusted or post hoc analyses.
Results
Screening commenced on 1 July 2020 and was ceased on 23 March 2022. There were 50 patients enrolled, representing 19% of the target sample size. The decision to stop the trial was made by the investigators based on their judgement that recruitment of the target sample was not feasible in the context of high levels of vaccination, 37 increasing availability of disease-modifying therapies for COVID-19, 38 and a dearth of eligible patients presenting to the two participating centres. Enrolments occurred during two distinct periods, separated by an almost 12-month hiatus, associated with waves of COVID-19 ICU admissions 39 and, at the time of stopping the trial, there had not been an enrolment for more than 50 days (Figure 1(a)). The investigators were not aware of the study outcomes when the decision to stop the trial was made. The HREC was notified of the decision.
(a) Cumulative enrolment. Screening commenced on 1 July 2020 and the first patient was enrolled on 9 July 2020. The 15th patient was enrolled on 13 September 2020, but it was almost a year until the 16th patient was enrolled on 4 September 2021. The 50th enrolment occurred on 29 January 2022 and there were no further enrolments when screening ceased on 23 March 2022. The dashed line indicates the period between the 50th enrolment and cessation of screening and (b) Trial profile. aInclusion criteria: age ≥18 years; in the intensive care unit (ICU) or scheduled for transfer to ICU; endotracheal tube in place; intubated yesterday or today; PaO2 to FIO2 ratio ≤300 while intubated; acute opacities not fully explained by effusions, lobar/lung collapse and nodules, affecting at least one lung quadrant on chest X-ray or computed tomography; the acute opacities on chest X-ray or computed tomography are most likely due to coronavirus disease 2019 (COVID-19); and there is a polymerase chain reaction-positive sample for severe acute respiratory syndrome-2 (SARS-CoV-2) within the past 21 days or results are pending or further testing planned. bThe patient was withdrawn by the responsible decision maker on day 6; permission to collect data was granted. cThe patient was withdrawn by the responsible decision maker early on study day 1; permission to collect data was granted.
The trial profile is shown in Figure 1(b). Of 134 patients who satisfied the trial inclusion criteria, 84 were not enrolled, including 30 who refused participant consent and 30 expected to be transferred elsewhere for caseload sharing. Of the 50 enrolled patients, 27 (54%) were randomly assigned to nebulised heparin and 23 (46%) standard care. One participant in the heparin group was withdrawn by their substitute decision maker from further study intervention, but approval for use of data was granted. All enrolled patients satisfied the eligibility criteria, although one patient, in the heparin group, who was rapid antigen test-positive for SARS-CoV-2 and had a clinical picture of COVID-19, did not undergo planned PCR testing due to the site testing regimen in place at that time. All enrolled patients were included in the analysis.
Ventilator separation to day 28. Kaplan–Meier estimates of the probability of separation from the ventilator to day 28. Hazard ratio in the nebulised heparin group 0.56 (95% confidence interval 0.31 to 1.01; P = 0.052). Those deceased by day 28 are deemed to have not had ventilator separation; there was one death in each group.
Baseline demographic and comorbid conditions of participants are shown in Table 1. Information pertaining to baseline characteristics, process of care assessments, efficacy and safety is for the nebulised heparin group versus the standard care group, unless otherwise specified. The nebulised heparin group and standard care group were close in age (56 (IQR 38–63) versus 53 (IQR 49–65) years). The heparin group had a preponderance of males (18/27 (67%) versus 10/23 (43%)). Vaccination against COVID-19 was available to those enrolled after 2020 but only three patients, all in the heparin group, had completed a course of vaccination, whilst another three, all in the standard care group, were partially vaccinated.
Baseline characteristics of patients by treatment allocation group and for all participants.
Data are number (percentage of total non-missing), mean (standard deviation), and median (interquartile range). The number of patients available for specific variables is stated in each cell if different from the total number in the treatment allocation group.
Each partially vaccinated participant had received one dose of the Pfizer/BioNTech vaccine.
Two of the fully vaccinated participants had each received two doses of the Pfizer/BioNTech vaccine, and one had received two doses of the Oxford/AstraZeneca vaccine.
During the 24 h before randomisation.
Janus kinase inhibitor use collected from 23 September 2021.
Oseltamivir, laninamivir, zanamivir or peramivir.
PBW males = 50 + 0.91 (cm of height −152.4); PBW females = 45.5 + 0.91 (cm of height −152.4).
Driving pressure = peak inspiratory pressure − PEEP.
Compliance = tidal volume ÷ (peak inspiratory pressure − PEEP).
Oxygenation index = (mean airway pressure × FIO2 × 100) ÷ PaO2; mean airway pressure = (peak inspiratory pressure + PEEP) ÷ 2.
Ventilatory ratio = (minute ventilation × PaCO2) ÷ (predicted minute ventilation × ideal PaCO2); minute ventilation = respiratory rate x tidal volume; predicted minute ventilation = PBW × 100; and ideal PaCO2 = 37.5 mmHg.
Each component score is out of 4. The final score is the sum of the component scores divided by the number of components.
APTT: activated partial thromboplastin time; ARDS: acute respiratory distress syndrome; COPD: chronic obstructive pulmonary disease; COVID-19: coronavirus disease 2019; CXR: chest X-ray; FEU: fibrinogen equivalent units; FIO2: fraction of inspired oxygen; IL: interleukin; IV: intravenous; PaCO2: arterial partial pressure of carbon dioxide; PaO2: arterial partial pressure of oxygen; PEEP: positive end-expiratory pressure; PBW: predicted body weight; SARS-CoV-2: severe acute respiratory syndrome coronavirus-2; SC: subcutaneous; UFH: unfractionated heparin.
Baseline use of heparin, immune therapies and antimicrobials, and coagulation markers are shown in Table 1. Subcutaneous or intravenous UFH or LMWH, typically enoxaparin, had been administered to most patients in each group (25/27 (93%) versus 20/23 (87%)). Enoxaparin more than 40 mg daily had been administered to a similarly considerable proportion of each group (9/27 (33%) versus 9/23 (39%)). The groups were alike in the proportions treated with corticosteroids (27/27 (100%) versus 20/23 (87%)), a Janus kinase inhibitor (14/17 (82%) versus 9/14 (64%)), remdesivir (12/27 (44%) versus 9/23 (39%)), a non-macrolide antibacterial (22/27 (81%) versus 16/23 (70%)), and a macrolide (15/27 (56%) versus 13/23 (57%)). D-dimer values were similarly elevated in each group (1.03 (IQR 0.80–1.48), n = 23 versus 0.80 (IQR 0.54–2.27) mg/L fibrinogen equivalent units (FEU), n = 19). The APTT was within normal limits and similar in the groups (34 (IQR 32–37) versus 33 (IQR 31–35) s, n = 21).
Baseline respiratory characteristics are shown in Table 1. The groups were similar in the substantial interval since onset of COVID-19 symptoms (8 (IQR 6–11) versus 9 (IQR 7 to 13) days) and in the time since intubation (10 (IQR 5–23) versus 10 (IQR 4–19) h). Similar proportions of each group were treated with neuromuscular blockers (10/27 (37%) versus 8/23 (35%)) and similarly small proportions with prone positioning (3/27 (11%) versus 2/23 (9%)). ARDS, irrespective of time since onset, was highly prevalent in each group (24/27 (89%) versus 22/23 (97%), but its prevalence reduced substantially when only cases whose onset occurred within 1 week were counted (6/27 (22%) versus 5/23 (22%)). Murray Scores of each group were consistent with severe lung injury (2.81, SD 0.51 versus 2.79, SD 0.50) and the number of affected lung quadrants was high in each group (3 (IQR 2–4) versus 4 (IQR 3–4)). The groups each had elevated driving pressure (13, SD 3 versus 13, SD 4 cm H20), oxygenation index (11.82 (IQR 8.76–14.33) versus 10.29 (IQR 8.47–14.64)) and ventilatory ratio (1.49 (IQR 1.19–1.78) versus 1.44 (IQR 1.31–2.05)).
Details regarding the use of nebulised UFH are shown in Table 2. All those assigned nebulised heparin received nebulised UFH and the therapy was administered on 6 (IQR 4–10) days at a median daily dose of 83 (IQR 75–88) kIU. However, seven participants, all in the heparin group, were transferred to the ICU of a non-participating hospital on day 2, 3, 3, 4, 7, 7, and 10, respectively, preventing further administration of the study medication, and another participant, also in the heparin group, was withdrawn from study treatment on day 6. The reason for transfer was the ongoing management of ECMO for one participant, to utilise private health insurance for another, and caseload sharing for five. One participant in the standard care group received a single dose of nebulised UFH in breach of the protocol.
Process of care assessments in ICU to day 10 by treatment allocation group.
Data are number (percentage of total non-missing), median (interquartile range), and [range]. The number of patients available for specific variables is stated in each cell if different from the total number in the treatment allocation group. Seven patients, all in the heparin group, were transferred to the ICU of another hospital prior to the end of day 10; APTT results are reported up to the time of transfer; use of therapies and organ supports that had not been administered by the time of transfer is reported as unknown.
Days on which the patient received one or more doses of the drug.
Janus kinase inhibitor use collected from 23 September 2021.
Oseltamivir, laninamivir, zanamivir or peramivir.
APTT: activated partial thromboplastin time; CI: confidence interval; ICU: intensive care unit; IL: interleukin; IV: intravenous; OR: odds ratio; SC: subcutaneous; UFH: unfractionated heparin.
Post-enrolment use of systemic heparin, immune therapies and antimicrobials, and APTT results are shown in Table 2. Most patients received subcutaneous or intravenous enoxaparin, and half of each group received more than 40 mg enoxaparin per day. Although peak APTT was similar in the groups, among the small number who received intravenous UFH, there was a large numerical difference in peak APTT with nebulised UFH (range 96–200, n = 2 versus range 48–65, n = 2; P = 0.12). Use of immune therapies and antimicrobial therapies after enrolment was more extensive than at baseline. The groups had similar proportions treated after enrolment with corticosteroids (27/27 (100%) versus 23/23 (100%)), a Janus kinase inhibitor (15/17 (88%) versus 11/14 (79%); P = 0.47), remdesivir (11/24 (46%) versus 11/23 (48%); P = 0.89), a non-macrolide antibacterial (25/27 (93%) versus 21/23 (91%); P = 0.87), and a macrolide (15/23 (65%) versus 18/23 (78%); P = 0.33).
Time to separation from invasive ventilation to day 28 adjusted for the competing risk of death, the primary outcome, did not differ significantly between the groups but was numerically slower in the heparin group (12.0, SD 10.4 versus 7.4, SD 6.9 days; HR 0.56, 95% confidence interval (CI) 0.31 to 1.01, P = 0.052); in the heparin group 1/27 died and 6/26 survivors failed to achieve ventilator separation, while in the standard care group 1/23 died and 0/22 survivors failed to achieve separation; 95% CI for the pointwise Kaplan–Meier estimates of each group were wide and there was a large overlap of the groups (Table 3 and Figure 2).
Efficacy outcomes by treatment allocation group.
Data are number (percentage of total non-missing), mean (standard deviation), median (interquartile range), and median [range]. The number of patients available for specific variables is stated in each cell if different from the total number in the treatment allocation group.
Non-survivors to day 28 are deemed to have never had separation from either invasive ventilation or the ICU.
Calculated by subtracting the baseline value from the highest value measured once daily on day 2 and day 4. Indices were analysed for invasively ventilated patients who were in the study hospital ICU and not receiving extracorporeal membrane oxygenation. Three patients, in the heparin group, were transferred to the ICU of another hospital prior to day 4; their data are reported up to the time of transfer.
Driving pressure = peak inspiratory pressure - PEEP.
Oxygenation index = (mean airway pressure × FIO2 × 100) ÷ PaO2; mean airway pressure
Ventilatory ratio = (minute ventilation × PaCO2) ÷ (predicted minute ventilation × ideal PaCO2); minute ventilation = respiratory rate x tidal volume; predicted minute ventilation = predicted body weight kg × 100; and ideal PaCO2 = 37.5 mmHg.
Instituted after randomisation. Seven patients, in the heparin group, were transferred to another ICU prior to the end of day 10; use of therapies that had not been administered by the time of transfer is reported as unknown.
Calculated by subtracting the baseline measurement from the highest measurement up to day 10 while in the study hospital ICU. Seven patients, in the heparin group, were transferred to the ICU of another hospital prior to end of day 10; their data are reported up to the time of transfer.
Another participant, in the heparin group, had a tracheotomy after day 28, which is not shown here.
Missing data for one participant, in the heparin group, who had a tracheotomy prior to day 28 but the precise date is not known.
CI: confidence interval; HR: hazard ratio; ICU: intensive care unit; Inf: infinity; MD: mean difference; OR: odds ratio.
Secondary efficacy outcomes at day 4 and day 10 are shown in Table 3. Deterioration in driving pressure to day 4 was numerically higher (worse) in the heparin group (+0.48, SD 4.44 versus −1.86, SD 5.32 cmH2O, n = 22; mean difference (MD) 2.35, 95% CI −0.52 to 5.21, P = 0.11) and deterioration in oxygenation index to day 4 was also numerically higher (worse) in the heparin group (−0.42, SD 5.00 versus −2.64, SD 5.25, n = 22; MD 2.22, 95% CI −0.76 to 5.20, P = 0.14), but the groups showed similar deterioration in ventilatory ratio. Post-randomisation initiation of neuromuscular blockade and prone positioning were similar in each group. One participant, in the heparin group and with pre-enrolment pneumomediastinum, had ECMO instituted after enrolment. Deterioration in the D-dimer was numerically lower (better) in the heparin group (+0.07 (IQR −0.20–1.30) mg/L FEU, n = 18 versus +0.66 (IQR −0.03–9.10) mg/L FEU, n = 14, P = 0.16), and the groups had similar deterioration to day 10 in white cell count, INR, and C-reactive protein.
Secondary efficacy outcomes at day 28 and day 60 are shown in Table 3. Tracheostomy by day 28 occurred somewhat more frequently in the heparin group (7/27 (26%) versus 3/23 (13%); odds ratio (OR) 2.33, 95% CI 0.53 to 10.33, P = 0.26). During 2020, tracheostomy occurred later in the heparin group (day 22 and 25, n = 2 versus day 11, n = 1) and another participant, also in the heparin group, was tracheostomised after day 28. Time to separation from invasive ventilation to day 28 among survivors was significantly slower in the heparin group (11.3, SD 10.0 days, n = 26 versus 6.4, SD 5.2 days, n = 22; HR 0.52, 95% CI 0.30 to 0.92, P = 0.024). Separation from ICU to day 28 was numerically slower in the heparin group (13.4, SD 9.5 versus 10.2 SD 6.7 days; HR 0.65, 95% CI 0.36 to 1.15, P = 0.14) as was separation from ICU to day 28 among survivors (12.9 SD 9.2 days, n = 26 versus 9.4 SD 5.5 days, n = 22; HR 0.61, 95% CI 0.35 to 1.07, P = 0.087). Day 28 mortality was similarly low in each group (1/27 (4%) versus 1/23 (4%), HR 0.85, 95% CI 0.05 to 13.23, P = 0.91) as was day 60 mortality (2/27 (7%) versus 1/23 (4%); HR 1.70, 95% CI 0.16 to 18.57, P = 0.66). The proportion residing at home at day 60 was numerically lower in the heparin group (22/27 (81%) versus 22/23 (96%); OR 0.20, 95% CI 0.02 to 1.85, P = 0.16) while the proportion of survivors residing at home at day 60 was similar.
Safety outcomes are shown in Table 4. One participant in each group received a red cell transfusion by day 10. The total red cell transfusion volume by day 10 was numerically larger in the heparin patients (810 mL, n = 1 versus 250 mL, n = 1; P = 0.32). Maximum absolute fall in haemoglobin by day 10 was similar in the groups as was the maximum percentage fall in platelet count by day 10. One participant in the heparin group experienced minor bleeding from a tongue laceration in the context of an elevated APTT while concurrently receiving intravenous UFH. This event was considered medically important and was the subject of a safety alert. There were no cases of pulmonary bleeding, major bleeding, or heparin-induced thrombocytopenia.
Safety outcomes by treatment allocation group.
Data are number (percentage of total non-missing) and mean (standard deviation). The number of patients available for specific variables is stated in each cell if different from the total number in the treatment allocation group.
Evaluated whilst in the study hospital ICU up to day 10. Seven patients, all in the heparin group, were transferred to another ICU prior to the end of day 10; if transfusion had not occurred by the time of transfer, it is reported as unknown; haemoglobin fall, and platelet percentage fall are reported up to the time of transfer.
Packed red cells and whole blood.
Difference of the baseline haemoglobin and the nadir haemoglobin.
Difference of the baseline platelet count and the nadir platelet count as a percentage of the baseline platelet count.
CI: confidence interval; ICU: intensive care unit; MD: mean difference; OR: odds ratio; SD: standard deviation.
Discussion
Between 1 July 2020 and 23 March 2022 we conducted, at two hospitals in Victoria, Australia, an investigator-initiated, randomised, parallel-group, open-label, controlled trial of nebulised UFH in ICU patients receiving invasive mechanical ventilation for COVID-19. When screening was ceased due to slow recruitment, 50 patients had been enrolled, less than one-fifth of the target sample size. Ninety-five percent CIs for all effect estimates were wide, reflecting the modest sample size and variability of the data. On the primary outcome, time to separation from invasive ventilation to day 28 adjusted for the competing risk of death, the groups were not significantly different but there was a numerical trend toward taking longer in the heparin group. One death occurred in each group by day 28, fewer than expected. Time to separation from invasive ventilation among survivors to day 28 occurred more quickly than anticipated in the standard care group and was, without correction for multiple comparisons, significantly slower in the heparin group. Other secondary outcomes were not statistically significant, but we found with heparin a numerical trend toward greater deterioration in driving pressure and the oxygenation index, less deterioration in D-dimer, longer time to ICU separation by day 28 when adjusted for the competing risk of death and separately among survivors, and toward a smaller proportion residing at home at day 60. There were no cases of pulmonary bleeding, major bleeding, or heparin-induced thrombocytopenia.
The small number of participating centres and small sample size are important limitations. Among those who met the trial inclusion criteria, over 60% were not enrolled, predominantly for refusal of consent or imminent transfer to manage caseload, increasing the potential for sampling bias. The day 28 mortality of 4% in each group that we observed contrasts with the findings of an Australian COVID-19 ICU registry, which reported ICU mortality for ventilated patients of over 20%, 35 and suggests we might have enrolled an atypical cohort. Our power calculation assumed 25% of patients would have either died, or be alive but on a ventilator, at the end of day 28. However, we found that, whilst 26% had died or remained ventilated in the heparin group (one death and six ventilated survivors out of 27), only 4% had died or remained ventilated in the standard care group (one death and no ventilated survivors out of 23), a better-than-expected result in the standard care group.
Although the study groups appeared approximately balanced in their baseline characteristics, some differences, such as the preponderance of males in the heparin group, might have prognostic significance. 40 Baseline differences in primary immunodeficiency and SARS-CoV-2 variant could differentially affect outcomes but were not reported.41,42 Recruitment into the study occurred during waves of COVID-19 ICU admissions. Significant changes between these waves in the characteristics and treatment of COVID-19 patients in Australian ICUs have been described and could contribute to baseline imbalance in outcome risk. 39
Between-group differences in clinical care after enrolment that were not likely to be related to the intervention could also have affected the direction of outcomes. The timing of tracheostomy was delayed in several heparin-group patients early in the study, and delayed tracheostomy in critically ill patients is associated with an increased incidence of ventilator-associated pneumonia, fewer ventilator-free days and more ICU days. 43 Also, seven patients, all in the heparin group, were transferred to non-participating ICUs after enrolment, exposing them to the hazards associated with transportation, including greater duration of mechanical ventilation and longer ICU stay;44 –46 and in only one case was the transfer to facilitate specialist-centre care (ECMO management). All participants assigned nebulised heparin received the intervention, which was administered for 6 days on average. Nevertheless, eight patients had their nebulised heparin treatment curtailed due to study treatment withdrawal (1), ECMO (1) and inter-hospital transfer (6), potentially limiting any benefit that might have been derived.
There was a large increase in peak APTT among the small number of patients who were administered both nebulised and intravenous UFH, consistent with our previous experience. 20 One of these patients developed bleeding from a pre-existing tongue laceration that was considered medically significant. We recommend caution if administering both nebulised and intravenous UFH.
The efficacy of heparin in COVID-19 might be affected by the duration and severity of infection and the use of concomitant therapies. The predominant histopathological feature of severe COVID-19 within 10 days of symptom onset is acute diffuse alveolar damage with widespread pulmonary oedema, fibrin deposition and hyaline membrane formation, but severe COVID-19 of longer illness duration is characterised by organising diffuse alveolar damage with alveolar type II cell hyperplasia, fibrotic expansion of interalveolar septa and squamous metaplasia, culminating in fibroblast proliferation, loss of alveoli and pulmonary fibrosis. 15 Nebulised UFH was previously shown to reduce pulmonary coagulation activation, 17 and might thereby reduce pulmonary fibrin deposition and hyaline membrane formation. However, if alveolar damage has progressed to the organising stage, the potential for benefit from nebulised UFH might be reduced and, in the current study, a third of participants had had COVID-19 for 10 days or longer at the time of enrolment. In our pre-pandemic phase 3 trial of ventilated ICU patients, just under half had ARDS at baseline, 21% were concomitantly administered therapeutic systemic UFH or LMWH and 54% were concomitantly administered corticosteroids. 20 In contrast, over 90% of participants in the current study had ARDS (omitting the requirement for onset within 7 days), over 50% were concomitantly administered therapeutic systemic UFH or LMWH, and all participants received corticosteroids. Additional benefit from nebulised UFH in mitigating pulmonary coagulopathy and inflammation might be limited in this context. In a case series of hospitalised COVID-19 patients treated with nebulised UFH, a more rapid improvement in SpO2 to FIO2 ratio was observed in non-intubated than intubated patients, 47 and a pilot randomised study in hospitalised non-intubated COVID-19 patients found a numerical improvement in 28-day mortality and in the World Health Organization modified Ordinal Scale for Clinical Improvement score with nebulised UFH. 48 Similarly, a large randomised trial in hospitalised non-critically ill COVID-19 patients found therapeutic-dose systemic anticoagulation with heparin, compared to usual thromboprophylaxis, significantly reduced use of organ supports, but the corresponding trial in critically ill patients found no benefit.49,50 Finally, heparin has been shown to inhibit SARS-CoV-2 replication in human nasal epithelial cells, 51 and intranasal administration has been demonstrated safe at potentially therapeutic doses. 52 Whether intranasal heparin can prevent COVID-19 infection in exposed individuals is currently being investigated in a multicentre trial. 53
Conclusion
Among intubated adult ICU patients with COVID-19, nebulised UFH did not improve time to separation from invasive mechanical ventilation to day 28. The trial was stopped due to slow recruitment and is limited by its small sample size and potential for sampling bias. Further study of nebulised UFH for COVID-19 is required.
Footnotes
Author Contribution(s)
Data Sharing
All data needed to evaluate the conclusions in this article are presented in the main text. The trial protocol and supporting documents can be obtained by contacting the corresponding author. Requests from bona fide researchers for de-identified individual raw data that underlie the results reported in this article can be made to the corresponding author for consideration by the study management committee. Such requests should be accompanied by a proposal that includes a detailed rationale and analysis plan. Proposals may be submitted up to 24 months after publication of this article. Ethics review board approval and execution of a data sharing agreement are required prior to sharing of data. Sharing of data will occur via a secure data access system.
