Systemically injected oxygen within rapidly dissolving microbubbles improves the outcomes of severe hypoxaemia in swine

Abstract

Acute respiratory failure can cause profound hypoxaemia that leads to organ injury or death within minutes. When conventional interventions are ineffective, the intravenous administration of oxygen can rescue patients from severe hypoxaemia, but at the risk of microvascular obstruction and of toxicity of the carrier material. Here we describe polymeric microbubbles as carriers of high volumes of oxygen (350-500 ml of oxygen per litre of foam) that are stable in storage yet quickly dissolve following intravenous injection, reverting to their soluble and excretable molecular constituents. In swine with profound hypoxaemia owing to acute and temporary (12 min) upper-airway obstruction, the microbubble-mediated delivery of oxygen led to: the maintenance of critical oxygenation, lowered burdens of cardiac arrest, improved survival, and substantially improved neurologic and kidney function in surviving animals. Our findings underscore the importance of maintaining a critical threshold of oxygenation and the promise of injectable oxygen as a viable therapy in acute and temporary hypoxaemic crises.

Fig. 1. Design and fabrication of LmD PMB gas carriers for IVO₂ therapy.

a, LmD PMBs are manufactured through a process of homogenization and simultaneous titration of acid for interfacial crosslinking of the LmD polymer, which is solidified through hydrogen bonding. Following contact with blood, the PMBs rapidly dissolve to deliver gas and their shells immediately revert into their low-MW and soluble molecular constituents, which are excreted via urine and hepatic clearance. b, Cryo-scanning electron microscopy imaging of LmD PMBs depicts their thin shell and smooth surface. c, The infrared absorption peak of carbonyl groups (black curve) in LmD PMBs indicates that they exist in various H-bound states (deconvoluted Gaussian peaks by taking second derivative; blue and red, bound; green, unbound). d, Light microscopy of PMB solutions homogenized in varying dextrose concentrations illustrates that PMB concentration (that is, yield) increases with increasing dextrose additives. Scale bars, 10 µm. e, The effect of osmolarity of carrier fluids on LmD PMBs that were fabricated under 30% dextrose. LmD PMBs originally made with 30% dextrose were not stable in water but were stable in D10 and solutions with higher dextrose concentrations. Data are means ± s.d., biological replicates. f, Size distributions (insets) and microscopy of LmD PMBs fabricated under various homogenization speeds under 30% dextrose. Scale bars, 10 µm. g, The gas carrying capacity of LmD PMB foams fabricated under various speeds. Data are means ± s.d., biological replicates. h, The aseptically fabricated LmD PMBs stored in D10 at room temperature did not change size distribution after 3 months. Source data

Fig. 2. LmD PMBs rapidly dissolve at physiologic pH.

a, DLS measurement of size following mixing of LmD PMBs in PBS solution for 2 min at varying pH. LmD PMBs fully dissolve above pH 5 and revert to their soluble components with a mean size <10 nm, similar to those of LmD solutions prepared from solid states. In contrast, the previous generation of IFNP MBs (made from more hydrophobic polymers) led to formation of much larger nanoparticles. b, Phantom sonography of aPMBs in aerated PBS shows the pH-dependent dissolution rate, evidenced by the disappearance of contrast intensity produced by the gas core under continuous ultrasound. The black dashed line indicates the timepoint when aPMBs were administered. In the absence of a gas sink, PMBs rapidly dissolved above pH 6 within seconds. Data (means ± s.e.m.) presented as change in contrast/bright area from baseline. (Of note, although the LmD shell is less soluble at pH 4.8 and 3.8, we noticed that the gas core of PMBs slowly becomes fluid filled as shown by the slow decrease in echo intensity, probably because the salts in PBS affected the swelling of the hydrogel-like shells.) c, Representative images from the phantom sonography study showing the dissolution profile of aPMBs at different timepoints at various pHs. BL, baseline. d,e, To account for any destructive effect that ultrasound itself has on PMBs, the experiment was repeated while only applying ultrasound at the expected dissolution time from b, showing similar dissolution times even without the application of continuous ultrasound (d). e, Data (means ± s.e.m.) presented as change in area of contrast/brightness from baseline, analysed using Student’s t-test. Cont., continuously applied ultrasound; Inter., intermittently applied ultrasound; NS, not significant. f,g, Dissolution of the shell and gas core was then studied using UV-Vis absorbance spectroscopy. Similar to the characterization using ultrasound (which detects only dissolution of the gas core), UV-Vis returns to baseline within seconds at pH above 6 (g), suggesting that the gas core has dissolved and that the shell has broken down into its constituent components. At more acidic pH, return of UV-Vis absorbance to baseline took 10 min or longer (f). Contrasting this with sonographic experiments (b) in which ultrasound scatter returned to baseline within 90–260 s, these findings suggest that following dissolution of the gas core, the remaining shell constituents take additional time to dissolve and revert to soluble components. All repeated measurements are biological replicates. Source data

Fig. 3. The pH-triggered, rapid dissolution mechanism of LmD PMBs avoids vascular obstruction and haemodynamic instability.

a, The rapid dissolution of PMBs is critical to their in vivo safety. Previously described gas carriers did not dissolve rapidly or coalesced following injection, leading to pulmonary vascular obstruction. b, In contrast, PMBs dissolve so rapidly following injection that they exist mainly as soluble molecular constituents by their first contact with the pulmonary circulation. c–f, Haemodynamic safety study. n = 4 for treatment, n = 5 for D10 control. Continuous measurements collected as biological replicates. PVR (c) was not significantly different from baseline following 5 injections of 5 ml 80% aPMBs (50% vol air/vol foam) over 1 min each. Data (means ± s.e.m.) presented as mean percent change from baseline. Mean PVR during each experimental period (d) did not change during or following infusions of aPMBs during either the injection or observation period. Data are means ± s.e.m., P values calculated using two-way ANOVA. MABP (e) increased during injection of PMBs and returned to baseline thereafter. Data (means ± s.e.m.) presented as mean percent change from baseline. Mean MABP during each experimental period (f) did not change during or following infusions of aPMBs during either the injection or observation period. Data are means ± s.e.m., P values calculated using two-way ANOVA. g–j, Representative transthoracic echocardiography images during infusion of PMBs and LOMs through the left parasternal window for a four-chamber view. g, Control animal, four-chamber view with right atrium (RA), left atrium (LA), right ventricle (RV) and left ventricle (LV). h, High opacification in both left and right ventricles of an animal injected with LOMs at a rate of 4 ml kg⁻¹ min⁻¹. i, Moderate opacification in the right ventricle of an animal injected with IVO₂ of oPMBs at a rate of 4 ml kg⁻¹ min⁻¹, with no visible signal noted in the left chambers. j, Moderate opacification in the right ventricle of an animal injected with intravenous aPMBs (IVAir) at a rate of 4 ml kg⁻¹ min⁻¹, with no visible signal noted in the left chambers. k,l, Percentage of opacified areas in the right (k) and left (l) ventricles relative to respective ventricle area was quantified during administration of IVO₂, IVAir and LOM at the flow rates shown, each during a 1-min infusion. Data are means ± s.e.m., comparisons using one-way ANOVA with Tukey’s multiple comparisons test. Source data

Fig. 4. IVO₂ via LmD oPMBs improves survival and meaningful outcomes in a swine model of severe hypoxaemic respiratory failure.

a, Study timeline. IVO₂ treatment (oPMBs) (n = 8) or control (D10) solution (n = 10) was administered at minutes 6, 8 and 10. ETT, endotracheal tube. b, IVO₂ rapidly and significantly increased arterial oxyhaemoglobin saturation (SaO₂) during the asphyxial period (grey shading). c, Arterial carbon dioxide tension (pCO₂) during asphyxia was significantly higher in IVO₂-treated swine. b,c, Groups compared using two-way ANOVA with Sidak’s multiple comparisons, with only significant P values shown. d, IVO₂ treatment increased the fraction of animals free of cardiac arrest and CPR during and post asphyxia. Treatment period, blue shading. e,f, IVO₂ treatment significantly decreased CPR time (e) and the required dose of epinephrine (f) used during resuscitation. Groups compared using Student’s t-test. g, IVO₂ treatment improved MABP during resuscitation. Groups compared using two-way ANOVA with Sidak’s multiple comparisons, with only significant P values shown. h, IVO₂ treatment significantly improved ROSC at 30 min (log-rank test P = 0.003; Gehan–Breslow–Wilcoxon test, P = 0.003) and 84-h survival (log-rank test P = 0.013; Gehan–Breslow–Wilcoxon test, P = 0.007). i, IVO₂ treatment significantly improved the SNDS in surviving swine. ACA, asphyxial cardiac arrest; POD, post-operative day. Groups compared using two-way ANOVA with Sidak’s multiple comparisons. j, GFAP, a marker of astrocyte injury, was significantly elevated at day 3 in the control group, whereas no difference from baseline was observed in the treatment group. Groups compared using two-way ANOVA with Sidak’s multiple comparisons test. k, Representative weighted T2 MR image at 84 h post asphyxia depicts total grey matter and white matter diffusion restriction (supratentorial/infratentorial) with T2 prolongation throughout the cortex. l, Representative image of an IVO₂-treated swine reveals faint T2 prolongation in the basal ganglia. m,n, Three-dimensional representation of the median injury from brain MRI in control (m) versus treated (n) animals. Areas of enhancement on axial and coronal T2 and diffusion coefficient images were manually processed on a voxel-per-voxel basis. o, Volume of abnormal enhancement on T2 and diffusion coefficient images was significantly lower in IVO₂-treated swine than in surviving control swine. Comparison using Student’s t-test. p, Representative gross photos from the control group showed swollen, friable brain tissue with severe maceration of the ventral surface, and their pathological sections showed an overall dusky colour, blurring of the grey–white junction and intraventricular discoloration; in contrast, representative photos from the IVO₂-treated group revealed well-preserved brain tissue with few apparent abnormalities. q, Histologic injury score was statistically significantly lower in the basal ganglia structures in IVO₂-treated swine than in controls. Scoring: 0, no damage; 1, rare hypereosinophilic neurons; 2, clusters of hypereosinophilic neurons; 3, >50% of neurons are hypereosinophilic; 4, >90% of neurons are hypereosinophilic; 5, cavitated infarction. Groups compared using two-way ANOVA with Sidak’s multiple comparisons test. DG, dentate gyrus; PL, pyramidal layer. r,s, BUN (r) and creatinine (s) were significantly higher in the control group on day 3 than in IVO₂-treated swine. Note that i–s reflect data collected only in surviving swine, which omits 7 of the 10 swine in the control group that did not survive. For all figures, data are means ± s.d.; measurements are biological replicates. Source data

Fig. 5. Biodistribution and safety study of LmD PMBs in rodents.

a–d, Representative PET/CT images of rats receiving ⁸⁹Zr-labelled polymers over time. a, Immediately after infusion, radioactivity was observed in the upper abdomen, liver and kidneys, with a significant portion being excreted via bladder. b, Continuous excretion via urine and hepatic clearance at 24 h. c, Bowel excretion continues via hepatic clearance at 48 h, with maximum accumulation in stools. d, Low radioactivity level on day 7, with continuous excretion via faeces. e, Biodistribution on day 7, with the residual polymer presented as injected activity per gram of a particular organ (n = 3; measurements are biological replicates). f–o, Major clinical markers for organ injury and toxicity were normal in animals receiving LmD PMBs compared with control group (n = 3–6 per group) in rodent safety study. Dosage 1, 32 ml of 70% oPMBs (40% v/v oxygen) per kg, equivalent to the efficacy dose, with endpoints at three timepoints (4, 7 and 14 days), and control groups receiving equal volume of D10. Dosage ×2, doubling of dosage 1, administered 30 min apart, a total of 64 ml of 70% oPMBs (40% v/v oxygen) per kg, 14-day single timepoint, with control group receiving equal volume of D10. Data are means ± s.d., compared using multiple Mann–Whitney tests. Weight gain (f). Liver function tests: (g) alkaline phosphatase, (h) alanine transaminase, (i) amylase. Renal function tests: (j) BUN, (k) creatinine. (l) Lactate. Complete blood count: (m) haemoglobins, (n) white blood cells, (o) platelets. All values of animals receiving PMBs of both dosages at all timepoints were within normal ranges and showed no significant difference from control groups. p–w, Coagulation analysis by ROTEM in external coagulation pathway (known as EXTEM) and intrinsic coagulation pathway (known as INTEM) showed that infusion of PMBs did not adversely affect clotting: (p,t) clotting time, (q,u) clot formation time, (r,v) maximum clot firmness, (s,w) maximum lysis. Data are means ± s.d., compared using multiple Mann–Whitney tests with q value shown. All measurements are biological replicates. Source data

Extended Data Fig. 1. Haemodynamic response in acute rodent safety study.

Changes in cardiac index (CI) (a) and the mean CI values during each experimental period (b), changes in stroke volume (SV) (c) and the mean SV values during each experimental period (d), changes in left-ventricular end-diastolic pressure (LVEDP) (e) and the mean LVEDP values during each experimental period (f), as well as changes in mean pulmonary arterial pressure (MPAP) (g) and the mean MPAP values (h) during each experimental period were presented as mean and SEM, and analyzed by two-way ANOVA. Source data

Extended Data Fig. 2. Changes in select clinical markers in the asphyxia efficacy study.

The values for various clinical biomarkers including alanine transaminase (a), alkaline phosphatase (b), total bilirubin (c), hemoglobin (d), white blood cells (e) and platelets (f) were among normal ranges for IVO2-treated animals and were not statistically different from the surviving animals from the control group. Data presented as mean ± SD, statistical analysis by two-way ANOVA. Source data