Phosgene-induced acute lung injury (ALI): differences from chlorine-induced ALI and attempts to translate toxicology to clinical medicine

Phosgene (carbonyl dichloride) gas is an indispensable chemical intermediate used in numerous industrial processes at a global annual production scale of approximately 15 million metric tons. In contrast to most irritant gases, phosgene is poorly soluble in water, reacts by acylation with nucleophilic moieties, and causes pulmonary edema with a typical clinical “latency” or, more appropriately described, an clinically “occult” period. At such production scales, phosgene is commonly produced on demand and solvent-free by high-yield catalyzed gas-phase reactions of carbon monoxide and chlorine. Phosgene is an essential building block for the synthesis of isocyanates, which serve as monomers for paints, coatings, insulating foams, and special-use chemicals, to mention but a few.

Phosgene and chlorine are most notorious as early chemical warfare agents, first used militarily in 1915 during World War I (WWI). For operational purposes, they were denoted as either G52 (during WWI) or CG (WWI and later) [1‐8]. A comprehensive overview of warfare agents, including phosgene, has been published [9‐13]. Within military communities [both United States (US) and non-US], phosgene has been the subject of many toxicological studies, human toxicity estimates (ranging from threshold mild effects to lethality), and other general reviews and commentaries [12‐15].

Therapeutic approaches have traditionally been based on hypothetical pathways elucidated through in vivo research on rodents. Putative mechanisms of phosgene-induced ALI range from direct interaction and deterioration of lung surfactants and associated changes in lung mechanics [16, 17] to free radical attacks on neuronal, endothelial and epithelial cells, resulting in tissue destruction and mediator release [18, 19]. Canine, ovine, and porcine models have been used when studying the cardiopulmonary and hematological effects of methods used in humans. These models included protective ventilation in terminally anesthetized animals and pharmacological interventions such as steroids and bronchodilators [18‐26]. Many of the drugs examined in these studies were shown to be ineffective in follow-up proof-of-principle studies. Concurrent with the change in paradigms for treating ALI/ARDS (acute lung injury/acute respiratory distress syndrome), the emphasis of more recent research has shifted from treating to preventing acute lung injury using triage-based preemptive, personalized ventilator strategies applied to maintain normal lung function in patients at high risk [27‐32]. Supportive treatment commonly includes increased fractional concentrations of FiO₂, the use of PEEP, and physical rest. However, preventive treatment of phosgene-induced ALI with ‘injury-adjusted’ partial pressures of O₂ and PEEP settings has not been systematically investigated. Countermeasures were further complicated because ‘trial and error’ types of treatment strategies were already applied in the absence of indications and robust diagnostic tools that would have given prognostic guidance to clinicians.

Mechanism-based causal countermeasures require an in-depth understanding of the adverse outcome pathway (AOP), including its concentration × time relationship, initiating and amplifying the respective life-threatening condition. While past approaches focused on pharmacological interventions to mitigate phosgene-induced pulmonary edema, the focus of the research described in this paper was to better characterize the onset and interrelationships of early types of physiological dysregulation as initiating events causing progressively developing pulmonary edema. Unlike other, more water-soluble irritant gases, such as HCl or chlorine, potentially lethal exposure to phosgene may not subjectively perceived as such. Thus, clinically occult lung edema might occur within the asymptomatic period of patients, which then changes precipitously with time after exposure, leading to respiratory failure and death. The odor threshold for phosgene is significantly higher than current inhalation exposure limits [5, 33‐35]. Thus, odor or sensory irritation provides insufficient warning or clinical evidence of hazardous exposure doses.

Despite overwhelming evidence from both toxicological and medical research, even recently published papers often begin with the following statement: “Owing to its poor water solubility, one of the hallmarks of phosgene toxicity is an unpredictable asymptomatic latent phase before the development of noncardiogenic pulmonary edema”. Notably, the “latent” or, more appropriately phrased, clinically “occult” period of phosgene poisoning is the largely asymptomatic interval between exposure and the onset of edema by conventional methods. This definition is a fallacy since the incipient anatomic and pathophysiologic lung injury occurs with exposure and steadily progresses until sufficiently severe to become phenotypically detectable. Its occurrence follows a typical reciprocal inhaled concentration x time relationship. At exposure intensities within the range of 300–500 ppm × min, pulmonary edema occurs few hours post-exposure, followed by lethality ≈12–24 h later. At much higher exposure intensities, this period may becomes markedly shorter [35, 36]. Delayed mortality was also observed in experimental models of phosgene examined 80 years ago [24]; however, it was absent in more recent studies [37, 38]. Accounting for the fact that the more recent industrial production of phosgene is by catalytic reaction of the high-purity gases anhydrous chlorine and carbon monoxide, the presence of irritant impurities causing airway injury can be ruled out. The largest-scale human exposures to chlorine occurred during World War I, when the gas was used as a chemical weapon. Chlorine-induced oxidative injury and normal repair of the respiratory epithelium of the airways was critical to preventing the long-term pulmonary pathology that can occur following acute injury [39, 40].

This review discusses the most salient findings from toxicological and pharmacological research on rats and dogs over a period of one decade [17, 20, 37, 38, 41‐50]. The objective of this project was not only to develop inhalation exposure systems to expose rats and dogs to phosgene under highly controlled conditions and similar modes of exposure [20, 33, 37, 38, 49, 51] but also to study the early physiological events involved in phosgene-induced ALI, including options for causal and preventive treatment strategies. This process included the identification of early biomarkers of pulmonary injury that predict life-threatening pulmonary edema. Although most of the mechanistic endpoints were invasive in nature, emphasis was also directed toward non-invasive diagnostic methods that are translatable to clinical practice. One of the ancillary objectives of this work was to search for diagnostic tools to provide integrated information as to how triage and countermeasures could be structured for patients exposed to mixtures of phosgene and chlorine, a precursor of phosgene. To achieve these objectives, methods used in toxicology must be translatable to those used in humans.

Rats were exposed to phosgene (COCl₂) using a directed-flow nose-only inhalation principle [33, 37, 51]. Current testing guidelines give preference to this mode of inhalation exposure [52]. Certified gas standards with specified stability in synthetic air were used in all studies. The gas was contained in 10 L cylinders @150 bar. The volume-to-mass conversion factor for phosgene is 1 ppm = 4.1 mg/m³. Throughout all studies, the exposure period was 30 min. Air flow, temperature, and humidity measurements in the inhalation chamber utilized a computerized data acquisition and control system. The exposure conditions were adjusted to maintain an airflow rate of 0.75 L/min per rat, which is threefold higher than the respiratory minute ventilation of the rat. Under the given conditions, inhalation chamber state–state was attained within the first minute of exposure. The analytical concentrations from the inhalation chamber were in agreement with the nominally calculated concentrations, which were targeted at 30–35 mg phosgene/m³ (≈1000 mg/m³ × min or ≈250 ppm × min). In studies aimed at toxicological endpoints, the characterization of test atmospheres utilized OSHA method no. 61 (http://​www.​osha-slc.​gov/​dts/​sltc/​methods/​organic/​org061/​org061.​html) using gas bubblers filled with a toluenic solution of the trapping agent 2-hydroxymethyl-piperidine (2-HMP). The resultant analyte was then analyzed by gas chromatography. For mechanistic and intervention studies, actual concentrations were determined in real time using a calibrated Gasmet Dx-4000 FT-IR (Fourier transform infrared spectroscopy) gas analysis system (for details see http://​www.​gasmet.​com/​images/​tiedostot/​product-downloads/​Gasmet_​DX4000_​Technical_​Data_​(v1.​6).​pdf). The spatial homogeneity and temporal stability of phosgene in exposure atmospheres were controlled in real time [37].

Rats exposed first to phosgene and then to the aerosolized drug aminoguanidine were exposed nose-only, similar to phosgene [44], or in a small whole-body inhalation chamber with dynamic air flow and aerosol generation at targeted and analytically verified concentrations of ≈300 mg drug/m³. The comparison of nose-only and whole-body exposed rats served the purpose of judging the impact of “psychological immobilization stress” and associated cardiovascular stimulation due to restraint relative to non-immobilized, whole-body-exposed rats. Under such exposure conditions, the inhaled dose rate of drug is equivalent to ≈16 mg/kg-rat/hour.

Rats were anesthetized by intraperitoneal injection of pentobarbitone, and blood was collected from the left ventricle at sacrifice. Animals were exsanguinated by severing the abdominal aorta. Then, the excised lungs were weighed, and bronchoalveolar lavage fluid (BALF) was obtained as detailed elsewhere [38, 42].

Details of the head-only chamber used for dog inhalation studies have been published elsewhere [17, 20]. This mode of exposure to phosgene differed from those of other authors using larger animals. For reference, the reader is advised to consult more detailed reviews and papers on larger animals used for studies with phosgene [9, 21, 22, 24, 53]. Larger inhalation chambers may be useful to accommodate larger animals or larger numbers of small animals. For technical reasons and the difficulty of generating homogeneous exposure atmospheres at short exposure durations, a more human-like exposure mode and regimen may jeopardize the outcome of the study due to technical shortcomings. Especially for pharmaceutical countermeasures delivered by the inhalation route, particular attention must be paid to maintaining similarities of the dosing regimen used in the bioassay with that used in humans. Otherwise, meaningful interspecies extrapolations and dosimetric adjustments are hampered. The endotracheal administration of phosgene and inhalation drugs may overcome some of these difficulties; however, due to the numerous manipulations required, this may cause additional uncertainties concerning the inhaled dose. Compared to small animals, dogs and pigs offer the advantage that these species have also been used in pre-clinical studies of inhalation pharmaceuticals. Their breathing physiology is closer to that of humans than that of rodents. The size and anatomy of their lungs, including the large tracheobronchial tree and vascular architecture, make it possible to use the same equipment as used in intensive care units (ICUs). Thus, when making judgements as to the extent to which a small or large animal model delivers the most significant information for human risk assessment, numerous methodological and species-specific factors must be considered. These factors include that the exposure and treatment of larger animals using endotracheal tubes and terminal anesthesia may not only complicate translation dosimetry but may also affect reflex-mediated responses to exposure and injury.

The incipient pathogenesis of phosgene-induced ALI/ARDS starts with loss of surfactant function, which is caused by direct chemical reactions of surfactant with phosgene [33, 45]. This loss then may lead to a heterogeneous collapse of alveoli and ensuing changes in lung mechanics. Although additional mechanisms cannot entirely be excluded, the alteration of the Starling fluid flux equation, e.g., due to increased interstitial pressure, further destabilizes the alveoli. Accordingly, the time for treatment is before, not after, edema has appeared, as previously concluded by Coman et al. [24] 80 years ago.

Such a series of events calls for preventive, rather than therapeutic, modes of treatment. However, any proposed therapies targeting prevention or early treatment of lung injury prior to respiratory failure require specialized diagnostic tools to identify early at-risk patients who will actually develop ALI/ARDS. Progress in specific treatments for ALI/ARDS beyond the lung-protective strategies of mechanical ventilation and conservative fluid management needs to be realized [31, 54]. Hence, the currently instituted reactive rather than proactive approaches regarding lung protection should be refocused on preventing the progression of worsening lung injury with time elapsed post-exposure rather than attempting to treat respiratory failure that becomes increasingly refractory to any type of treatment.

Evidence collected from studies in small animals supports the conclusion that phosgene-induced pulmonary edema was initiated and propagated by events that affect the lungs’ overall control of lung mechanics and was indirectly initiated by tissue congestion and decreased vascular compliance/airway patency. Secondary changes in cardiovascular control and hemodynamics add another dimension of complexity. These conclusions from small-animal models were further substantiated in exploratory proof-of-concept studies in dogs [17] and pigs [22] with protective positive PEEP ventilation. This mode of ventilation strategy was demonstrated to improve survival in patients with ARDS [31, 120, 121], especially when begun early after the initiating encounter. Dogs were exposed at clearly lethal levels (135 mg/m³ × 25 min, equivalent to 3350 mg/m³ × min or 820 ppm × min). Shortly after exposure, the dogs were anesthetized, intubated and then subjected to PEEP (V_T = 10–12 mL/kg body weight, 40 breaths/min; FiO₂: 0.21) at 0, 4, or 12 cm H₂O over a post-exposure period of 8 h (one dog per setting). The lung edema was markedly alleviated at 4 cm H₂O, but not at 0 cm H₂O of PEEP. Microscopy confirmed reduced hemorrhage, neutrophilic infiltration, and intra-alveolar edema. Thus, the time-dependent progression into life-threatening pulmonary edema can be effectively suppressed by protective, low-pressure PEEP when implemented sufficiently early after exposure to phosgene [17]. Anesthetized pigs were instrumented and exposed to phosgene (concentration × time (Cxt), 2350 mg × min/m³ or 573 ppm × min) and then ventilated with intermittent PEEP (V_T = 10 mL/kg; PEEP 3 cm H₂O; 20 breaths/min; FiO₂: 0.24), monitored for 6 h after exposure, and then randomized into treatment groups: V_T at 8 or 6 mL/kg; PEEP 8 cm H₂O; 20 or 25 breaths/min; FiO₂: 0.4. This study aimed to examine the pathophysiological changes observed with low-V_T-protective ventilation strategies compared to conventional ventilation. Pathophysiological parameters were measured for up to 24 h. The results showed improved oxygenation and decreased shunt and mortality, with all the animals surviving to 24 h compared to only three of the conventional ventilation animals. Microscopy confirmed reduced hemorrhage, neutrophilic infiltration, and intra-alveolar edema [22].

From phosgene inhalation studies in dogs at 1880 ppm × min (7708 mg/m³ × min), it was concluded that, under the given experimental conditions, immediate therapy with O₂ is vital and FiO₂ of 0.4–0.5 is sufficient [25]. Timely correction with NaHCO₃ infusion was recommended for base deficit; however, the associated negative consequences must be thoughtfully considered (for details, see ‘permissive hypercapnia’ below). There was no apparent benefit from cortisone, theophylline, PGE₁ or atropine. Jugg and coworkers published a more comprehensive comparison of large animal models using therapeutic approaches [9, 25, 26].

As exemplified for phosgene, the most critical phase for prognostic triage and successful preventive treatment is the asymptomatic, rather than the symptomatic phase. The comparison of the predominantly airway irritant chlorine with the alveolar irritant phosgene demonstrated appreciable differences in injury patterns. This result justifies not only different countermeasures but also the appropriate diagnostic tools to guide optimal treatment. Elevated concentrations of fibrin and hemoglobin in blood as well as CO₂ and NO measured in expired gas were shown to be practicable and sensitive biomarkers of site-specific injuries within the respiratory tract. Re-triage by time-course measurements of CO₂ and NO in exhaled breath using protocols distinguishing the fraction of breath from the airways and alveoli may increase the diagnostic power of this assay [92, 122]. Bedside quantification of dead space could be used to titrate countermeasures at the asymptomatic stage of injury. In cases of exposure to mixtures of irritant gases, late complications cannot be entirely excluded. Therefore, prior to discharge of patients or before changing treatment strategies from anti-edema to anti-inflammatory, these readily available analyses may deliver important information to clinicians regarding which course to take. These methods appear to be easy to manage and suitable for both triage and re-triage.

Commonly, practitioners and clinicians alike are guided by the symptoms elaborated in putatively exposed subjects for the identification of high-risk patients. Most often, treatment follows reactive rather than proactive approaches, with an emphasis on treating rather than preventing the progression of worsening lung injury. Frequently, countermeasures appear to focus on PaO₂ or saturation [32] to determine whether treatment strategies are effective. However, PaO₂ may not be an accurate surrogate of alveolar stability; therefore, reliance on PaO₂ as a marker of lung function presumes that there is no self-perpetuating and progressing occult ALI leading to alveolar instability and eventually lethal edema. As shown by the preventive PEEP applied to dogs and pigs, there is evidence that oxygenation as a method to optimize PEEP is not necessarily congruent with the PEEP levels required to maintain an open and stable lung [31, 32]. Thus, optimal PEEP might not be personalized to the lung pathology of an individual patient using oxygenation as the physiologic feedback system. Likewise, non-personalized, unreasonably high PEEP pressures may block lymph drainage. Ideally, titration of PEEP by volumetric capnometry at low V_T appears to be the most promising strategy [92, 123]. Volumetric capnometry was shown to be helpful for monitoring the response to titration of PEEP, indicating that the optimal PEEP should provide not only the best oxygenation and compliance but also the lowest V_D while maintaining the V_T below a level that over-distends lung units and aggravates V_D and lung injury [92]. Thus, the improvements in oxygenation and lung mechanics after an alveolar recruitment maneuver appear to be better preserved by using injury-adapted PEEP than by any ‘one size fits all’ standardized approach. Notably, protective lung ventilation strategies commonly involve hypercapnia. Thus, permissive hypercapnia has become a central component of protective lung ventilatory strategies [121, 124‐128].

Due to the complex interactions among (patho)physiological events, it seems unrealistic to assume that any monocausal, drug-related treatment regimen will be identified in the near future to mitigate the particular type of ALI attributable to inhaled phosgene gas. This conclusion matches those of other authors [120, 129‐131]. Collectively, the wealth of published evidence supports the conclusion that, if the acute stage of pulmonary edema with its attendant anoxic anoxia is survived, circulatory failure may become a more important factor in the ultimate outcome [65]. Likewise, a countermeasure identified to be efficacious for a non-water-soluble gas, such as phosgene, may not necessarily be the best countermeasure for a highly water-soluble airway and alveolar irritant gas, such as chlorine, and vice versa.

Multiple approaches for drug-related interventions, most of them anti-inflammatory and sympathomimetic, have been examined [9, 19, 22, 23, 25, 26, 55, 96, 132, 133]; however, none of these drugs have found their way into the clinic. To the contrary, as could be expected for phosgene, anti-inflammatory treatment with steroidal or non-steroidal drugs was either ineffective or even aggravated phosgene-induced ALI [21, 22, 44, 46]. More recent exploratory preclinical investigations have identified TRP inhibitors, NOS inhibitors, and statins as novel pharmaceutical approaches that prevent ALI; these drugs merit being studied in greater detail in the future [19, 31, 83, 84, 96, 134].

As exemplified by many experimental studies in rats, an excess of water in the lung is not a consequence of too much water in the body; rather, it is a consequence of dysfunctional cardiovascular control to prevent excess fluid from accumulating in the septal interstitium and subsequent alveolar flooding. Hence, any use of diuretics may further aggravate the phosgene-induced hemoconcentration, rather than having any beneficial effect on the increasing pulmonary edema. Equally deleterious therapeutic results were obtained with bleeding or venesection (phlebotomy) and argue against these therapeutic options [65]. Notably, despite its vulnerable blood-air barrier, the lung is relatively resistant to the onset of pulmonary edema. This resistance is ascribed to several safety factors, which include increased lymph flow to drain fluids away from the lung and decreased interstitial oncotic pressure and interstitial compliance. These safety mechanisms are quite effective as long as surfactant prevents alveolar collapse [135‐138]. The supine position increases gravity-related hydrostatic pressure and lung edema, which supports the prone positioning of patients [31]. The symptomatic treatment of hemoconcentration by non-conservative fluid resuscitation may change a non-lethal to a lethal lung edema, as this surplus fluid was shown to settle in the lung as edema [54, 139], as shown in previous dog inhalation studies with phosgene [65, 139‐141]. Hence, fluid resuscitation should be handled most conservatively [115, 140]. The use of nebulized sympathomimetics may further contribute to reflexively induced changes in cardiac output and pulmonary hydrostatic pressure. Nebulized salbutamol treatment following phosgene-induced ALI did not improve survival and worsened various physiological parameters, including arterial oxygen partial pressure and shunt fraction. Anti-inflammatory corticoids have shown little benefit in patients with this type of cardiogenic lung edema in the absence of an inflammatory etiopathology. In summary, most of these types of “symptomatic treatments” might transform phosgene-induced ALI into iatrogenic ALI, rather paving the road to recovery [21, 25].

Data from multiple animal species and mechanistic studies have coherently demonstrated that phosgene-induced ALI is unique compared to ALI induced by other, more water-soluble irritant gases. Phosgene-induced ALI is initiated with exposure and remains occult for hours post-exposure, depending on the dose inhaled. During this asymptomatic period, a range of reflex-related cardiovascular responses appears to be involved in triggering progressive changes in cardiopulmonary and hemodynamic homeostasis. This imbalance of neurophysiological control may progressively shift fluid from the peripheral to the pulmonary circulation, leading to potentially lethal alveolar edema. Any proposed therapies targeting the prevention or early treatment of lung injury prior to respiratory failure require triage to identify patients at high risk, as resources are limited. CO₂ and NO in exhaled breath were shown to be prognostic for edema occurring hours later.

Most importantly, clinicians should refrain from non-rationalized or common symptomatic treatments that could accelerate the progression of ALI. Preventive and personalized treatment strategies of mechanical ventilation with feedback loops focusing on lung function and conservative fluid management should be given preference.

In summary, current knowledge about the sequelae of phosgene-induced ALI has clearly positioned the field to undertake steps toward preventive or causal treatment, rather than mere symptomatic treatment; however, much work and communication remain necessary to make therapies effective, practical, and safe for asymptomatic subjects. The objective of the course taken in this paper was to challenge the often-exercised ‘trial-and-error’ type of symptomatic treatment in the absence of any mechanistic understanding.