Acute Severe Anaphylaxis in Nepali Patients with Neurotoxic Snakebite Envenoming Treated with the VINS Polyvalent Antivenom

Diagnosing and treating acute severe and recurrent antivenom-related anaphylaxis (ARA) is challenging and reported experience is limited. Herein, we describe our experience of severe ARA in patients with neurotoxic snakebite envenoming in Nepal. Patients were enrolled in a randomised, double-blind trial of high vs. low dose antivenom, given by intravenous (IV) push, followed by infusion. Training in ARA management emphasised stopping antivenom and giving intramuscular (IM) adrenaline, IV hydrocortisone, and IV chlorphenamine at the first sign/s of ARA. Later, IV adrenaline infusion (IVAI) was introduced for patients with antecedent ARA requiring additional antivenom infusions. Preantivenom subcutaneous adrenaline (SCAd) was introduced in the second study year (2012). Of 155 envenomed patients who received ≥ 1 antivenom dose, 13 (8.4%), three children (aged 5−11 years) and 10 adults (18−52 years), developed clinical features consistent with severe ARA, including six with overlapping signs of severe envenoming. Four and nine patients received low and high dose antivenom, respectively, and six had received SCAd. Principal signs of severe ARA were dyspnoea alone (n=5 patients), dyspnoea with wheezing (n=3), hypotension (n=3), shock (n=3), restlessness (n=3), respiratory/cardiorespiratory arrest (n=7), and early (n=1) and late laryngeal oedema (n=1); rash was associated with severe ARA in 10 patients. Four patients were given IVAI. Of the 8 (5.1%) deaths, three occurred in transit to hospital. Severe ARA was common and recurrent and had overlapping signs with severe neurotoxic envenoming. Optimising the management of ARA at different healthy system levels needs more research. This trial is registered with NCT01284855.


Introduction
Snake antivenoms are the only specific treatments for snakebite envenoming; they save lives but are associated with acute pyrogenic reactions, due to endotoxin contamination whilst in production, and acute anaphylaxis [1].
The mechanisms underlying antivenom-related anaphylaxis (ARA) are uncertain and are probably a combination of complement activation, a type I hypersensitivity reaction, non-allergen-specific activation of mast cells triggered by the antivenom impurities and immune priming due to the venom itself [2,3]; indeed, anaphylaxis has been reported in snake 2 Journal of Tropical Medicine handlers after >1 envenoming [4]. The reported rates of ARA range from <5 up to <90% depending on the quality of the antivenom used and the diligence in recording events [5][6][7][8][9][10][11][12]. One report from Thailand suggested higher rates in cobrabitten (∼12%) vs. viper-bitten (∼2%) patients [13] whereas a larger Indian study found the opposite relationship [14].
Intramuscular adrenaline (IMAd) is the treatment of choice for acute ARA, but ARA may respond poorly (protracted anaphylaxis) or recur later (biphasic anaphylaxis) [22]. Giving additional doses of antivenom to treat progressive or recurrent envenoming is another important cause of recurrent ARA [9]. The optimal management of protracted or recurrent ARA is unclear. One recommendation is an intravenous adrenaline infusion (IVAI) [16] with tight control of the blood pressure to avoid adrenaline-induced toxicity like hypertension and intracranial haemorrhage [23].
Most venomous snake bites in Nepal are caused by spectacled and monocled cobras, Naja Naja and Naja kaouthia, and the common krait, Bungarus caeruleus [24]. Their neurotoxic venoms cause death by paralysing the bulbar and respiratory muscles. Indian-manufactured, equine-derived, polyvalent antivenom raised against the venoms of B. caeruleus, Daboia russelii (Russell's viper), Echis carinatus (saw-scaled viper), and N. naja is the only antivenom available in Nepal and, in one study, was associated with severe ARA in ∼22% of recipients [19].
We reported previously the efficacy and tolerability of high vs. low dose antivenom in neurotoxic envenomed patients [25]. Herein, we focus on severe ARA. n ∘ ACA-575-/067/068), the Nepal Health Research Council (approval n ∘ 986), and the Geneva University Hospitals Ethics Committee (approval n ∘ 08-192). All study participants gave written informed consent to participate in this study.

. . Study
. . Inclusion/Exclusion Criteria. Recruited patients were aged ≥ 5 years (y), with signs of neurotoxic envenoming, assessed using a neurotoxicity score (NS, Box 1), who/whose guardians gave signed, informed consent. Excluding criteria were (i) pregnant or breast feeding women, (ii) presentation > 24 h, (iii) patients needing immediate mechanical ventilation [respiratory distress, no gag reflex, paradoxical breathing and/or oxygen saturation (SpO 2 )] < 90% on room air], (iv) known allergy to horse proteins, (v) patients (a) with an underlying neuromuscular disease, (b) with a proven viper bite, and (c) who had received antivenom earlier at another health centre.
Patients were randomised to either the Nepali recommended low dose (LD) regimen: 6 vials, 2 by IV push over 5-10 m, then 4 infused over 4 h (1h & 3h infusions), or high dose (HD) antivenom: 10 vials by IV push over 5-10 m, then 8 vials infused over 1 h and 3 h normal saline infusion to maintain the blind. Persisting neurotoxic signs were treated with 4h infusions of 4 vials (LD arm) or saline (HD arm). Acute neurological deterioration (i.e., increased NS) was treated with 2 vials (LD arm) or 5 vials (HD arm) by IV push.
All patients received neostigmine to enhance neuromuscular transmission [27] and atropine to prevent the muscarinic effects of neostigmine (hypersalivation, colic, pulmonary oedema, and sinus bradycardia).
. . Patient Monitoring. Patients were monitored hourly until the NS became 0 (i.e., complete resolution of neurotoxic . . Reporting of Severe Anaphylaxis. Severe acute anaphylaxis meets the definition of a serious adverse event (SAE), i.e., life-threatening, prolongs inpatient stay, or results in death [28]. All clinical details were recorded on SAE forms by the site investigator under the supervision of the study site principal investigator (SKS) who was on call 24 h/day. All SAE reports were reviewed by EA, FC, & WRJT and then sent to the ethics committees of the BPKIHS and Geneva University Hospitals, the Nepal Health Research Council (NHRC) and the Drug Safety and Monitoring Board (DSMB) for comments.
For this report, the senior author reviewed again all SAE forms and associated communications to reconstruct the clinical picture from patient admission to discharge or death. Clinical events, defined as important symptoms or signs, were identified and the following details were recorded: (i) their timing, (ii) how often they occurred, (iii) what their causes were, (iv) how and at what times they were treated, and (v) outcomes. Data were entered into Microsoft Excel and analysed [descriptive analyses and Mann-Whitney U test (continuous data)] in Stata v14 (Stata Corporation, USA).
. . Prevention of Anaphylaxis. Premedication with SCAd was given in the second snakebite season in 2012 following the report by de Silva et al. [6]. The dose was 0.25 mL (patients aged ≥ 13 y), 0.2 mL (11-12 y) and 0.125 mL (5-10 y) of a 1:1,000 solution of adrenaline.
. . Definition and Treatment of Anaphylaxis. We used the definition recommended by the National Institute of Allergy and Infectious Disease and Food Allergy and Anaphylaxis Network, "anaphylaxis is a serious allergic reaction that is rapid in onset and may cause death" [29].
All clinical teams received prestudy and refresher trainings in the recognition and treatment of anaphylaxis, including (i) a list of key symptoms and signs (itching, urticaria, swollen lips or tongue, angioedema, dry cough, wheezing, stridor, hoarse voice, 'lump in throat' , nausea, vomiting, abdominal colic, diarrhoea, hypotension, and shock), (ii) ABC of resuscitation (airway: obstruction/compromise, breathing: tachypnoea, wheezing, and circulation: hypotension or shock +/-poor peripheral circulation), (iii) intubation and the use of an Ambu bag. We stressed the urgency of treating anaphylaxis when the first sign/s consistent with ARA appeared, irrespective of their severity [30], which consisted of IMAd, IV hydrocortisone, and IV chlorphenamine, following international guidelines [30][31][32]. Training did not include performing a cricothyrotomy or a tracheotomy.
For patients who had experienced ARA but needed more antivenom to treat envenoming, antivenom infusions were resumed when patients were either haemodynamically stable or when the treating physician thought the anaphylaxis was clinically resolved. A reappearance of ARA was treated as above, but we later replaced this practice with IVAIs to "cover" additional doses of antivenom infusion, following the regimen of Brown et al. [16]. IVAIs were adopted because (i) it was sometimes difficult to decide when ARA had resolved fully and so whether it was "safe" to restart the antivenom infusion and (ii) to prevent multiple injections of IMAd if physicians thought patients were having ongoing ARA during the antivenom infusions. Antivenom pushes (as above) for acute neurological deterioration were not covered by adrenaline.

Results
. . Demographic and ARA Summary Data. 155 patients with neurotoxic envenoming received at least one dose of antivenom and are included in this analysis, including one patient who later withdrew from the study ( Figure 1).
Over time, these 13 patients had 64 clinical events: two patients had 3 events, six had 4 events, two had 5 events, and three had 8 events (Table 2). Some events led rapidly to a cascade of additional events whilst others were separated by large time periods (Figure 2). There were 8 deaths (#s 6-13,  (Table 2). Rash, documented as urticaria, erythema, or 'itchy rashes' were the most common ARA manifestation and occurred during the first ARA episode in 10 patients (Tables 2, 3, and 4). Five patients had other concomitant signs of ARA whilst the other five all went on to develop other features of ARA over time. Severe ARA signs included dyspnoea alone or with wheezing, hypotension, shock, restlessness, and respiratory or cardiorespiratory arrest; two patients with dyspnoea manifested as gasping respirations (#8, #12). Early (#1) and late (#11) laryngeal oedema (E & LLO) were seen in two patients. The ELO occurred 35m into the resumption of his first antivenom infusion (interrupted because of an itchy rash), corresponding to 95 m after the IV push; clinical signs were noisy breathing, cough, and a fall in SpO 2 to 63%. The  patient with LLO developed a hoarse voice 8 h 35 m after his antivenom infusion had finished (11h after IV push). Stridor was not a feature of either E or LLO. In most ARA-diagnosed patients, antivenom was stopped and IMAd given almost immediately after the ARAs were recognised (Figure 1), followed by IV hydrocortisone and IV chlorphenamine. Nebulisers were given to one patient (#8) without IMAd who continued to receive antivenom without deterioration. Two patients (#s5 & 10) received IMAd after IV hydrocortisone at 6 and 15 m after the start of their first ARAs without deleterious effects. One and three patients were given IV adrenaline and SCAd, respectively, instead of IMAd. Four patients (#s1, 3, 10, 12) were admitted to Bharatpur Hospital and treated with five adrenaline infusions so antivenom could be restarted following antecedent reactions: (i) patient 1: initial antivenom induced rash had resolved, (ii) patient 3: it was unclear if the initial episode of severe ARA had resolved; the patient was ventilated and needed additional antivenom, (iii) patient 10: the patient had partial resolution of antivenom induced rash, and (iv) patient 12: despite two injections of IMAd, patient's rash remained unresolved; IVAI was used to complete the initial IV push of antivenom.

. . Deaths in Patients with Indeterminate Clinical Features.
Six patients (#s 6,7,8,9,12,13) had events with clinical features consistent with severe ARA and severe neurotoxic envenoming, although they were all considered to be antivenom-related at the time by the treating physicians (Table 3). Patient 7, whose presentation included dyspnoea, suffered a respiratory arrest and sinus bradycardia 15 m after starting antivenom; despite IMAd, intubation and oxygen, she had a cardiac arrest and died in the ambulance on the way to hospital. The other five patients had increasing neurotoxic scores (n=4), sudden cardiac arrest (n=1), dyspnoea without wheeze (n=1), gasping respiration (n=1), falling oxygen saturation (n=4), restlessness (n=2), hypotension (n=1), and sinus bradycardia (n=4, 5 episodes) that was followed rapidly by death. At the time of these clinical events, patient 9 was on an antivenom infusion, three had received IV push injections for neurological deterioration, and one was under observation in the ICU.
. . Other Deaths. One patient died of LLO (#11) and the other of anaphylactic shock (#10). Both patients died in the ambulance during hospital transfer.  Developed vomiting 15m after IV push followed by urticaria and dyspnoea with wheezing 5m later. Treated with salbutamol and ipratropium inhalations. No IM adrenaline given. Antivenom continued (NS=2). 40m later developed increased NS of 4 (IV AV push given) that, 20m later, increased to 6 associated with frothy secretions and muscle weakness (IV AV push given again).

1.9
Clinical picture dominated by rapidly progressive envenoming despite treatment with antivenom pushes. Patient did not receive IM adrenaline for initial episode of mild anaphylaxis nor adrenaline cover for the IV pushes, nor IMAd for possible ARA.

Female 5 y Y H
Presented with abdominal pain, vomiting, ptosis, tachypnoea (RR 40/m), tachycardia (120/m) and central cyanosis (SpO 2 60%) treated with oxygen (SpO 2 rose to 90%) before antivenom. 15m after AV push & while on AVI had a respiratory arrest associated with sinus bradycardia (50/m). Immediate intubation was followed by a cardiac arrest. Resuscitated with IV adrenaline & 300 mL IV fluid bolus; AVI continued. Pulse detected by oximeter but no recordable blood pressure. Decision made to transfer to hospital. Second cardiac arrest (exact time not noted) followed by death despite resuscitation in ambulance.

1.3
Clinical picture dominated by poor respiratory status before antivenom associated with NS score of 3. Respiratory arrest after antivenom followed by cardiac arrest. Given the rapidity of the events, ARA may have contributed to the clinical picture.

Discussion
Our study has documented a relatively high rate (∼8%) of VINS antivenom associated ARA that was clinically clearcut in seven patients but clouded in six mostly by a mix of dyspnoea, restlessness, and increasing neurological scores, all consistent with envenoming.
Acute anaphylaxis is a predictable toxicity of antivenom. The Indian-manufactured antivenoms, used commonly in south Asian countries, are associated with rates of severe ARA as high as ∼20 to 40% [3,6,10,19]. Recurrent ARA may recur acutely when additional doses of antivenom are given as infusions or IV pushes and "unexpectedly" as biphasic anaphylaxis hours after the apparent resolution of antecedent episode of ARA [9,21]. Moreover, patients may "unexpectedly" deteriorate clinically because of a recrudescence of their envenoming despite an initially good clinical response to antivenom. We faced such challenges in several patients.
The most difficult patients were those who developed restlessness and/or acute dyspnoea/gasping respiration with (e.g., #s13, 8) or without (e.g., #12, 9) an increase in the NS score and patients who deteriorated rapidly, culminating in a cardiac/respiratory arrest, whether they were treated for ARA (#6) or not (#7). Sinus bradycardia was an ominous sign that preceded cardiac arrest and death. With an acute increase in NS, clinicians may believe the acute clinical deterioration is exclusively due to worsening envenoming. However, they should consider whether an earlier antivenom infusion or IV push is contributing either acutely or as biphasic ARA and should have a low threshold for treating with IMAd.
To increase clinical awareness, we suggest several "danger" signs that require urgent assessment such as sinus bradycardia, relative bradycardia and hypotension [33], dyspnoea alone or with wheezing, hoarse voice, restlessness, and nonspecific acute clinical deterioration. These signs may occur in isolation or within a picture of worsening neurotoxicity that may be confusing and lead clinicians not to consider ARA.
Five patients developed dyspnoea without wheezing; two (one as gasping respiration) did not have a concomitant decline in NS whilst one gasping patient did. Although wheeze is a classic sign of acute anaphylaxis, dyspnoea alone is well documented and was more common than wheeze in two large clinical series. In one ARA study, ∼16% (32/198) patients developed hypoxia, 3% wheeze and hypoxia and 3.5% wheeze alone [10] and, in a study of all cause anaphylaxis (n=1,149), dyspnoea alone was present in 29% patients vs. 13% who had wheeze [34]. Acute dyspnoea without wheezing is also a well-characterised feature of progressive neurotoxic envenoming (leading to insufficient respiratory muscle strength and increasing NS) which requires rapid treatment with antivenom +/-assisted ventilation. Moreover, dyspnoea should always prompt the exclusion of mechanical blockage of the airway (e.g., a prolapsed tongue), and faulty intubation (e.g., into one bronchus) in ventilated patients.
Two patients developed laryngeal oedema neither of whom had stridor. The LLO patient had recovered from his earlier mild ARA and envenoming and his hoarse voice developed 8.5 h postantivenom infusion. It is unclear whether LO developed late and progressed rapidly (i.e., biphasic anaphylaxis) or insidiously to become clinically manifest only when the oedema was severe. He was unresponsive to adrenaline and hydrocortisone and the extent of the LO precluded endotracheal intubation. Without surgical equipment and skilled staff to relieve his upper airway obstruction (UAO), he died. The risk of laryngeal obstruction/oedema/stridor/throat tightness was 2% (1/48) in snakebite victims in Australia [15].
UAO is classed as both moderately severe [34] and lifethreatening [31]. In settings where intubation is problematic, mechanical ventilators are several hours away, and death during hospital transfer is well described [35], UAO is better classed as life-threatening. In retrospect, had we anticipated that LO might prevent intubation, we would have added training on cricothyrotomy/tracheotomy together with the necessary equipment.
During the study, our guidelines on treating recurrent ARA changed from treating each episode with IMAd to IVI to cover subsequent antivenom infusions [16]. Four patients, all enrolled at Bharatpur Hospital, received adrenaline infusions. Given the potential dangers of IVAI (e.g., cardiac arrhythmias, acute hypertension, and haemorrhagic stroke), the need for close monitoring, e.g., 3−5 m [36] and infusion titration, IVAIs are suitable only where there are well-trained ICU or emergency staff. More research is needed to optimise the management of patients with antecedent ARA who require repeated antivenom infusions or IV antivenom pushes for acute neurological deterioration.
We had a crude fatality rate (CFR) of ∼5%, higher than the 1.3% reported by de Silva et al. in their large (n=1007) trial that was conducted in hospitals with experience of snakebite management [6] but lower than the 13 [37] to 20%, with marked interclinic variation [38], reported previously in Nepal. Nevertheless, these data support a relationship of limited resources and skills and higher CFRs.
To conclude, we have highlighted the risks of severe ARA in neurotoxic envenomed patients and challenges in managing patients where resources are limited. Clinicians should have a low threshold for treating of ARA when the clinical picture is not clear-cut and ARA cannot be excluded. More research is needed to define optimal treatment strategies in different health settings, especially snakebite centres.

Data Availability
The data that are presented in this anaphylaxis paper are available in Microsoft XL. All data requests should be addressed to Professor F. Chappuis: Francois.Chappuis@hcuge.ch.