Efficacy of Indian polyvalent snake antivenoms against Sri Lankan snake venoms: lethality studies or clinically focussed in vitro studies

In vitro antivenom efficacy studies were compared to rodent lethality studies to test two Indian snake antivenoms (VINS and BHARAT) against four Sri Lankan snakes. In vitro efficacy was tested at venom concentrations consistent with human envenoming. Efficacy was compared statistically for one batch from each manufacturer where multiple vials were available. In binding studies EC50 for all VINS antivenoms were less than BHARAT for D. russelii [553 μg/mL vs. 1371 μg/mL;p = 0.016), but were greater for VINS antivenoms compared to BHARAT for N. naja [336 μg/mL vs. 70 μg/mL;p < 0.0001]. EC50 of both antivenoms was only slighty different for E. carinatus and B. caeruleus. For procoagulant activity neutralisation, the EC50 was lower for VINS compared to BHARAT - 60 μg/mL vs. 176 μg/mL (p < 0.0001) for Russell’s viper and 357 μg/mL vs. 6906μg/mL (p < 0.0001) for Saw-scaled viper. Only VINS antivenom neutralized in vitro neurotoxicity of krait venom. Both antivenoms partially neutralized cobra and didn’t neutralize Russell’s viper neurotoxicity. Lethality studies found no statistically significant difference in ED50 values between VINS and BHARAT antivenoms. VINS antivenoms appeared superior to BHARAT at concentrations equivalent to administering 10 vials antivenom, based on binding and neutralisation studies. Lethality studies were inconsistent suggesting rodent death may not measure relevant efficacy outcomes in humans.

Scientific RepoRts | 6:26778 | DOI: 10.1038/srep26778 Other in vitro and in vivo investigations are available to test the efficacy of antivenom to neutralise relevant pathophysiological effects induced by snake venoms and may provide more clinically useful information [13][14][15][16] . For snakes that cause coagulopathy in humans it would seem more appropriate to test the efficacy of antivenom to neutralise procoagulant venom activity in human plasma, rather than the efficacy of antivenom in preventing death in a rodent. The fields of venomics and antivenomics are also beginning to improve our understanding of the different composition of venoms and their interaction with antivenoms 17,18 . However, it will be important to ensure that antivenoms are tested in functional studies of clinically relevant venom effects.
Currently all antivenom used in Sri Lanka is supplied from India and there are limited data regarding the efficacy of Indian antivenoms against Sri Lankan snakes 19 . Concerns were raised about the effectiveness of Indian polyvalent antivenoms in Sri Lanka in 2012 and there were sufficient concerns about the efficacy of antivenom for the treatment of Russell's viper (Daboia russelii) envenoming that the Government requested independent testing of the antivenoms. Treating doctors had observed an increased number of hospital deaths and reactions to the antivenom.
The aim of this study was to investigate the efficacy of two different Indian polyvalent antivenoms comparing a range of in vitro studies to classical lethality studies. In doing so we aimed to test the ability of the antivenoms to neutralise 1) the procoagulant effect of two viper venoms-Russell's viper (D. russelii) and the saw-scaled viper (Echis carinatus); 2) neurotoxic effect of three snake venoms-common krait (Bungarus caeruleus), common cobra (Naja naja) and Russell's viper (D. russelii); comparing these to LD 50 /ED 50 studies.

Results
All in vitro experiments were done at venom concentrations measured in cases of human envenoming which was 1.7 μ g/ml for D. russelii, 0.5 μ g/ml for E. carinatus, 1.4 μ g/ml for N. naja and 0.15 μ g/ml for B. caeruleus. Multiple batches of antivenoms from both VINS Bioproducts Limited and BHARAT Serum and Vaccines Limited were tested and compared. To statistically compare the efficacy of the two antivenoms one batch from each manufacturer was compared where multiple vials were available-VINS (01011/10/11; 2010) and BHARAT (A5311006; 2011Jan).
Protein quantification. The median dry powder weight (mg) of antivenom, the median percentage of protein per mg of antivenom and the median dry powder weight (mg) of protein per antivenom vial were measured (Table 1). All batches of VINS antivenom had a higher protein concentration than the batch of BHARAT antivenom tested ( Table 1). The mean protein content of 10 vials of VINS antivenom from the same batch (01011/10-11) was 198 mg (Standard Deviation [SD]: 28 mg), which was significantly higher than the mean protein content of 10 vials of BHARAT antivenom from the same batch (A5311006), 98 mg SD: 29 mg; p < 0.0001; unpaired t-test).

Neurotoxicity
Neutralization of neurotoxicity of B. caeruleus, N. naja and D. russelii venoms. Addition of the recommended concentration of VINS and BHARAT antivenom did not prevent the inhibition of twitches induced by any of the venoms, so five times the recommended concentration of the antivenoms was used as per previous studies 20 . At this concentration VINS antivenom effectively prevented B. caeruleus venom induced twitch inhibition compared to no effect with BHARAT antivenom (Fig. 3B, Table 4). There was partial recovery of the response to exogenous ACh and CCh with VINS and BHARAT antivenoms following B. caeruleus venom (Fig. 4B). However, both VINS and BHARAT antivenoms had minimal effect in preventing N. naja venom induced twitch inhibition (Fig. 3C, Table 4), and no effect in preventing the partial inhibition of twitches induced by D. russelii venom (Fig. 3D, Table 4).     (Table 5). VINS antivenom appeared to be slightly more effective in the neutralization of lethality induced by D. russelli and B. caeruleus venoms compared to BHARAT antivenom, but this was not statistically significant (Table 5). VINS was almost twice as effective against N. naja venom compared to BHARAT, which was almost statistically significant. There was no numerical difference in the ED 50 values for E. carinatus venom

Discussion
This study has shown that VINS antivenom has a higher protein content and overall a greater in vitro efficacy against the medically important effects of most snake venoms in Sri Lanka, compared to BHARAT antivenom. VINS antivenom performed better against the clinically relevant effects of three snakes, being more efficacious   against the in vitro procoagulant activity of D. russelii and E. carinatus venoms, and the neurotoxic effects of B. caeruleus venom. A concerning finding was that the lethality and ED 50 studies did not reflect these findings. The ED 50 for the two antivenoms was not statistically significantly different for D. russelii, B. caeruleus and E. carinatus, in contrast to VINS being more efficacious against clinically relevant effects-coagulopathy and neurotoxicity. The study also found some variation in the protein content and efficacy of the different VINS antivenom batches from 2008 to 2012, with decreased protein content and reduced capacity of antivenoms binding venom antigens, for antivenoms manufactured in 2008 and 2010.
In a previous study of Russell's viper envenoming where antivenom concentrations were measured in 86 patients after administration of 10 vials of antivenom, the median antivenom concentration was 2.2 mg/ml a median of two hours after administration of 10 vials of antivenom 21 . Tables 2-4 show that this concentration is sufficient for complete binding of all four venom antigens, neutralisation of procoagulant effects and neutralisation of neurotoxicity by VINS antivenom. However, this was not true for BHARAT antivenom in which larger concentrations were required to bind the venom antigens and neutralise the procoagulant effects of E. carinatus venom. Both antivenoms required higher concentrations to neutralise the procoagulant effect of E. carinatus venom suggesting they are not as efficacious against this venom, although this concentration appears to be sufficient for VINS antivenom. The inferior efficacy of BHARAT antivenoms was consistent with the lower protein content in these antivenoms.
An unusual finding was that BHARAT antivenom had a significantly higher venom antigen binding capacity for N. naja venom than VINS antivenom. However, BHARAT antivenom was less effective in neutralising the neurotoxic effects of N. naja venom and had a higher ED 50 compared to VINS. This differed to the other venoms and suggests that BHARAT antivenom has a higher titre to cobra venom antigenic components that may not be toxic. One study found that BHARAT antivenom was ineffective against the neurotoxicity of Pakistani cobra (N. naja) and Pakistani krait (B. sindarus) at similar concentrations 22 . Unfortunately this study did not test other antivenoms and concluded that antivenom in general was ineffective against Pakistani snake neurotoxicity 22 . A more recent study found that higher titres of three different antivenoms (Indian, Thailand and Taiwan) were required to neutralise the neurotoxic effects 23 . Another study found that VINS antivenom bound more avidly to Indian compared to Sri Lankan cobra venom, also suggesting possible geographical variability in the venoms, particularly the neurotoxic activity 24 .
A limitation of the study was that only the most recent batches of BHARAT antivenoms were available to be tested and only multiple vials from one batch of each type of antivenom was available. However, the two recent batches of BHARAT were inferior to the recent batches of VINS tested against the important clinical effects of the medically important snakes in Sri Lanka and a direct comparison between batches with multiple vials found VINS to be statistically significantly more efficacious than BHARAT antivenoms. In addition, the variability within these two batches was much greater for BHARAT than for VINS. We have previously shown that expired antivenoms up to 10 years old, formulated as liquids, have lost minimal activity, even after extended periods at room temperature 25 . Antibodies in the solid form would be expected to be as least as stable.
The units of antivenom differed for the lethality studies compared to the in vitro studies because the lethality studies are dosed on mouse body weight. This is another reason that lethality studies are problematic because it is difficult to relate μ g/g body weight to the amount of venom in human bites. For the binding and in vitro studies we used concentrations measured in human snake envenoming cases. The difference in units did not affect the conclusions of the study because the relative efficacy of the two antivenoms differed in lethality studies compared to the in vitro studies. In addition, the ER 50 was also calculated to provide another comparison without units.
Another limitation was that the study did not test the efficacy of the antivenom against other known venom effects, such as myotoxicity or nephrotoxicity 26 . There are no well tested methods of assessing nephrotoxicity 26 . One study reports only minor changes in renal function in an in vivo murine model, evidenced by protein and erythrocytes in urine, but not renal histology or measurements of creatinine. Myotoxicity is only a minor clinical problem in envenoming by all of these snakes 27 . One study reported an unusually high early increase in creatine kinase in mice 3 hours post-injection of D. russelii venom, which is too early to be due to systemic myotoxicity and not consistent with a previous in vivo examination of systemic myotoxicity due to snake venoms in rats 28 . Coagulopathy is the most important clinical effect in Sri Lankan Russell's viper and saw-scaled viper envenoming, so testing the efficacy of antivenom against the procoagulant effect is most appropriate. Other haemotoxic venom effects were also not tested, such as haemorrhagic effects and platelet toxicity. Although haemorrhagic effects are important for E. carinatus neither of the other haemotoxic effects are important in D. russelii bites.
Neurotoxicity is the most important clinical effect in B. caeruleus envenoming 27 , so testing antivenom against neurotoxicity was most appropriate. In addition, we tested the efficacy of the antivenoms to bind to venom antigenic components as a general assessment of antivenom efficacy. There was good correlation between binding efficacy and efficacy against medically important clinical effects, except for cobra neurotoxicity where BHARAT antivenom was found to bind more effectively. VINS antivenom had excellent binding efficacy for D. russelii, and more recent vials had better binding efficacy for B. caeruleus and E. carinatus. The binding efficacy was statistically significantly better for VINS for all snakes except E. carinatus. VINS antivenoms were also more efficacious in neutralising procoagulant effects for D. russelii and E. carinatus, and neurotoxic effects for B. caeruleus. Neither antivenom was able to neutralise the neurotoxic effects of D. russelii venom. Although neurotoxicity occurs in about half of D. russelii bites in Sri Lanka it is rarely life-threatening 29 . Neurotoxicity is only reported for Russell's viper bites in Sri Lanka and Southern India 30 , so the venoms used in making the Indian antivenoms may not contain these neurotoxins 31 . This supports assessing antivenoms using tests of in vitro efficacy against clinically relevant toxicity.
In contrast to efficacy as assessed by in vitro binding and neutralisation efficacy, efficacy assessed by traditional ED 50 testing against lethality in mice, was not statistically significant between the antivenoms for any of the venoms. The relative efficacy of the two antivenoms based on ED 50 values was not consistent with any of the testing against important clinical effects in any of the snakes, suggesting that relying on such testing is problematic. The reason for this is that death in animals (e.g, mice in this study) could be due to a range of toxicities including some clinically irrelevant toxic effects important in human envenoming. It is entirely possible that post-synaptic neurotoxins or early cardiovascular collapse are major causes of lethality in rodents 32 , but are far less important in humans, in which presynaptic neurotoxins and procoagulant toxins are more prominent. A recent study has found that the procoagulant toxins in snake venoms have different effects on human and animal plasmas, making interpretation of efficacy of antivenom in rodent models problematic 12 . Based on the results of the current studies it would appear to be more appropriate to use clinically relevant in vitro studies of antivenoms against venom effects, and great care should be taken when interpreting in vivo animal models. However, further work is required on other snakes worldwide to confirm our findings for all antivenoms.
VINS antivenom appears to be the more efficacious compared to BHARAT antivenom. A dose of 10 vials is sufficient to bind all free venom antigens from these four snakes for venom concentrations found in patients with human envenoming. In addition, this dose was also able to neutralise the procoagulant effects in vitro of D. russelii venom and E. carinatus venom, and the neurotoxic effects of B. caeruleus. In contrast, the lethality studies did not appear to provide as useful an assessment or comparison of the efficacy of the two antivenoms. Bradford reagent (Bio-Rad, Catalogue # 500-0205) and Bovine Gamma Globulin (Bio-Rad, Catalogue # 500-0208) were used for protein quantification. Tris-buffered saline (TBS) was used to make up dilutions of antivenom for neutralization of D. russelii and E. carinatus venom procoagulant activity studies. Fresh frozen plasma was obtained from the Australian Red Cross and aliquots of 10 mL were thawed at 37 °C. Tetramethylbenzidine (TMB) from Sigma, bovine serum albumin (BSA) from Bovogen, Australia and Streptavidin-conjugated horseradish peroxidase (Streptavidin HRP) from Millipore Chemicon were used for the binding studies. Blocking solution is 0.5% BSA in phosphate buffered saline (PBS). Washing solution is 0.02% TWEEN 20 in PBS. Polyclonal monovalent rabbit IgG to Russell's viper venom was obtained by injection of rabbit with D. russelii venom followed by purification of the serum on a Protein G-Sepharose column and was carried out at the Western Australian Institute of Medical Research. Polyclonal monovalent rabbit IgG to E. carinatus, N. naja and B. caeruleus venom were obtained by injection of rabbits with the corresponding venoms, followed by purification of the serum on a Protein G-Sepharose column and was carried out at the Faculty of Medicine and Allied sciences, Rajarata University, Sri Lanka. Rabbit IgG antibodies were biotinylated using EZ-Link Sulfo-NHS-LC-Biotin (Pierce # 21335).

Methods
Binding and neutralisation studies of D. russelii venom were undertaken at venom concentrations measured in cases of human envenoming and taken as 1.7 μ g/ml, the 97 th percentile of venom concentrations in a previous study of 257 patients with pre-antivenom venom concentrations ranging from 0.0033 to 14.8 μ g/mL 21 . For E. carinatus venom the venom concentration used was 0.5 μ g/mL based on the maximum concentrations measured in one study of Echis ocellatus envenoming 33 . For N. naja and B. caeruleus venoms the venom concentrations was again taken as the 97 th percentile which was 1.4 μ g/ml for N. naja and 0.15 μ g/ml for B. caeruleus based on nine envenomed patients from Sri Lanka for each venom. Only one batch of antivenom from each manufacturer were compared statistically -VINS (01011/10-11; 2010) and BHARAT (A5311006; 2011Jan)-because multiple vials were available.
Protein quantification. Quantification of the protein content in each antivenom was undertaken using the Bradford protein assay method 34 Inter-and intra-batch protein quantification was carried out for all 36 vials of antivenom. Bradford reagent (150 μ l) was added to a solution of antivenom in PBS (150 μ l). After 10 minutes absorbance at 595 nm was measured on a Bio-Tek ELx808 plate reader. Concentrations of proteins were calculated with reference to a standard curve based on bovine gamma-globulin. Samples were measured at three dilutions.
Venom-antivenom binding studies. The following antivenom vials were used for antivenom venom binding studies-VINS 1061; VINS 01011/10-11; VINS 01013/10-11; VINS 01023/10-11; VINS 01024/10-11; VINS 01AS11112; VINS 01AS11114; BHARAT A5311006; BHARAT A5311013; BHARAT A5311014. Solutions of increasing concentrations of antivenom (0 to 4.3 mg/mL for E. carinatus venom and 0 to 17.1 mg/mL for D. russelii, N. naja and B. caeruleus venom) in blocking solution (0.5% Bovine Serum Albumin in PBS) were incubated with venom (D. russelii 1.7 μ g/mL, E. carinatus 0.5 μ g/mL, N. naja 1.4 μ g/mL and B. caeruleus 0.15 μ g/mL) for one hour at room temperature. Unbound venom antigens were detected using a sandwich enzyme immunoassay (EIA) as previously described 35 . In brief, Greiner Microlon 96-well high-binding plates were coated with the four different monovalent rabbit anti-snake venom IgGs (1 μ g/mL) in carbonate buffer (50 mM, pH 9.6), kept at room temperature for 1 h and then at 4 °C overnight. The plates were then washed once with PBS containing 0.02% TWEEN 20 and 300 μ L of blocking solution of 0.5% BSA in PBS was added. After 1 h the plates were washed again, and 100 μ L of venom-antivenom mixture was applied, after first diluting appropriately in blocking solution applied as (1:400 for D. russelii, 1:25 for E. carinatus, 1:140 for N. naja, 1:15 for B. caeruleus) dilutions in blocking solution. The plates were allowed to stand for 1 h and then washed three times. Next, biotinylated anti-snake venom IgG (D. russelii 0.5 μ g/mL, E. carinatus 8 μ g/mL, N. naja 0.12 μ g/mL and B. caeruleus 1 μ g/mL in blocking solution) was added. After standing for a further hour the plates were washed again. Streptavidin-horseradish Scientific RepoRts | 6:26778 | DOI: 10.1038/srep26778 peroxidase (100 μ L, 0.1 μ g/mL in blocking solution) was added and left for 1 h. The plate was then washed three times and 100 mL of TMB reagent added and colour allowed to develop for 3.5 min. The reaction was stopped by the addition of 50 mL of 1 M H 2 SO 4 . All samples were measured in triplicate, and the averaged absorbance converted to a concentration of the venom of interest by comparison with a standard curve based on serial dilutions of venom.

Neutralization of procoagulant activity of D. russelii and E. carinatus venom. The same
antivenom vials used for D. russelii and E. carinatus venom binding studies were used for neutralization studies. The ability of antivenom to neutralise the procoagulant activity of D. russelii and E. carinatus venoms was measured using the turbidimetric method 13 . Solutions of increasing concentrations of antivenom in TBS (0 to 1 mg/mL for D. russelii venom and 0 to 30 mg/mL for E. carinatus venom) were incubated with 1.7 μ g/mL of D. russelii venom or 0.5 μ g/mL of E. carinatus venom for 30 min at 37 °C in a 96 well plate. Fresh frozen plasma (100 μ l) containing 40 μ l of 0.4 M CaCl 2 /mL and venom-antivenom solution were added simultaneously to each well using a multichannel pipette. After a 5-second shaking step, the optical density at 340 nm was monitored every 30 s for 20 min. The clotting time was defined as the time until the rapid increase in absorbance, as calculated by Gen5 software (supplied with the Biotek ELx808 plate reader).
In vitro neurotoxicity studies. VINS (01AS11114) and BHARAT (A5311014) antivenom batches were used to investigate the neutralisation of the neurotoxic effects of Russell's viper (D. russelii), Common cobra (N. naja) and Indian krait (B. caeruleus) venoms. Chicks (4 to 10-day-old males) were killed by CO 2 inhalation and exsanguination, and the two biventer cervicis muscles were removed from the back of the neck. Each muscle was attached to a wire tissue holder and placed in a 5 mL organ bath filled with physiological salt solution with the following composition (mM): NaCl, 118.4; NaHCO 3 , 25; glucose, 11.1; KCl, 4.7; MgSO 4 , 1.2; KH 2 PO 4 , 1.2 and CaCl 2 , 2.5. The organ baths were bubbled with carbogen (95% O 2 , 5% CO 2 ) and maintained at a temperature of 33-34 °C under a resting tension of 1 g. Motor nerves were indirectly stimulated every 10 s (0.2 ms duration) at supramaximal voltage using a Grass S88 stimulator. The tissues were equilibrated for 10-15 min after which d-tubocurarine (10 μ M) was added, and the subsequent abolition of twitches confirmed the selective stimulation of the motor nerves. The tissues were then washed repeatedly until twitch height was restored. Contractile responses to acetylcholine (ACh; 1 mM for 30 s), carbachol (CCh; 20 μ M for 60 s), and KCl (40 mM for 30 s) were measured in the absence of stimulation 36 . At the conclusion of the experiment, responses to ACh, CCh, and KCl were measured again. Each of the four venoms at two concentrations (3 and 30 μ g/mL) were initially added to the organ bath without antivenom to determine the neurotoxic potency of each venom. Only B. caerulus venom had significant neurotoxicity at concentrations seen in human envenoming (3 μ g/mL). Neurotoxicity potency was measured as the t 90 which is the time required to cause 90% inhibition of the initial twitch height, for a given concentration of venom, expressed as a mean + /− SD. For some venoms, where the twitch height inhibition for a particular venom/treatment did not reach 90%, the time to maximum twitch inhibition observed in all tissues of that treatment/venom group was considered for all the groups for comparison.
Antivenom was added to the bath and incubated for 10 min. D. russelii (30 μ g/mL), N. naja venom (30 μ g/mL) or B. caeruleus venom (3 μ g/mL) was then added and left in contact with the tissue for 90 min. The antivenom concentration used was based on the manufacturer's instructions (1 ml of antivenom neutralises 0.6 mg N. naja venom, 0.45 mg B. caerulus venom and 0.6 mg of D. russelii venom). Experiments were done for both the recommended dose and five times the recommended dose.
Lethality neutralization studies. Lethality neutralization studies on antivenoms were tested by calculating the median lethal dose (LD 50 ) and the median effective dose (ED 50 ) values in mouse experiments. ICR (Institute of Cancer Research) mice, both sexes, weighing 18-20 g were used for all experiments. The following antivenom vials VINS0101110/11 and Bharat A5311006 were used for lethality studies. For this study, 250 mg/ml solutions of the antivenoms were prepared for venom-neutralization studies which was higher than the manufacturer's recommended concentration because the recommended concentration was too dilute and therefore too large a volume for administration to mice Assessment of the LD 50 and ED 50 experiments followed the methods described by the World Health Organisation (WHO) (2010) 9 . After dose ranging studies, varying doses of venom or venom and antivenom mixtures were injected to multiple groups of five mice in both LD 50 and ED 50 experiments. In all ED 50 studies, five times the LD 50 of the respective venom was mixed with varying amounts of antivenom and incubated at 37 °C for 30 min before injection. All injections were intravenous to the tail veins and were made to a final constant volume of 250 μ l by adding normal saline. In both LD 50 and ED 50 studies, death/survival rates were recorded for 48 hours. The ED 50 values were expressed in μ g of antivenom per g body weight of mouse (μ g/g) to neutralize the challenge dose of venom. The median effective ratio (ER 50 ) was also calculated using the following formula, where n is the number of LD 50 doses given (5 in this study).
Ethics Approvals. Ethical  Analysis of results. Standard curves for enzyme immunoassays and calculations of EC 50 were fitted by non-linear regression. For the in vitro neurotoxicity data the twitch heights were analysed by one way ANOVA followed by Bonferroni post-hoc tests, and the time to a reduction in twitch heights by Kruskal-Wallis test followed by Dunn's multiple comparison test. Standard error of the mean (SEM), standard deviations (SD) and 95% confidence intervals (95% CI) were all calculated for parametric and non-parametric data respectively. The difference in protein content between vials from one batch of each of the two antivenoms was compared using an unpaired t-test. Log EC 50 values were compared using the extra sum of the squares F test in Prism when comparing one antivenom from each manufacturer. The 95% confidence intervals of the ED 50 values were compared to determine if the ED 50 values were significantly different. Statistical significance was set at p < 0.05. All analyses and graphics were done in GraphPad Prism version 6.03 for Windows, GraphPad Software, San Diego California USA, www.graphpad.com, except for the calculation of the LD 50 and ED 50 which was done using the probit method 38 using SPSS statistical software version 20.0.