Development, validation, and application of a fully quantitative method for the determination of pyridine in crustacean tissues (and application of the same method in sediments).

Date in format: 25/09/2023

Crown copyright 2022

This information is licensed under the Open Government Licence v3.0. To view this licence, visit [[www.nationalarchives.gov.uk/doc/open-government-licence/]] 

This publication is available at [[www.gov.uk/government/publications]]

www.cefas.co.uk

{TOC}
===== 1. Executive Summary =====
	* Pyridine was implicated as a cause in the mass mortalities events (MMEs) that occurred during the autumn 2021 in the NE coast of England involving mainly crabs and lobsters. This was inferred from high pyridine levels reported in some crabs from the affected area, following analysis by the Environment Agency (EA).
	* The analytical screening method used by the EA (although accredited for water samples), was neither quantitative nor validated for biota (or for sediment) samples but was used within this incident to identify lines for potential follow up.
	*Due to continuing external concerns over pyridine, Defra commissioned this work at Cefas to develop and validate a robust quantitative method for pyridine in these environmental matrices.
	* This method was used to re-analyse stored samples collected during the MMEs.
	* Cefas analysts developed and validated a method, using a head space injection gas chromatography coupled to a mass spectrometer, HS-GC-MS technique. The limit of detection (LOD) and the limit of quantification (LOQ) of the method was 0.006mg/kg and 0.02mg/kg wet weight (ww) for the shellfish matrix and 0.002mg/kg and 0.008mg/kg ww for the sediment matrix, respectively.
	* The analytical method developed in this study, demonstrated fit-for-purpose performance criteria for biota matrices, including a recovery range of 89-101% and an associated within batch coefficient of variation (relative standard deviation) of 2-3% across three concentration levels (5, 25 and 500mg/kg).
	* Reanalysis of crustacean samples that had originally returned high indicative pyridine levels with the EA method (3-429mg/kg) demonstrated very low concentrations of the chemical (<0.22-0.077mg/kg, over 3 orders of magnitude lower). Analysis of additional crab samples, unrelated to the events, also demonstrated the presence of pyridine at very low levels (<0.02-0.139mg/kg).
	* A single mussel sample returned a value of 2.36mg/kg.
	* Pyridine levels in sediments collected in November 2021 all returned values between the LOD and LOQ. Three sediment samples collected in January 2022 returned values above the LOQ (0.014, 0.008, 0.009mg/kg ww), although below 2x LOQ.
	* Both sediment and biota samples returned levels of pyridine within an expected range based on the low environmental persistence, and high biodegradation rate of the chemical.
	* It is therefore considered very unlikely that pyridine, as a single chemical entity, was the cause of the crab and lobster mortalities during autumn 2021.
===== 2. Background =====

In the autumn of 2021, a mass mortality event (MME) concerning primarily, marine decapods (crabs and lobsters), occurred along the NE coast of the UK. As first responders for marine mortality incidents, the Environment Agency (EA), launched a major pollution investigation, which involved, amongst other things, collecting samples (water, sediment, and biological material (referred to hereafter as biota) to analyse for chemical pollutants to determine possible causes. Screening for several hundreds of contaminants1 did not reveal the presence or levels of toxic substances(s) which could definitively explain the event. Pyridine was detected in tissues of some of the affected crabs using the screening qualitative method of analysis which initiated further investigation. The United Kingdom Accreditation Service (UKAS) accredited ISO/IEC 17025 GCMS screening method for chemicals in waters 2. The EA method include pyridine in the target database but is not validated (or intended for routine use) in either biota or sediment matrices. Application of this method to the tissues of some affected and unaffected crab tissues indicated the presence of unexpectedly high levels of pyridine. It was acknowledged from the outset that the method could only provide a qualitative data. The reasons for this were (1) the sample matrix, in this case water compared to biota, can have a substantial effect upon the reliability of results in non-target matrices, (2) the analytical approach, qualitative versus fully quantitative, was not designed to determine precise numerical values. Thus, the accuracy of pyridine levels indicated by these exploratory screening tests was highly uncertain. Confirmation was required to verify the potential significance of these findings in the context of the MME.

In December 2021 Defra commissioned the Centre for Environment, Fisheries and Aquaculture Science (Cefas) to develop and validate a fully quantitative method for pyridine in biota and sediments. This report describes the development, validation, and performance characteristics of the method, and presents quantitative analyses of archived samples, collected, and analysed originally during the initial investigation.

1 Evidence gathered under the Defra-group investigation into Crustacean Mortality of Autumn 2021 - [[https://www.data.gov.uk/dataset/120483de-417e-4e4d-9a61-1fb81aea15e0/evidence-gathered-under-the-defra-group-investigation-into-crustacean-mortality-of-autumn-2021|data.gov.uk]]

2 [[https://www.ukas.com/wp-content/uploads/schedule_uploads/00002/0754Testing-Multiple.pdf|0754Testing Multiple (ukas.com)]]
===== 3. Development and validation of a method of analysis for pyridine =====
==== 3.1. Instrumental analysis ====

There are several methods that can be used for the analysis of pyridine in environmental samples (including biota and sediments) (IARC 2000). Following internal literature review, method development in this study focused upon analysis using headspace injection gas chromatography (HS-GC) coupled to a mass spectrometer (MS) used in tandem mode (MS/MS). The choice of HS-GC-MS over the more traditional injection techniques that use gas chromatography (GC), was informed by the need to minimise loss of pyridine during initial sample processing. Standard GC methods involve injection of a liquid sample or sample extract onto the front end of a GC column, whilst the head space technique injects part of the sample that has been volatilized whilst contained within a sealed vial. This means that volatile chemicals can be “extracted” from the sample directly without the need for additional steps (such as Soxhlet and further clean-up of the extracts). HS-GC-MS approaches therefore bypass the risk of loss of target chemicals during sample preparation steps and reduces interferences.

The instrument used for sample injection was a Shimadzu HS-20 headspace system. Separation of pyridine was performed with a Shimadzu GC-2010Plus (Shimadzu, Japan) using a Rxi-5Sil MS column 60m x 0.25mm id x 1.0µm (Restek, Bellefonte, PA, USA). This very long column with relatively thick stationary phase was needed to separate pyridine from the solvent front and from the deuterated form. The analyser was a Shimadzu TQ8050 triple quadrupole MS used in electron ionization (EI) mode. Instrumental parameters are summarized in Table 1.

Identity of native pyridine (not labelled) and deuterated pyridine (d5 labelled pyridine) were confirmed by their spectrum, obtained running the MS in full scan (Figure S1, Annex I). To improve sensitivity and selectivity, quantification of pyridine was carried out in multiplereaction monitoring mode (MRM) using the parameters summarized in Table 2.

Table 1: Instrumental parameters used for the analysis of pyridine by HS-GC-MS.
|t ,13 Head space|Injection volume|0,0 1µL|
|Oven temperature|0,0 80°C|
|Sample liner|0,0 150°C|
|Transfer line|0,0 150°C|
|Shaking level|0,0 3|
|Pressurizing gas pressure|0,0 76kPa (nitrogen)|
|Equilibrating time|0,0 30min|
|Needle flush time|0,0 2min|
|Pressurizing time|0,0 0.5min|
|Pressure equilibration time|0,0 0.1min|
|Load time|0,0 0.5min|
|Load equilibration time|0,0 0.1min|
|Injection time|0,0 0.5min|
|t ,3 GC|Oven temperature|0,0 100°C (isocratic, for 15min)|
|Carrier gas|0,0 278.2kPa (helium)|
|Split ratio|0,0 200|
|t ,4 MS|Detector|0,0 70eV|
|Voltage|0,0 0.7kV relative to the tuning|
|Transfer line temperatures|0,0 300°C|
|Ion source temperatures|0,0 250°C|



Table 2: MS/MS parameters (transitions and collision energies) used for the analysis of native pyridine and deuterated pyridine.
|,2 MS/MS|Quantitative|CE (V)|Qualitative|CE(V)|
|(m/z)||(m/z)||
|Pyridine|0,0 79>79|0,0 5|0,0 79>52|0,0 15|
|Pyridine d5|0,0 84>84|0,0 5|0,0 84>57|0,0 15|
=== 3.1 Quality Control and Quality Assurance ===
=== 3.1.1 Preparation of standards for calibration and reference materials ===

The stability of pyridine during laboratory storage is not known, so fresh standards for native pyridine (stock standards) were prepared weekly to improve accuracy of both reference and sample quantitation. To account and correct for any losses of pyridine during the sample preparation and analysis, d5-pyridine was added to samples, blanks, and standards as an internal standard at the beginning of the process and was used as an injection standard for quantification of the native pyridine, thereafter, referred to as pyridine. The d5 standard was prepared <14 days before use (since providing the standards, samples, blanks, and reference materials were run in the same batch and were spiked simultaneously, the d5 standard concentration is not a critical quantification parameter).

The volatility of chemicals in head space usually varies depending on the composition/type of matrix (interfering or enhancing the signal of the pyridine). To account for this expected variation, the calibration curves used in this study were carried out using a matrix matched calibration approach. In brief, both clean biota (lobster, Homarus gammarus, shellfish matrix) collected from the Isle of Man in 2022) and sediment (previously collected, characterised as clean and stored sediment from Shoebury East beach, Essex, (sediment matrix)) were utilised as blank matrixes. These blank samples were spiked at different levels to create both the calibration curves (9 levels) and the in-house reference materials (3 levels). During samples analysis, only the lowest 5 points of the calibration curve were used for quantification due to all samples registering pyridine levels close to the bottom of the calibration range.

Note: The number of samples from the mortality event required that analyses extended over several days. To ensure consistency of test results, samples were divided in batches. For each batch samples, calibration standards and reference materials were all prepared together (same day) and they were all analysed in the instrument together.
=== 3.1.2 Method validation ===

Validation was carried out to evaluate the performance characteristics of the quantitative pyridine method in biota (using lobster as shellfish matrix). Whilst previous analyses undertaken at the EA were considered only a qualitative screen, pyridine values from crustacean tissues were recorded within the range 3mg/kg (low) to 439mg/kg (high). Therefore, it was considered important to ensure that the quantitative method was reliable over an equivalent linear concentration range, and that validation was carried out over that range. Consequently, the blank matrix (lobster muscle tissue) was spiked at 3 concentration levels: 5mg/kg (low), 25mg/kg (medium) and 500mg/kg (high). Six replicates were used at each concentration level. Results for the validation are given in Table 3. Recoveries for all levels ranged between 89% and 101% and the associated within batch Relative Standard Deviation (RSD) between 2% and 3%. Consequently, performance characteristics were well within the accepted range of target values as per Cefas analytical quality requirements (Annex II, Table S1). Method standardised uncertainty was 8.11% and the expanded uncertainty 16%, both calculated following the equations in Annex II. These criteria were based on those of Eurachem (2014).

These results confirmed the suitability of the method used for the quantitation of pyridine in biota samples.



Table 3: Concentrations and method performance characteristics for the validation tests using lobster as blank shellfish matrix spiked with native pyridine at 5, 25 and 500mg/kg (wet weight).
|Conc. (mg/kg)|R1|R2|R3|R4|R5|R6|Average|SD*|RSD** (%)|Recovery (%)|
|Low (5mg/kg)|0,0 5.03|0,0 5.03|0,0 4.83|0,0 4.97|0,0 5.25|0,0 5.12|0,0 5.04|0,0 0.143|0,0 3|0,0 101|
|Medium (25mg/kg)|0,0 24.8|0,0 25.0|0,0 23.9|0,0 24.9|0,0 23.5|0,0 23.9|0,0 24.3|0,0 0.651|0,0 3|0,0 97|
|High (500mg/kg)|0,0 454|0,0 431|0,0 439|0,0 442|0,0 447|0,0 456|0,0 445|0,0 9.72|0,0 2|0,0 89|
=== 3.1.3 Linearity, Limit of quantification (LOQ) and limit of detection (LOD) ===

Limit of quantification (LOQ) and limit of detection (LOD) were determined using repeat injections of the lowest calibration point. The injection of 0.01 and 0.05mg/kg of pyridine resulted in the determination of the limit of detection (LOD) of 0.006mg/kg and the limit of quantification (LOQ) of 0.02mg/kg ww for the lobster. For the sediment, the LOD was 0.0023mg/kg and the LOQ 0.008mg /kg ww.

Quantification was undertaken using a bracketing technique. In brief, for each run, a 5-point serial log dilution calibration curve, with pyridine concentrations ranging from 0.01mg/kg to 1000mg/kg, was analysed followed by matrix blank/s, matrix reference material and samples, followed by a repeated calibration curve. At least one matrix blank and one in-house reference material were analysed for every 10 samples.

Low concentrations of pyridine were found to be present naturally in the unspiked lobster and sediment calibration samples and blanks were quantified using a standard addition method (as the intersect with the X axis or y=0). Concentrations for the in-run calibration curves were subsequently corrected to account for these background levels, subtracting 0.007mg/kg for sediment and 0.010mg/kg for lobster.

The choice of the concentration ranges used for the calibration curves was based on the expected concentrations in the samples. Independently of the range, a minimum regression correlation coefficient, R2 of > =0.98 was required to consider a calibration curve acceptable for quantification. A linear or a quadratic fitting were chosen depending on the response of the instrument to ensure a good R2 and accuracy of result. To minimise the potential for instrumental carryover (i.e., residual pyridine from high concentration standards impacting the results of subsequent sample runs), several instrumental blanks (empty vials or vials with only water) were analysed after each high concentration sample. No values above LOQs were detected in any of these instrumental blanks.
===== 4. Sample Analysis =====
==== 4.1 Sample collection, transport, and storage ====

Biota and sediment samples were collected by the Environment Agency (EA) during the mortality event in autumn 2021 and stored at ~-20°C. Samples were transported frozen to the Cefas Lowestoft laboratory in March (sediments) and April (biota) 2022. All samples were transferred directly (as received) to a freezer and kept frozen at ~-20°C until analysis commenced. The biota samples for analysis consisted of fresh, frozen or processed, chopped crabs and mussels in glass jars. Sediment samples were wet, frozen in either glass jars or plastic bags.

In addition, two claw samples derived from a single crab collected by the Cefas Weymouth Laboratory as part of a separate national monitoring programme for aquatic animal health assessment in late November 2022 from a location close to the original mass mortality area were analysed. Following euthanasia using a CrustaStun (a humane stunning device designed to administer a lethal electric shock to crustacean species), the two claws were treated differently. In brief, following confirmation of death, claws were aseptically removed from the animal, one was immediately frozen (-80°C), whilst the second claw was left at room temperature for ~60 h before transferring to -80°C. The aim of this differential treatment was to explore the hypothesis that pyridine levels were affected by the freshness of the sample. This was considered important given that the time of sample collection in relation to the time of death following the events was unknown. Following frozen transport to the Cefas Lowestoft laboratory, the claw samples were chopped as prepared for analysis.

All sample information is provided in Tables 4 (biota) and 5 (sediments).

Table 4. Information about the EA-collected biota samples that were analysed at Cefas.
|Sample ID|Date collection|Latitude|Longitude|Species|Sample type|Additional info|
|Bran Sands Tees|0,0 8/10/21|0,0 54.633511|0,0 -1.140149|Blue mussels|Incident impacted area||
|Seaton Carew Shore|0,0 25/10/21|0,0 54.679577|0,0 -1.197454|Shore crabs|Incident impacted|0,0 2 crabs|
|Runswick Bay Shore|0,0 8/11/21|0,0 54.533203|0,0 -0.747815|Shore crabs|Incident impacted|0,0 2 crabs (sampled live)|
|Bran Sands|0,0 8/10/21|0,0 54.635471|0,0 -1.141705|0,0 1 Edible crab + 1 shore crab|Incident impacted||
|Saltburn|0,0 25/10/21|0,0 54.588394|0,0 -0.976879|Shore crabs|Incident impacted|2 crabs|
|St Mary’s Island|0,0 5/11/21|0,0 55.069736|0,0 -1.450062|Shore crabs|Comparison (outside of impacted area)|Tiny|
|Penzance shellfish merchant|-|-|-|Edible crab|Comparison (outside of impacted area)|0,0 1 crab|
|South Shields|0,0 9/11/21|0,0 55.002472|0,0 -1.377162|Edible brown|Comparison (outside of impacted area)|0,0 1 crab|
|Eastern IFCA (small bag)|-|Norfolk Wash|-|-|Comparison (outside of impacted area)|0,0 1 crab|
|Eastern IFCA (big bag)|-|Norfolk Wash|-|-|Comparison (outside of impacted area)|0,0 1 crab|
|Old crab – shore crab (small)|-|-|-|Shore crab|EA method development samples|-|
|Old crab – edible (big)|-|-|-|Edible crab|EA method development samples|-|
|Blue crab Atlantic boiled|-|Leeds market|-|Edible crab|EA method development samples|Boiled|
|Flamborough boiled|-|Leeds market|-|-|EA method development samples|Boiled|
|Scarborough edible (live)|-|Leeds market|-|Edible crab|EA method development samples|Sampled live|
|Crab claw fresh|0,0 30/11/22|0,0 53.51571N|0,0 00025262E|Edible crab|Storage test|Immediately frozen|
|Crab claw decomposing|0,0 30/11/22|0,0 53.51571N|0,0 00025262E|Edible crab|Storage test|Frozen after storage at 18°C for 60hours)|

Table 5. Information about the EA-collected sediment samples that were analysed at Cefas.
|Sample ID|Date of collection|Latitude|Longitude|Type|
|0,0 2 Control site - Robin Hoods Bay|0,0 15/11/2021|0,0 54.423462|0,0 -0.503459|Sand|
|0,0 3 Runswick Bay MCZ South East|0,0 15/11/2021|0,0 54.540504|0,0 -0.671226|Sandy mud|
|0,0 6 Sea off Saltburn|0,0 18/11/21|0,0 54.607643|0,0 -0.973968|Sand|
|0,0 9 Dredged channel|0,0 17/11/21|0,0 54.655060|0,0 -1.126675|Sand|
|0,0 10 Seaton Carew WFR otter trawl site|0,0 18/11/21|0,0 54.659367|0,0 -1.161645|Muddy sand|
|0,0 12 Hartlepool Bay|0,0 17/11/21|0,0 54.682388|0,0 -1.148107|Sand|
|0,0 14 N Gare Sands|0,0 18/11/21|0,0 54.644302|0,0 -1.156022|Sand|
|0,0 15 Phillips Buoy|0,0 18/11/21|0,0 54.628849|0,0 -1.162838|Soft mud|
|0,0 3. Runswick Bay MCZ South East|0,0 18/01/22|0,0 54.540604|0,0 -0.671097|Muddy sand|
|0,0 4. Runswick Bay MCZ North West|0,0 18/01/22|0,0 54.580597|0,0 -0.773950|-|
|0,0 5. Skinning Grove Wick|0,0 18/01/22|0,0 54.581200|0,0 -0.859756|Sand|
|0,0 6. Sea off Saltburn|0,0 18/01/22|0,0 54.607574|0,0 -0.973987|Sand|
|0,0 7. Sea off Redcar|0,0 18/01/22|0,0 54.634083|0,0 -1.082852|Sand|
|0,0 8. Dredger spoil ground|0,0 18/01/22|0,0 54.680363|0,0 -1.048587|Sand|
|0,0 9. Dredged channel|0,0 18/01/22|0,0 54.654949|0,0 -1.126897|Sand with mud layer beneath|
|0,0 10. Seaton Carew|0,0 18/01/22|0,0 54.659397|0,0 -1.161566|Slightly muddy sand|
|0,0 11. Sea off Seaton Carew|0,0 18/01/22|0,0 54.669201|0,0 -1.174547|Sand|
|0,0 12. Hartlepool Bay|0,0 18/01/22|0,0 54.682175|0,0 -1.148116|Sand|
|0,0 13. Tees Bay|0,0 18/01/22|0,0 54.658611|0,0 -1.081027|Slightly muddy sand|
|0,0 14. N Gare Sands|0,0 18/01/22|0,0 54.644276|0,0 -1.156318|Fine sand|
|0,0 15. Phillips buoy|0,0 18/01/22|0,0 54.628784|0,0 -1.163617|Anoxic mud|
|0,0 16. Teesport|0,0 18/01/22|0,0 54.598446|0,0 -1.171763|Anoxic mud|


==== 4.2 Sample preparation ====

On each day of analysis, samples were left to defrost at room temperature for a few hours (24h). Defrosted samples were mixed thoroughly to produce a homogeneous mixture. Approximately 1g of sample (weighed to the 3rd decimal point) was introduced into a 20mL solvent-rinsed head space vial containing 2 mL of UPLC filtered water Hypersolv Chromanorm (VWR chemicals, Lutterworth, Leicestershire, England). Each sample was pushed to the bottom of the vial and mixed with the water to improve surface contact. Deuterated pyridine (100µL of d5 at 300µg/mL) prepared in acetone was then added to each sample with a glass syringe as an injection standard and the vial was immediately sealed with a head space crimp cap. All samples were prepared sequentially to minimize loss of volatiles during the preparation process.

Calibration standards and in-house reference materials were prepared as follows. Blank matrix materials (samples collected from clean areas (section 3.2.1)), consisting of lobster muscle or clean sediment (~1g of blank sample) were placed into a 20mL solvent rinsed head space vial containing 2mL of UPLC filtered water Hypersolv Chromanorm (VWR chemicals, Lutterworth, Leicestershire, England). After mixing the sample with the water, pyridine standards (prepared in acetone as the deuterated pyridine standard) were added with glass syringes to achieve a range of concentration levels. Concentration levels were 5mg/kg for the reference material, and a range between 0.01mg/kg to1000 mg/kg for the calibration curve standards. Injection of pyridine was quickly followed by the addition of 100uL of injection standard (deuterated pyridine, d5). Blanks were prepared the same way but without adding any pyridine. Spiking of native and labelled pyridine was carried out sequentially and vials were capped immediately after spiking to minimize any loss of chemical during the preparation process.

All blanks, calibration standards, reference materials and samples were shaken in a rotatory shaker for 1h before being analysed in the instrument. Environmental samples (biota and sediment), processed in batches, were sonicated for 10min before shaking to further aid the transfer of pyridine from the sample to the water.
==== 4.3 Determination of pyridine in biota and sediment samples ====
=== 4.3.1. Biota samples ===

A total of 17 samples were analysed in this study, including 15 environmental biota samples (supplied by EA and collected during the MMEs) and two crab claws (supplied by colleagues from Cefas Weymouth laboratory) (Table 4). The samples received by the EA included nine crab samples previously analysed by the EA collected from the affected area during the events and six additional biota samples, not analysed previously and not related to the affected area or mixed biota samples.

The results obtained in this study are displayed graphically in Figure 1 and set out in Table 6. For comparison, the results found by the EA in their study have also been included. Recoveries obtained for the three reference materials (spiked blank samples) analysed with the samples ranged between 98 and 104%.

Example chromatograms obtained for some biota samples (Bran Sands Tess, Seaton Carew and Saltburn) are provided in Annex I (Figure S2).

{html}<img src="./DEFRA/2311Pyridine/Images/Figure1.jpg" alt="Figure 1" width="100%" />{/html}

**Figure 1.** Levels of pyridine (mg/kg wet weight) found in the samples analysed in this study.

//Light yellow colour identifies the extra samples, green the ones labelled as comparison and blue the storage test samples. The red line states the limit of quantification of the method.//

For comparison (at a different scale), levels obtained from by EA are also included in red (obtained from the “WFA 170322 Pyridine slides for discussion” sent by EA.

Table 6. Levels found above the limit of quantification (0.02mg/kg) for the biota samples analysed in Cefas using head space-GC-MS.
|Sample ID|Pyridine Cefas (mg/kg ww)|Pyridine EA (mg/kg)|
|Bran Sands Tees|0,0 2.36|na (not analysed)|
|Seaton Carew Shore|0,0 0.061|0,0 204|
|Runswick Bay Shore|0,0 <0.02 between LOD & LOQ|0,0 20|
|Bran Sands|0,0 0.077|0,0 255|
|Saltburn|0,0 0.047|0,0 439|
|St Mary’s Island|0,0 <0.02 between LOD & LOQ|0,0 78|
|Penzance shellfish merchant|0,0 0.031|0,0 6|
|South Shields|0,0 0.025|0,0 35|
|Eastern IFCA (small bag)|0,0 <0.02 between LOD & LOQ|0,0 3|
|Eastern IFCA (big bag)|0,0 <0.02 between LOD & LOQ|0,0 195|
|Old crab – shore crab (small)|0,0 0.139|na|
|Old crab – edible (big)|0,0 0.067|na|
|Blue crab Atlantic (boiled)|0,0 0.046|na|
|Flamborough (boiled)|0,0 0.068|na|
|Scarborough edible (live)|0,0 <0.02 between LOD & LOQ|na|
|Crab right fresh|0,0 0.022|0,0 na|
|Crab left decomposing|0,0 0.031|na|

Applying the validated quantitative method developed in this study to a range of frozen stored biota samples collected during the mortality event by the EA (both within and outside the affected area), and to a limited number of processed or fresh samples indicated detectable levels of pyridine in all samples, ranging between <0.02mg/kg to 2.361mg/kg ww (one mussel sample). The highest single value determined in crustacean tissues, designated ‘Old crab shore crab (small)’ identified as a shore crab and collected from the affected area was 0.139mg/kg ww. Whilst the sample size is small, an unpaired t-test comparing values obtained from all crustacean tissue samples (where exact sample provenance was known and using a conservative estimate of results between LOD and LOQ of 0.02mg/kg), derived from the within the affected area and, those collected from unrelated areas revealed no significant differences in pyridine content (mg/kg) (P<0.05). Equally, again whilst sample sizes are extremely small, no substantial change in pyridine content was measured in fresh (sample Crab claw fresh, 0.022mg/kg) and stored (sample – Crab claw decomposing, 0.031mg/kg).

These levels differ in some cases by up to 3 orders of magnitude from the indicative levels found by EA in their investigation to identify further lines of enquiry, which ranged, if interpreted as quantifiable samples, from 3 to 439mg/kg.
=== 4.3.2 Sediment samples ===

A total of 15 environmental sediment samples supplied by EA were analysed (Table 5). The results obtained in this study are displayed graphically in Figure 2 and set out Table 7. Recoveries obtained for the 3 reference materials (spiked blank samples) analysed with the samples ranged between 91 and 96%.

{html}<img src="./DEFRA/2311Pyridine/Images/Figure2.jpg" alt="Figure 2" width="100%" />{/html}

**Figure 2.** Levels of pyridine (mg/kg wet weight) found in the samples analysed in this study.

//The light blue bars identify the samples collected in November 2021, while the dark identify the ones collected in January 2022. The red line marks the limit of quantification of the method, the green lines the method limit of detection.//

As Figure 2 and Table 7 display, pyridine levels in sediments collected in November 2021 all (100%) returned values between the LOD and LOQ (0.008mg/kg). Three results (21%) from sediment samples collected in January 2022, returned values above the LOQ-Sea off Saltburn (0.014mg/kg ww), Sea off Seaton Carew (0.008mg/kg ww) and Teesport (0.009mg/kg ww) (all collected in January 2022), although below 2xLOQ.

Table 7. Pyridine levels (LOQ 0.008mg/kg wet weight) found in sediments.
| ,2 Cefas Analysis ID|Pyridine (mg/kg ww)|Pyridine (mg/kg ww)|
|Collected Nov 2021|Collected Jan 2022|
|0,0 2. Control site - Robins Hoods Bay|0,0 <0.008|na (na: not analysed)|
|0,0 3. Runswick Bay MCZ SE|0,0 <0.008|0,0 <0.008|
|0,0 4. Runswick Bay MCZ NW|na|0,0 <0.008|
|0,0 5. Skinning Grove Wick|na|0,0 <0.008|
|0,0 6. Sea off Saltburn|0,0 <0.008|0,0 0.014|
|0,0 7. Sea off Redcar|na|0,0 <0.008|
|0,0 8. Dredger spoil ground|na|0,0 <0.008|
|0,0 9. Dredged channel|0,0 <0.008|0,0 <0.008|
|0,0 10. Seaton Carew WFR otter trawl site|0,0 <0.008|0,0 <0.008|
|0,0 11. Sea off Seaton Carew|na|0,0 0.008|
|0,0 12. Hartlepool Bay|0,0 <0.008|0,0 <0.008|
|0,0 13. Tees Bay|na|0,0 <0.008|
|0,0 14. N. Gare Sands|0,0 <0.008|0,0 <0.0076|
|0,0 15. Phillips Buoy|0,0 <0.008|0,0 <0.008|
|0,0 16. Teesport|na|0,0 0.009|
===== 5. Discussion =====
==== 5.1 Development and validation of a method for ====

quantitative determination of pyridine in biota and

sediments

A quantitative method for pyridine analysis in biota and sediment samples, using head space injection gas chromatography coupled to a mass spectrometer (HS-GC-MS) in tandem mode, was developed and validated at the Cefas Lowestoft laboratory. The limit of detection (LOD) and the limit of quantification (LOQ) of the method is 0.006mg/kg and 0.02mg/kg ww for the shellfish matrix and 0.002mg/kg and 0.008mg/kg ww for the sediment matrix, respectively. Linearity range was from 0.01 to 25 mg/kg, with an R2 of >0.99. The validation of the method at three concentration levels (5, 25 and 500mg/kg ww) showed recoveries between 89-101% and RSDs of 2-3%. For this concentration range, the acceptable recoveries were 70-110% and RSDs <10%, see Annex II and Eurachem (2014)).

Pyridine is a volatile compound with a vapour pressure c.a. 2.kPa at 20°C and boiling point 115°C (NIH, 2023)- volatile is a compound with a pressure higher of 0.01 kPa at 20°C (EU, 1999) or boiling point between 50°C and 240°C (David and Niculescu, 2021).

As a volatile compound, pyridine has been analysed mainly by gas chromatography (GC) following distillation (Kroner et al., 1952; EPA 2006, EPA 2007). However, other methods, such as extraction with solvents (dichloromethane; EPA, 1996a, 1996b) and headspace approaches (ATSD, 1992) can also be used with GC as long as their suitability is tested (EPA, 2006, 2014).

Traditional extraction and enrichment procedures, such as solvent extraction, steam distillation and vacuum distillation, often have problems due to interferences (co-extraction of matrix components and introduction of extraneous compounds from the extracting medium) (Robards et al., 2004). Losses of chemical are also possible during sample evaporation steps (EPA 1996b). Headspace GC has been used since the 1950s for the determination of volatile compounds in solid and liquid samples (Sparkman et al., 2011). The advantage of this technique is that, because only the gas phase above the sample is introduced into the GC column, it is a very clean way of introducing the volatile analytes of the sample into a gas chromatograph (Robards et al., 2004; Sparkman et al., 2011). Consequently, it is frequently used for the determination of analytes in trace levels. The method used here for the analysis of pyridine is based on Gopalakrishnan and Devi (2016) for the determination of pyridine in pharmaceuticals using headspace. The method has been adjusted and adapted for application to the samples analysed and the equipment used for the analysis, following EPA “method 5021A for volatile compounds using headspace analysis” (EPA 2014).

Alongside the validation studies, stringent quality control included in the analysis included use of suitable blanks, in-house reference materials, confirmation of lack of interferences, identification and quantitation MS/MS transitions, labelled pyridine for recovery correction, suitable extraction solvent, pre-extraction before analysis, all of which contributed to a high confidence in the test results.

We suggest that the HS-GC-MS method developed here is robust and suitable for environmental samples as a) it minimises losses during sample preparation by extractive methods and b) is potentially more specific as solvent extraction methods are often associated with matrix effects that make quantification problematic.


==== 5.2 Application of HS-GC-MS method to stored biota and sediment samples ====
=== 5.2.1 Biota ===

Application of the HS-GC-MS method to biota samples collected and stored during the event period indicated low levels of pyridine in all crustacean samples (irrespective of geographical area (i.e., those collected from the Northeast coast as part of the MME from animals demonstrating clinical pathology and apparently healthy crabs from different regions of England).

The most probable reason explaining the differences in biota pyridine levels between original analysis by the EA (qualitative screening approach) and reanalysis by Cefas (using the fully quantitative and validated matrix specific methodology described here) involves the wellknown interferences from the matrix and methodological approach.

Nevertheless, degradation, or loss of pyridine during frozen storage (samples analysis at Cefas following method development took place more than a year later to the EA analysis) should also be considered as a potential factor, and may explain some of the variability.

Further time series analyses would be required to determine if pyridine is progressively lost during frozen storage and samples have been stored to allow this analysis in the future. However, two observations may suggest that this is unlikely (1) the lack of a proportional numerical change in pyridine levels between analysis and reanalysis, and (2) that comparative data obtained from subsamples of crab claws derived from the same animal were not significantly different when analysed fresh or following storage.

The pyridine levels recovered from biota samples in this study are consistent with regulatory data on bioaccumulation, as the water-soluble nature of pyridine suggests very low bioaccumulation potential. An octanol/water partition coefficient (Kow) estimates the likelihood of a chemical to partition to organisms in an aquatic environment. Octanol is believed to best imitate the fatty structures in plants and animal tissues. There are various methods that can be used to provide indication for bioaccumulation potential of chemicals.

**Table 8:** Parameters indicative of bioconcentration potential for pyridine
|Parameter|Description|Tool|LogBCF|
|logPACD|logP value calculated by ACD software|ACD software|0,0 1.3329885|
|logPKowwin|logP value calculated by Kowwin software|Kowwin software|0,0 1.2674849|
|logPMDL|logP value from MDL descriptors Moriguchi octanol-water partition coeff.|MDL descriptors|0,0 1.1825917|
|MLOGP|(logP)|Dragon|0,0 1.1469135|

Pyridine has a very low bioconcentration potential in biota tissues (Table 8), this indicates that pyridine will not partition to fatty tissues in plants or animals. The logBCF for pyridine has also been experimentally established and determined as 1.9 (de Vooght et al., 1991; BCF=88), which suggests an underestimation of pyridine’s bioaccumulation potential (within an order of magnitude) by modelling approaches. Nevertheless, the BCF of 88, is low considering criteria for bioaccumulation. USEPA for example, require a BCF>1000<5000 to characterise a chemical as bioaccumulative, whilst EU REACH defines a substance with a BCF>2000 as bioaccumulative. In this context, the original pyridine values reported by the qualitative EA method, would have been theoretically unlikely to reach without a concentration of pyridine in water of greater than 400mg/L. One of the highest recorded environmental concentrations of pyridine in water has been reported as 5mg/L in Australian oil-shale retort water (Dobson et al., 1985). Additionally, the biodegradation rate constant for pyridine in water is 0.00330.018/hour, which corresponds to half-lives of 39-210 hours (Gherini et al.; 1989). Pyridine was completely degraded in about 8 days in a river die-away test (Cassidy et al., 1988).
=== 5.2.2 Sediments ===

Application of the method to determine levels of pyridine in stored sediments also yielded very low results (below 0.015mg/kg ww). Although the lack of data on pyridine stability in frozen sediments remains uncertain.

There are few published studies on pyridine in marine sediments. Krone et al., (1986) reported concentrations of 0.22mg/kg in creosote-contaminated sediments in Puget Sound, Washington. However, there is evidence that pyridine is readily degraded in soil, and numerous bacteria isolated from soils or sludges can use pyridines as sole sources of carbon and/or nitrogen (Jerry et al., 1989). Most pyridine biodegradation studies have been concerned with transformations in soils and sewage sludges. In an experimental study of pyridine biodegradation in unfiltered river water, the rate of removal depended on the initial concentration of pyridine, but in general, at lower concentrations (less than 20mg/L), pyridine degradation was virtually complete in 8 days or less (Cassidy et al., 1988). No information concerning the microorganisms present in the water was given, but this study suggests that

biodegradation may be a much more rapid mechanism for the removal of pyridine from the environment than abiotic mechanisms. There is also evidence that pyridine can be biodegraded in soil where a branching bacterium (Proactinomyces) that can utilize pyridine as a source of carbon and nitrogen, and energy are present (Moore 1949). Pyridine was biodegraded in an aqueous extract of a garden soil, but the process was slower with complete degradation of pyridine under aerobic conditions requiring 66 -170 days, whereas under anaerobic conditions, degradation was faster (32-66 days) (Naik et al., 1972). However, in a soil incubated with low concentrations of pyridine, the compound was completely degraded in 8 days (Battersby and Wilson, 1989). Taken together these studies indicate the probability that pyridine is unlikely to persist for extended periods (years) in aquatic sediments. No evidence of high-level persistence was determined in sediments in the current study.
=== 5.2.3 Summary ===

This study has successfully achieved the aim of establishing a reliable method for quantitative determination of pyridine in biota and sediments and therefore extends the suite of analytical tools available for future investigations. The methodological approach (head space injection gas chromatography coupled to a mass spectrometer, (HS-GC-MS)) relies upon the direct volatilisation of synthetic pyridine and thus is not impacted by matrix effects, which are common in solvent-based extraction approaches. The performance characteristics of the method were satisfactory and quality assurance criteria were met providing good confidence in the quantitative test results.

Low levels of pyridine were determined in biota samples both from the MME and samples collected from spatially distinct areas. All levels were close to the LOQ, or between the LOD and LOQ, with the exception of a single mussel sample (2.36mg/kg) (Bran Sands, October 2021). Pyridine levels in sediments collected in November 2021 and January 2022 were marginally above or below the LOD. No measurable difference in pyridine results between samples derived from the MME area or geographically distinct locations, or patterns or relationships between time of collection and/or time of storage, for either biota or sediment samples was found.

The EA initial qualitative analysis was undertaken to identify further lines of enquiry only, and as such data should not be directly compared to the quantitative findings in the present study. The limitations of the application of a validated water screening method to complex non-target matrices was well understood and acknowledged during the initial investigation. One limitation of the current study is the potential impact of frozen storage of samples on pyridine content, it is however considered most likely that the variation between EA qualitative analysis and data generated in this study can be explained by methodological approach and matrix effect.

The low levels of pyridine measured in both crustacea, and sediments using a validated quantitative method are within an expected range based upon the predicted low environmental persistence, and high biodegradation rate of the chemical. These levels would not be expected to cause effects in otherwise healthy crustacea. Furthermore, no significant difference in pyridine content between animals impacted in the MME and those sampled from unimpacted areas was determined. Importantly, these low values are consistent with the global regulatory view, which considers pyridine of low environmental toxicological concern. The lack of pyridine toxicity concerns at low levels, such as those present in the environment, can perhaps be demonstrated from the fact that pyridine is approved by the Food and Drug administration (FDA) for use as a flavouring agent (ATSD, 1992). Data presented in this study provide additional evidence supporting the conclusion that it is very unlikely that pyridine, as a single chemical entity, was the cause of the crab and lobster mortalities during autumn 2021.
===== 6. References =====

ATSD (Agency for Toxic Substances and Disease), 1992. Toxicological profile for pyridine. Registry U.S. Public Health Service, (https://wwwn.cdc.gov/TSP/ToxProfiles/ToxProfiles.aspx?id=535&tid=96 Access 13/04/23)

Battersby NS, Wilson V. 1989. Survey of the anaerobic biodegradation potential of organic chemicals in digesting sludge. Appl Environ Microbiol 55:433-439.

Cassidy RA, Birge WJ, Black JA. 1988. Biodegradation of three azaarene congeners in river water. Environ Toxicol Chem 7:99-105.

David E, Niculescu VC. 2021. Volatile Organic Compounds (VOCs) as Environmental Pollutants: Occurrence and Mitigation Using Nanomaterials. Int J Environ Res Public Health 18(24): 13147.

de Voogt, P, van Hattum B, Leonards P, Klamer JC, Govers H. 1991. Bioconcentration of polycyclic heteroaromatic hydrocarbons in the guppy (Poecilia reticula). Aquat Toxicol 20(3):169-194.

Dobson KR, Stephenson M, Greenfield PF, et al. 1985. Identification and treatability of organics in oil shale retort water. Water Res 19:849-856.

EPA (Environmental Protection Agency), 1996a. Method 8015C: Nonhalogenated organics using GC/FID

EPA (Environmental Protection Agency), 1996b. METHOD 3540C Soxhlet extraction

EPA (Environmental Protection Agency), 2006. Method 8260C: Volatile organic compounds by gas chromatography / mass spectrometry (GC/MS)

EPA (Environmental Protection Agency), 2007. Method 8261: Volatile organic compounds by vacuum distillation in combination with gas chromatography/mass spectrometry (VD/GC/MS)

EPA (Environmental Protection Agency), 2014. METHOD 5021A: Volatile organic compounds in various sample matrices using equilibrium headspace analysis.

EU, 1999. EU Council Directive 1999/13/EC of 11 March 1999 on the limitation of emissions of volatile organic compounds due to the use of organic solvents in certain activities and installations

Eurachem, Guide: The Fitness for Purpose of Analytical Methods: A Laboratory Guide to Method Validation and Related Topics: Second edition (2014)

Gherini SA, Summers KV, Munson RK, Mills WB. 1989. Chemical data for predicting the fate of organic compounds in water: Volume 1, Technical basis (No. EPRI-EA-5818-Vol. 1). Electric Power Research Inst., Palo Alto, CA (USA); Tetra Tech, Inc., Lafayette, CA (USA).

Gopalakrishnan J. Devi SA. 2016. Determination of triethylamine, pyridine and dimethyl formamide content in telmisartan by headspace gas chromatography using flame ionization detector. Indian J Pharm Sci 78(3): 413-416.

IARC, 2000. Pyridine. Some Industrial Chemicals, IARC Monographs on the Evaluation of Carcinogenic Risks to Humans Volume 77, Monograph Pyridine 503-528. ISBN 978-92832-1277-5

Krone CA, Burrows DG, Brown DW, et al. 1986. Nitrogen-containing aromatic compounds in sediments from a polluted harbor in Puget Sound. Environ Sci Technol 20:1144-1150.

Kroner, R C, Ettinger MB, Moore, WA. 1952. Determination of Pyridine and Pyridine-Base Compounds in River Water and Industrial Wastes. Anal Chem 24(12):1877–1881.

Moore FW. 1949. The utilization of pyridine by micro-organisms. J Gen Microbial 3:143-14.

Naik, MN, Jackson, RB, Stokes, J, Swaby RJ. 1972. Microbial degradation and phytotoxicity of picloram and other substituted pyridines. Soil Biology and Biochemistry, 4(3): 313-323.

NIH, 2023. Pyridine, National Library of Medicine, PubChem, https://pubchem.ncbi.nlm.nih.gov/compound/Pyridine#section=Vapor-Pressure, accessed 14/04/23

Robards, K, Haddad, PR, Jackson, PE. 2004. 3 - Gas chromatography Principles and practice of modern chromatographic methods. Academic Press, London, 2004. Principles and Practice of Modern Chromatographic Methods, Pages 75-177 - ISBN 0-12-589570-4, https://doi.org/10.1016/B978-0-08-057178-2.50006-6

Rudine, AB, Walter, MG, Wamser, CC, 2010. Reaction of Dichloromethane with Pyridine Derivatives under Ambient Conditions, J Org Chem 75(12): 4292–4295.

Sparkman OD, Penton Z, Kitson FG. 2011. Gas chromatography and mass spectrometry: a practical guide. Academic press; 2011 May 17. ISBN 978-0-12-373628-4
===== 7. Annex I =====

{html}<img src="./DEFRA/2311Pyridine/Images/FigureS1.jpg" alt="Figure 1" width="100%" />{/html}

**Figure S1:** Chromatogram and spectra for native and deuterated pyridine found in our study.

{html}<img src="./DEFRA/2311Pyridine/Images/FigureS2.jpg" alt="Figure S2" width="100%" />{/html}

**Figure S2:** Pyridine chromatograms in Brand Sands Tees mussel (left), Seaton Carew crab (middle) and Saltburn crab (right). Quantification transition (79>79) can be seen in black, confirmation transition (79>52, to confirm identity) in pink.

{html}<img src="./DEFRA/2311Pyridine/Images/FigureS3.jpg" alt="Figure S3" width="100%" />{/html}

**Figure S3:** Pyridine chromatogram in sediments from Sea off Seaton Carew (left), Sea off Saltburn (middle) and Teesport (right).
===== 8. Annex II =====

Cefas SOP 1405 “Method Validation (including uncertainty of measurements) and AQC”, taken from Eurachem, Guide: The Fitness for Purpose of Analytical Methods: A Laboratory Guide to Method Validation and Related Topics: Second edition (2014).

Table S1: Acceptable precision and accuracy criteria for quantitative methods depending on the analyte level.


|,2 Concentration range (mg/kg)|2,0 Repeatability|,2 Mean recovery range (%)|
|CVA%|CVL%|
|0,0 ≤0.00|0,0 35|0,0 36|0,0 50 - 120|
|0,0 0.001 - 0.01|0,0 30|0,0 32|0,0 60 - 120|
|0,0 0.01 – 0.1|0,0 20|0,0 22|0,0 70 – 120|
|0,0 0.1 – 1|0,0 15|0,0 18|0,0 70 – 110|
|0,0 >1|0,0 10|0,0 14|0,0 70 – 110|



Calculation of expanded uncertainty (%EU):

%EU = %SU x 2



Where:

%SU (standard uncertainty):

  %𝑆U =  Sqrt(PBIASp^2 + %RSDp^2)

%RSD (relative standard deviation):

  %RSD = ((SD of mean values)/(mean observed value)) x 100

%RSDp (pooled relative standard deviation):

  % RSDp = Sqrt({(RSDlvl1^2 x (nlvl1 - 1)) + (RSDlvl2^2 x (nlvl2 - 1)) + (RSDlvl3^2 x (nlvl3 - 1))} / {(nlvl1 - 1) + (nlvl2 - 1) + (nlvl3 - 1)})

PBIAS (percent relative bias):

  PBIAS = (bias / mean pooled value) x 100

PBIASp (pooled percent relative bias): Same equation as pooled %RSD.

**World Class Science for the Marine and Freshwater Environment**

We are the government’s marine and freshwater science experts. We help keep our seas, oceans and rivers healthy and productive and our seafood safe and sustainable by providing data and advice to the UK Government and our overseas partners. We are passionate about what we do because our work helps tackle the serious global problems of climate change, marine litter, over-fishing and pollution in support of the UK’s commitments to a better future (for example the UN Sustainable Development Goals and Defra’s 25 year Environment Plan).

We work in partnership with our colleagues in Defra and across UK government, and with international governments, business, maritime and fishing industry, non-governmental organisations, research institutes, universities, civil society and schools to collate and share knowledge. Together we can understand and value our seas to secure a sustainable blue future for us all, and help create a greater place for living.

© Crown copyright 2022

__________________________________________________________________

Pakefield Road, Lowestoft, Suffolk, NR33 0HT

The Nothe, Barrack Road, Weymouth DT4 8UB

www.cefas.co.uk | +44 (0) 1502 562244