Memorandum submitted by Integrin Advanced
Biosystems (M2)
SUMMARY
The "atypical" problem
stemmed from CEFAS and DARD's inexperience in running the DSP
Mouse Bioassay. This led to solvent-related deaths of mice most
probably because of anaesthetic effects and heart failure.
The lack of controls in the assay
meant that this was not immediately recognised as an artefact
and CEFAS and DARD did not conduct an appropriate investigation
of possible matrix effects. Nor apparently did they seek and heed
the advice of the more experienced FRS.
The initial response of the FSA in
closing cockle beds was correct. There was the threat of a genuine
toxin being present and this merited further investigation even
if it was a remote possibility.
As the DSP MBA was uninformative
as to the cause of the atypical response, the FSA should have
suspended its use in cockles and banned both harvesting and imports
of cockles into the UK until it was satisfied that there was not
a risk to the public.
The FSA did commission research to
investigate possible reasons for the atypical response other than
toxins but by giving much of this work to CEFAS it meant that
methodological artefact was not investigated as a possible cause.
By the autumn of 2002 the FSA were
receiving strong representations from both industry and from FRS
that solvent carry-over was the most likely cause of the atypical
response. FSA rejected this advice and it took almost a year before
it commissioned a proper investigation of this possibility.
Solvent carry-over was demonstrated
by these investigations. Despite solvent carry-over being a known
cause of artefact in the DSP MBA and causes symptoms similar to
those seen in the atypical response, the FSA continued to maintain
that solvent carry-over was unrelated to the atypical response,
going as far to say that they were satisfied that they had done
enough work to eliminate it as a cause.
The FSA did respond to the findings
of solvent carry-over and an audit of procedures by introducing
a new SOP which produced minimal solvent carry-over. Since the
introduction of this new SOP there have been no atypical positives.
Despite this firm evidence that solvent carry-over was involved
in producing the atypical response, the FSA are still looking
for a novel toxin.
While the FSA as an organisation
has probably acted in good faith throughout this episode, it can
be criticised for the lengthy time taken to properly investigate
the atypical response and for inconsistent reaction to it. Of
particular concern are the public statements made by the FSA to
the press and on their web site. These were often misleading and
were not justified by the scientific information available, suggesting
a lack of objectivity.
The FSA should be subject to an independent
scientific review to highlight lessons from this episode and to
recommend robust scientific working practises to help avoid this
sort of problem in the future.
INTRODUCTION
1. I am a marine biologist with over 20
years professional experience and over 50 scientific publications.
In September 1999 I founded Integrin as the first marine biotechnology
company in the UK. Integrin is focussed on two areas: drug discovery
from marine invertebrates and seafood safety. Our seafood safety
activities include providing testing services for marine toxins;
contract research and consultancy relating to seafood safety and
development of improved methods for the detection of marine toxins.
2. Our customers in seafood safety include
shellfish harvesters, processors and retailers and we undertake
work for major players including Sainsburys. We also do contract
research for the FSA and are currently engaged in a desk study
for FSAS on domoic acid testing in scallops. We pride ourselves
on our independence, integrity and impartiality.
3. Integrin was by shellfish industry to
provide the Committee with our professional opinion on the science
commissioned by the FSA. We have had access to the correspondence
between Industry and the FSA; have discussed the problem with
persons within the FSA and Government laboratories; read the appropriate
reports and discussed the matter at length with overseas experts.
This letter gives a synopsis of our findings.
4. The Committee should be aware that Integrin
were commercially commissioned by Kershaws to produce this piece
of research. However, we could equally have been commissioned
by the FSA to do this and our findings would have been the same.
Kershaws have not asked for any input into the content of this
letter and the opinions within it are solely those of Integrin.
5. Integrin are not intending to give verbal
evidence to the Committee but I would be happy to attend to expand
on any of the points in this letter if the Committee would find
it useful.
THE MOUSE
BIOASSAY AND
QUALITY SYSTEMS
6. The UK government has a statutory responsibility
to test bivalve shellfish for the presence of a number of different
toxins that come (mostly) from marine algae. These produce the
toxins, are eaten by the bivalve molluscs then humans can become
ill when they consume the shellfish. The situation is complex
because of the number and variety of toxins that can be found
in shellfish and research is continually finding new toxins.
7. The main method used to detect shellfish
toxins in Europe is the Mouse Bioassay (MBA). It is not universally
used, even within Europe, and a number of countries (eg Germany,
Sweden and Holland) prefer alternative methods for ethical or
practical reasons.
8. The MBA falls into two main categories. the
PSP (Paralytic Shellfish Poisoning) MBA which looks for hydrophilic
(water loving) toxins (mostly saxitoxins) and the DSP (Diarrhetic
Shellfish Poisoning) MBA which looks for lipophilic (fat loving)
toxins. The atypical response relates solely to the DSP MBA. It
should be noted that the MBA is not used for all shellfish toxins.
For detecting domoic acid a chemical method (HPLC) is used as
the MBA is insufficiently sensitive.
9. The DSP MBA is a very crude assay. It
involves injecting mice with a large volume of extract produced
from shellfish. The extraction method aims to concentrate any
toxins present so they can be detected. Extraction of the toxins
is a crucial step and is very sensitive to changes in either the
shellfish matrix or the extraction conditions.
10. The DSP MBA uses too few mice for any
sort of meaningful statistical analysis to be done. It is not
a quantitative, nor a discriminatory assay; ie the DSP MBA does
not tell us how much of a toxin is present, nor what types of
toxins. The assay is so extreme in terms of its severity that
if one of the three mice used dies then the assay is considered
to have passed. If two or three die then the assay fails. The
assay thus accepts that 33% of the mice used may die by accident.
An important point is that the MBA is an analytical assay designed
to detect a specific group of toxins. It is not a general toxicity
assay for assessing the safety of a foodstuff.
11. Crucially, the DSP MBA as used in the
UK did not involve the use of controls. In any scientific experiment
there must be adequate controls to guard against artefact. The
importance of controls cannot be over-stated. The lack of controls
was very important in the genesis of the atypical story.
12. For example, at Integrin we use chemical
assays to detect DSP toxins. These are quantitative and always
involve the use of both a negative and a positive control. The
positive control is a material which is known to contain a defined
amount of the toxin of interest. This is usually a reference material.
If the assay does not detect the positive control or not at a
level that is considered within the assays parameters then the
test is rejected. A negative control is a material which is known
to lack the toxin of interest. If the negative control produces
a positive reaction then the assay is rejected.
13. Proper validation of the assay is important.
Ideally this should be done on all the matrices that the assay
is to be used on. In practise this can be difficult because of
the time and expense involved. However, if an assay has been validated
on a single matrix (eg mussels) any unusual result coming from
a different matrix (eg cockles) should be subject to a separate
validation to eliminate the possibility of a matrix effect.
14. Another important safeguard is the requirement
of reproducibility. A sample being tested must always give a positive
or a negative result, it must not give both a positive and a negative
result. A lack of reproducibility is a clear indication of a methodological
problem.
15. The last safeguard is external validation.
In short, does another laboratory give similar results to your
laboratory when they test the same material? If they do not then
attention has to be paid to the reasons why.
16. The FSA has been making UKAS accreditation
a formal requirement for laboratories tendering for work from
them. They have also been supporting the development of new toxin
standards, reference materials and ring trials. However, another
crucial element in the atypical history was the relative inexperience
of many of the bodies and individuals involved in shellfish toxins;
the MBA and in running QA systems.
A HISTORY OF
THE ATYPICAL
RESPONSE
17. Through the '90's up till 2001, responsibility
for monitoring shellfish in England, Scotland and Wales rested
with the Fisheries Research Services (FRS) Laboratory in Aberdeen
(and its progenitors).
18. In 2001 the FSA put out for tender four
year contracts for both the Scottish Shellfish monitoring programme
and a similar programme for England and Wales. CEFAS were awarded
the England and Wales contract and FRS the Scottish contract.
DARD maintained the contract to provide testing in Northern Ireland
but introduced the MBA to bring them in line with the other laboratories.
FRS also host the National Reference Laboratory for marine biotoxins.
19. CEFAS commenced the contract in the
Spring of 2001 and almost immediately started having a much higher
incidence of mouse deaths in the DSP MBA than FRS had previously
encountered. The FSA was informed and they instructed local councils
to close affected shellfish areas. Most (though not all) of the
DSP MBA deaths were associated with cockle samples.
20. With time it emerged that cockles from
all areas sampled by CEFAS were producing deaths despite their
wide geographic separation. No seasonality appeared and the deaths
were apparently random. It also emerged that the deaths were unusualrapid,
very severe and quite unlike "typical" DSP MBA deaths.
21. No known UK DSP toxins were found in
the cockles samples, nor were any of the known contaminants (such
as some fatty acids) that cause false positives in the MBA found.
Exotic toxins (eg spirolides) were eventually eliminated as possible
causes.
22. The FSA announced that continued closures
of cockle beds was justified on the grounds of public safety and
that the most likely explanation of the atypical positives was
the presence of a novel neurotoxin (mostly) in cockles.
23. Between the Spring and Autumn of 2002
there was growing scepticism in many quarters that a novel toxin
was responsible. This suspicion was strengthened by the failure
of other European laboratories using the MBA to replicate the
atypical results.
24. Of particular concern to the FSA was
that FRS was not producing atypical responses while CEFAS contaminants
(such as some fatty acids) that cause false positives in the MBA
found. Exotic toxins (eg between the laboratories, the obvious
implication was that the atypical response was an artefact produced
by these differences. To head off this criticism, the FSA commissioned
a comparative study between the laboratories all using the same
cockle samples. This was undertaken in October 2002. FRS found
no atypical positives, while a technician from CEFAS undertaking
the MBA at FRS managed to kill all bar one of the mice used. At
first sight it looked as if DARD and CEFAS obtained the same results
but closer inspection reveals that this is because they are using
a different scoring system. Depending on who is doing the assay,
the same sample could be positive or negative.
25. This was a pivotal moment. FRS reacted
strongly to a report from CEFAS on the experiment (that I have
not seen but presumably said that FRS methods were inadequate).
FRS countered that the lack of an overnight "de-gassing"
in the DARD/CEFAS methods meant that the atypical results could
be due to solvent carry-over; a known reason for the MBA to be
compromised. They also provided some anecdotal evidence that this
was occurring. They also suggested a number of other differences
that could be producing the atypical responses.
26. The FSA was thrown into a degree of
turmoil as evidenced by the summoning of an emergency meeting
of the labs and FSA on the 6 November. Despite FRS having more
experience than CEFAS, DARD or the FSA combined, the "official"
Government line to emerge from this meeting was that the FRS SOP
was failing to extract the toxin and that the methodology used
in the DSP MBA was not the root cause of the atypical response.
All parties were said to have agreed this. FRS, in private at
least, certainly did not agree.
27. The main reason that the FSA rejected
the possibility of methodological artefact appears to be that
they could not square this with the observation that the atypical
problem was mostly associated with cockles. Put simply, if it
was a method problem why were not all species of bivalves producing
the response? And if it was a matrix problem why were CEFAS and
DARD obtaining negatives as well as positives? What they had missed
was that the same sample of cockles was producing both negatives
and positives. a sure indication of a problem with the methodology.
28. Not only would solvent carry-over "in
trace amounts" (Quote from Dr Godfrey Howard, FRS) cause
mouse deaths, it would also explain the lack of an obvious candidate
molecule; the rapid recovery of mice sometimes witnessed and the
apparent loss of activity when samples were stored. Despite FRS
putting up a number of testable hypotheses, experimental work
on solvent carry-over seems to have been put to one side by the
FSA.
29. In December 2002, FSA released a detailed
"updating report" on the atypical response. It gives
a good synopsis of why the FSA rejected ether carry-over as a
factor (it did not consider acetone). We see here the emergence
of a FSA paradigm based on argument rather than experimental science.
This resulted in the "artefact hypothesis" having to
pass a much higher bar of proof with the FSA than the alternative
"novel toxin hypothesis". This is bad science but it
is a crucial factor in explaining FSA's actions.
30. Another contributory factor was the
claim by Prof Yasumoto to have isolated a polar toxin from cockle
samples. This work has not been published so it is impossible
to judge its relevance but it definitely had a major influence
in FSA thinking. However, the DSP MBA is designed to look for
non-polar (lipophilic) molecules and eliminate polar (hydrophilic)
molecules. Polar toxins are checked for by the PSP MBA and no
problems have been found with it. Work by the LGC to isolate and
identify a toxin in the cockle extracts has so far failed to produce
a candidate toxin molecule.
31. The next major development was the attempt
by FSA to introduce a common extraction protocol in June 2003.
This led to immediate atypical-like responses at FRS from mussel
samples. FRS reported solvent carry-over and that the deaths were
consistent with solvent toxicity.
32. FSA Scotland funded an immediate study
by the Macaulay Institute and FRS. This found carry-over of DEE
(ether) but the report was not released until a second and more
detailed study, this time funded by FSA London and undertaken
by CSL (York), to look at solvent carry-over using head space
GC-MS was conducted. This also found considerable solvent carry-over
at all three laboratories.
33. The FSA then claimed that the lack of
statistically significant correlation between solvent carry-over
and the atypical response proved that solvent carry-over was not
responsible for the atypical response. This was a misuse of statistics
(the report gave no indication of whether or not a statistically
significant result could be expected given the very limited data
set) and an inappropriate use of the data as the experiment had
not been designed to look for correlations.
34. The use of head space GC-MS was appropriate
for determining whether solvents were being carried over but,
as was made clear in the Macaulay report, it may not provide a
quantitative assessment of the amount of solvents actually in
the extracts. Also, there must be some sort of cockle "factor"
involved. Without knowing what this is and quantifying it, seeking
simple correlations is a misleading waste of time.
35. Duplication of analyses showed that
the same sample could produce very different levels of solvent
carry-over and that the same sample could also produce both atypical
results and negatives. All of this pointed to the atypical response
being an artefact and the presence of solvent (a known problem)
meant that the obvious conclusion was that solvent carry-over
was at least involved in the atypical response. Despite this,
the summary of the report went as far to say that the FSA was
satisfied that solvent carry-over was not a factor in the atypical
response!
36. The FSA also commissioned Prof Makin
to do an audit of procedures at the different laboratories. He
criticised the lack of controls generally and was severely critical
of QA standards at CEFAS.
37. The FSA web site acknowledged the problem
of solvent carry-over but was emphatic that it was not causing
the atypical response. This was not an accurate reflection of
the scientific information available to the FSA.
38. The Home Office had been unaware of
any solvent carry-over until the release of the report and had
immediate talks with the FSA insisting that solvents be removed.
39. Before a stakeholder meeting due to
take place in October, the FSA released an addendum to their original
report. This was necessary as it transpired that CEFAS had made
mistakes in the experiment so that the data from one week was
not comparable with the other two weeks and had been producing
very high values for solvent carry-over. CSL had produced a new
dataset correcting for this and this had been hurriedly re-analysed.
40. After re-analysis, several of the conclusions
of the previous report were no longer valid. In addition, analysis
of the new dataset by Integrin showed that week three of the CEFAS
data was clearly different from weeks one and two, at least for
acetone. In the new dataset these data had been produced by a
different calculation than for the other two weeks and it was
agreed at the meeting that all the data would be recalculated
using the same method and re-released for analysis. To date this
has not been done.
41. At the stakeholder meeting in October,
I outlined how solvents could be causing the atypical results
either directly or indirectly and predicted that once solvents
were removed the atypicals would disappear.
42. At the end of October a new unified
SOP was introduced at all three laboratories. This was designed
to eliminate solvent carry-over and the presence of solvents was
to be routinely monitored for the first time. Since its introduction
there have been no atypical responses reported from any of the
laboratories. For the same period in 2002 (November and December)
there were 36 atypicals and in the month before the introduction
of the new SOP (October 2003) there were 13.
43. The FSA still does not acknowledge that
solvent carry-over is the underlying cause of the atypical response.
WHAT IS
DIFFERENT ABOUT
COCKLES?
44. It is easy with hindsight to assert
that the atypical problem should never have arisen and certainly
should not have taken as long as it has to resolve. However, what
is clear from reviewing the history of the atypical response is
that virtually all of the normal scientific checks and balances
that exist to prevent us being distracted into wild goose chases
such as this were absent or ignored.
45. These include:
Controls. there were no controls
used in the DSP Mouse Bioassay. This means that there was no way
of ensuring the assay was being conducted properly. It is alarming
to read that FSA officials did not believe that a negative control
was necessary as they were getting negative results. This shows
a lack of understanding of what a negative control is and what
it is for.
Blind studies. an important source
of artefact is operator bias. This is controlled for by blind
studiesthe operator does not know which samples are being
looked at. This is vital in experiments such as the DSP MBA where
subjective endpoints are increasingly used. The only evidence
of the use of blind trials was in the Macaulay report.
Internal reproducibility. samples
must always be positive or negative, not both. This was overlooked
or ignored.
External reproducibility and validation. other
laboratories must be able to reproduce your results. This was
not the case with the atypical response. When other laboratories
could not replicate the atypical result the FSA response was that
these labs were wrong. The correct response is to critically examine
your own procedures with the presumption that these are wrong.
Proper use of hypotheses to guide
experimentation. Basically a hypothesis is an idea that might
explain the phenomena under study. A good hypothesis is testable
by experiment and a good experiment provides a yes/no answer.
Another discipline is that you should design experiments that
will reject your hypothesis rather than try to prove it. The FSA
studies did not conform to these standards. The FSA were prepared
to reject solvent carry-over as a factor without experimentation
to test it and did not test their central hypothesis by carrying
out an experiment to show that the atypical response could occur
in the absence of solvent carry-over.
Proper understanding of fore knowledge
and experience. If we do not learn the lessons of the past
we are condemned to repeat them in the future. Solvent carry-over
was a well-understood problem and should have been investigated
much sooner even if the FSA could not see how this could be squared
with the predominance of cockle samples producing atypicals.
46. The FSA were able to diligently eliminate
a large number of possible interferences and possible toxins.
This left them with their major hypothesis. that the atypical
results were being produced by toxin. The null hypothesis that
should have been pursued with equal vigour was that the atypicals
were the result of some methodological artefact. The Table below
shows that evidence for and against both hypotheses. Clearly there
is more evidence for a methodological artefact and against the
presence of a toxin than the opposites.
Method Problem For
| Method Problem Against | Lipophilic Toxin For
| Lipophilic toxin Against |
Good evidence of solvent carryover (known problem with MBA)
| Problem mostly associated with cockles (why not all shellfish?)
| Problem mostly associated with cockles |
No known algal toxins in samples |
Good evidence of water carryover (known problem with MBA)
| No simple correlation between individual results and solvent concentrations
| New algal toxins are frequently encountered
| No association with known toxic algae |
Association of atypical response with solvent and/or water carryover
| Yasumoto's opinion that atypical response seen by him in cockles not caused by solvents
| | No candidate lipophilic molecule for toxin despite intensive search
|
Reproducibility of atypical response from replicate samples poor
| | | Atypical response seen in cockles from all areas studied but not in every cockle sample
|
Solvent carryover from same samples very variable
| | | No evidence of toxicity in humans or by rat bioassay
|
Similar problems seen by other labs and cured by improvements to solvent carryover.
| | | "toxin" very labile
|
Symptoms consistent with solvent toxicity; solvent abuse, saxitoxin poisoning or anaesthetic overdose
| | | Toxicity profile unlike other lipophilic toxins including known alternatives such as fatty acids (too fast)
|
Toxicity decreases on storage suggesting toxicity associated with volatile compound
| | | Rapid recovery of mice that survive
|
Atypical responses start when new labs initiate DSP MBA
| | | |
MBA protocols between labs confusedperhaps not looking at same thing
| | | |
| | |
|
47. If we focus entirely on the evidence supporting the
FSA position (Method problem against and lipophilic toxin for
columns): Prof Yasumoto has not published his findings so we cannot
comment further on his opinion but it may be a red herring; new
algal toxins are frequently discovered but usually when humans
become sick after eating shellfish; the datasets available do
not allow us to explore whether there are simple correlations
present or not but in any case the complexity of the problem may
mean that simple correlations cannot be found.
This leaves in both columns the observation that the atypical
response is associated primarily with cockles. This is the cockle
conundrum.
48. To resolve the cockle conundrum we need hypotheses
that fit the facts. In particular the suggested mechanisms should
fit the atypical profile:
2. Volatile or extremely labile
3. Rapid recovery from non-fatal attacks (suggests volatility)
4. Always associated with solvents
5. Cockle factor heightens probability of atypical response
49. At the October 2003 stakeholder meeting I proposed
four possible hypotheses of the atypical results. This is not
an exhaustive list, nor does it matter which one (if any) is correct.
50. Hypothesis I: Carry-over of a hydrophilic interfering
factor
A known problem of the DSP MBA is the carry-over into the
mouse of hydrophilic molecules which result in mouse death. The
best understood problem is with saxitoxin and related molecules.
Tiny amounts of saxitoxin will produce an atypical-like response
if injected into mice during the DSP MBA. Poor extraction technique
will allow this to happen.
This hypothesis would fit three of the criteria, including
the cockle factor: cockles live in the sediment and will feed
on bacteria at the surface of the sediment that would not usually
form the diet of mussels. Saxitoxin can be produced by bacteria
as well as algae. However, saxitoxin is not particularly labile
nor would rapid recovery be expected. There is a wide range of
other hydrophilic substances that might also be producing the
response if they are injected in the MBA. Tightening up extraction
procedures should prevent any of these causing the atypical response.
51. Hypothesis II: Toxicity of solvents
DEE and acetone are both toxic, and especially so by IP injection.
There is scant quantitative data on the toxicity of these molecules
and none on their co-toxicity. The mouse deaths could be the result
of toxic effects of the solvents present in the injection. This
explanation would fit all of the criteria except as to why cockles
would be different. It is especially good in explaining the rapid
recovery of some mice.
The Macaulay report suggested a possible model that would
explain the cockle factor: intra/inter species differences in
the water/lipid content could influence the amount of solvent
(particularly DEE) trapped in the extract and thus injected into
the mice. One of the interesting findings of the solvent carry-over
studies was that some solvent could remain in the extracts even
after they had been left overnight. This is difficult to understand
unless there is a mechanism for trapping the solvent.
52. Hypothesis III: Drug effect of solvents
DEE and acetone both have an anaesthetic action and this
will be additive if both are present together. Anaesthetic effects
can occur at levels well below toxic levels but can be just as
deadly. Side effects of anaesthesia include malignant hypothermia;
cardiac arrhythmia and ventricular fibrillation leading to sudden
death. Not only are the side effects of anaesthesia consistent
with the atypical response, vets witnessing it commented on it
resembling the effects of poor induction of anaesthesia. This
hypothesis would explain all of the atypical criteria but like
solvent toxicity requires the invocation of a cockle factor based
around some trapping mechanism.
53. Hypothesis IV: Volatile substance abuse response
Solvents are abused, mostly by young teenagers (glue sniffing)
resulting in fatalities. Over 55% of the fatalities in the UK
are associated with "sudden sniffing death" (the rest
tend to be accidents while intoxicated, choking on vomit etc).
This is manifest by the sniffer being startled and then dropping
dead from a heart attack. Clinical presentation is severe cardiac
arrhythmia then collapse in ventricular fibrillation with death
within 1-2 minutes. This thought to be because the solvents sensitise
the myocardium (heart muscle) to endogenous epinephrine or other
catecholamines leading to arrhythmias.
This is an interesting hypothesis as it fits all of the atypical
criteria including the cockle factor. Shellfish in general tend
to have high levels of chemicals that will act as neurotransmitters
in humans and cockles may have higher amounts than mussels because
of the greater stress on them between harvesting and processing
for the MBA. If present in the injection along with solvents,
these neurotransmitters could produce a mouse equivalent of the
"sudden sniffing death".
54. The symptoms witnessed in the atypical response seem
to fit a combination of anaesthetic and cardiac arrest. Working
out exactly what is going on would not be a trivial task and probably
unethical. The importance of these hypotheses is, however, to
illustrate that explaining the atypical response does not require
the invocation of an elusive novel toxin. These hypotheses depend
on either solvent carry-over or carry-over of water-soluble fractions.
Eliminating these should eliminate the atypicals. This is what
has happened with the introduction of the new SOP in late October.
55. This is not scientific proof that the atypical response
is produced by solvent carry-over as there is no control in the
experiment. The new SOP introduced a number of other changes which
may also have influenced the atypical response. Nevertheless,
the fact that the atypicals have disappeared after the new SOP
was introduced demonstrates that they are the result of a methodological
artefact. Of the possibilities, solvent carry-over remains the
most probable cause of the atypical responses, either directly
or indirectly.
CONCLUSIONS
56. The nature of the mouse bioassay was a fundamental
reason for the genesis of the atypical episode. The lack of controls
was especially important. Without the normal QA mechanisms to
guard against artefact, experience was the only defence. FRS had
this experience but the other laboratories did not draw on it
when setting up their DSP MBA programmes and so were deceived
into thinking that an artefact was a real result. It is possible
that inter-laboratory rivalry played a part here, with the new
labs either consciously or unconsciously not wanting to confront
the possibility that they were incompetent in running the assay.
57. The nature of the MBA, which involves mice dying,
tends to dramatise the result. However, the assay is very subject
to a "garbage ingarbage out" scenario. Mice deaths
do not necessarily indicate the possibility of human harm and
the possibility of artefact must be eliminated before over-reacting
to assay deaths.
58. The FSA were not the first regulatory authority to
have been foxed by an inappropriate response by the MBA but they
were still correct to take the possibility of a new toxin seriously.
Previous experience (such as the domoic acid poisonings in Canada)
showed that we have to be alert to novel toxins. The FSA response
was, however, inconsistent. If there was genuine concern that
a health problem was indicated, the FSA should have shut all cockle
beds and banned imports. This would have caused a lot of fuss
and triggered European Commission interest but it would have shown
consistency. It would also have speeded up finding a solution.
59. The FSA were not guilty of devoting insufficient
resources to the problem. There was a lot of good work done to
eliminate possible interfering factors and known toxins. Where
they did fall down is in not having a structured response that
was designed to tackle the question of possible artefact. In particular
the decision to fund CEFAS (without tendering) to investigate
possible causes of the atypical response when CEFAS was the main
originator of these results was a mistake. This meant that the
question of artefact was not likely to be properly tackled.
60. FSA was also too reliant on presumption and argument
in attempting to unravel the cause of the atypical response. This
had disturbing resonance with the original MAFF response to finding
scrapie-like disease in cattle where presumption and inter-lab
rivalry led to a delay of a crucial year in recognising what was
going on. Presumption is no substitute for experiments.
61. FSA did seek help from international experts but
they would have assumed that the MBA was being performed correctly
as this should have been the first thing checked. I have sympathy
with the FSA in regard to Prof Yasumoto's input. An international
expert is telling the FSA that he thinks there is a toxin present
and he may have isolated one.
62. After October 2002, the FSA's position and conduct
of the investigations is much more open to criticism. The comparative
study between the laboratories showed a lack of reproducibility
in the atypical response. FSA should have paid more attention
to the concerns of FRS as they had the most experience in this
area. FSA failed to follow up FRS's suggestions to look at methodological
artefact for nine months and then only returned to it when more
or less forced to by the actions of FRS.
63. Had artefact been tackled seriously from the beginning
then it would have taken no more than six months (and probably
far less) for the FSA to have resolved the situation to their
satisfaction that there was no risk to public health.
64. When solvent carry-over was established, the MBA
should have been withdrawn until a new SOP that eliminated solvent
carry-over was established and validated. FSA should have concluded
at this point that solvent carry-over was the most likely cause
of the atypical response. That they did not is worrying as it
suggests a loss of objectivity.
65. Also worrying were the public pronouncements of the
FSA. Statements were made that a toxin existed when there was
no direct evidence for a toxin being present in cockles. The link
with solvent carry-over was repeatedly dismissed but the evidence,
as opposed to FSA opinion, increasingly pointed to solvent carry-over
as the cause. Public presentation of complex scientific issues
is always a tricky art but it is crucial that FSA properly represent
the science, including areas of doubt.
66. The UK needs a strong, independent food safety body.
It should be fearless in protecting public safety and should err
on the side of caution. Food safety issues tend to be complex
and mistakes will occur. What the atypical episode seems to indicate
is that there was a lack of a structured response to emerging
issues. This leads to inconsistency and a poor rate of progress
to actually solving the problem.
67. I would recommend that an independent scientific
enquiry team be brought together to examine the atypical episode
with a remit to identify areas of the FSA approach that are inconsistent
with good scientific practise and to recommend codes of practise
and procedure that will give the FSA the tools it requires to
fulfil its role in protecting the UK population.
Dr J Douglas McKenzie
Managing Director
December 2003
|