Opinion on the results of the Risk Assessment of: Hydrogen Fluoride [HF], CAS N° : 7664-39-3, EINECS N°: 231-634-8 carried out in the framework of Council Regulation (EEC) 793/93 on the evaluation and control of the risks of existing substances - Opinion expressed at the 15th CSTEE plenary meeting, Brussels, 5th of May 2000.
Terms of reference
The CSTEE on the basis
of the examination of each Risk
Assessment Report is invited to
examine the following issues:
1. Does the CSTEE agree with
the conclusions of each Risk
Assessment Report?
2. If the CSTEE
disagrees with such
conclusions, the CSTEE is
invited to elaborate on the
reasons for this divergence of
opinion.
GENERAL COMMENTS
ENVIRONMENT
The CSTEE concludes that
the available information is
sufficient and allows an
assessment of the environmental
risk, in spite of some gaps.
The CSTEE also agrees with the
conclusions made.
The RAR only includes
the assessment for the local
scenario on the basis that
fluoride emissions, not related
to the production and use of
hydrogen fluoride, designated
"other", are expected to be the
most relevant emissions. These
additional emissions cannot be
evaluated with the information
provided for the assessment and
therefore the decision of the
authors of the report is
justified. The CSTEE agrees
with the report that a proper
in-depth risk assessment of the
environmental emission of
fluorides, considering all
potential sources, is required.
The consequences of
omitting to model regional and
continental exposures (PEC)
could have been discussed
further, especially when
earlier it is stated that
naturally high background
levels can be significant and
should be accounted for.
HUMAN HEALTH
THE CSTEE agrees with
the conclusions made in the
results section. It supports
the request to make available
the full test report on
reproduction. However, the
CSTEE also asks for further
valid volunteer studies for
deriving the NOAEL for
irritation after acute and
repeated exposure.
In general, the Risk
Assessment Report is well
written and gives a good
documentation of the database.
The CSTEE indicates, however,
that the SCOEL (Scientific
Committee on Occupational
Exposure Limits) recently
published a documentation and
proposed an OEL using
additional data on local and
systemic effects which have not
been considered in the RAR
(SEG/SUM/56F).
SPECIFIC COMMENTS
Environment
Exposure assessment
The absence of import
and export data can undermine
the basis of modelling at
local, regional and continental
levels. In the absence of such
data being provided it may be
possible to consider a
worst-case scenario.
The local exposure
assessment follows the
recommendations of the TGD. The
measured emission and
monitoring data are considered
in the assessment.
The approach to
determining PEC for non-river
waters, variously called
"estuary", "ocean", "sea",
"tidal estuary (lagoon)" seems
reasonable although ad-hoc in
the absence of TGD methods.
However, the terminology at
least should be defined and
standardised as there may be
significant differences between
a lagoon, estuary, open sea and
ocean, and the tidal flushing
in each can vary greatly
between localities (i.e. tidal
range varies from cm to m).
Information on
partitioning at different pH is
mentioned but the implications
of this for ecotoxicology are
not discussed. Why the
'bioavailability of fluorides
in sediments is expected to be
low' (pg.43) is not explained.
For the terrestrial
(soil) and atmospheric
estimations, and taking into
account that the risk
assessment is restricted to the
vicinity of the emission
sources, a higher tier
assessment, if possible, should
be recommended, i.e. estimating
the PEC as a function of the
distance to the emission source
instead as a single value. The
raw information is not
available to the CSTEE and
therefore the Committee cannot
evaluate whether or not this
higher tier assessment is
possible with the available
information.
Effect assessment: Aquatic
compartment.
The assessment is based
on a relatively large set of
data on aquatic organisms
belonging to all trophic levels
of the aquatic ecosystem (fish,
invertebrates, algae,
bacteria). Short and long term
toxicity data are available for
freshwater fish and
invertebrates but not for
marine animals. In some cases,
test on soft and hard water,
relevant for this kind of
chemical, are available. Many
data are produced with
analytically controlled
concentrations.
It can be concluded that
the information available is
suitable for a sound set up of
PNEC for water. Calculated PNEC
values (0.9 and 0.4 mg/L for
hard and soft water
respectively) are close to
background levels measured in
many natural water bodies and
are lower than natural
concentrations measured in
catchement areas rich on
fluorine. Thus, PNEC for water
can be considered as reliable
and no more information is
required. Taking into account
that, on a local basis, the
PEC/PNEC ratio is often higher
than 1, conclusion (iii) for
the aquatic environment is
justified if exposure is
correctly assessed.
A low concern for
treatment plants is also
justified.
Effect assessment:
Terrestrial
compartment
Following the TGD
recommendations, the
terrestrial ecosystem is
evaluated separately in
different compartments:
terrestrial (soil) compartment,
atmosphere, secondary
poisoning.
Soil compartment
The effect assessment
for the terrestrial (soil)
environment is based on
toxicity data on soil
organisms. The CSTEE recognises
the efforts of the authors of
the report to expand the
information submitted with the
HEDSET in order to conduct a
proper risk assessment.
The toxicity for plants
is limited to a non standard
assay of difficult
interpretation. Therefore the
CSTEE suggests to consider a
requirement on additional
toxicity tests on terrestrial
plants exposed via soil.
The estimation of the
PNEC for the terrestrial
compartment according to the
TGD produces a value that is
considered in the report as
negligible when compared to the
background concentration. The
ecotoxicological meaning of the
word "negligible" for a PNEC is
not clear. However, in some
situations low incremental
increases above high background
concentrations could produce
significant effects. The tests
on soil microbial activity
validated in the RAR were
presumably conducted with
natural soils with natural
fluoride background
concentrations. If this
assumption is right, the
results could suggest that low
increases in the background
concentration can produce
significant effects.
The comparison of model
PECsoil estimations with
toxicity data could support the
report conclusion on potential
low risk for soil dwelling
organisms. However, monitoring
data seems to be several orders
of magnitude higher than model
estimations and therefore a
clarification is required.
More information on the
expected exposure levels and on
the toxicity of fluoride for
soil dwelling organisms and the
effects of soil characteristics
and background concentrations
on this toxicity should be
required for a proper
assessment before reaching
conclusion ii).
Atmospheric compartment
As expected, the effect
assessment for the atmosphere
is mostly restricted to the
toxicity of plants, a single
study on invertebrates is
included. The assessment for
plants covers a compilation of
NOECs proposed by different
institutions from The
Netherlands and Germany. These
NOECs are not direct results of
toxicity assays but estimations
on concentrations expected to
be safe for plants. However,
the scientific basis of these
proposals are not included, and
therefore the CSTEE cannot
produce an opinion on the
acceptability of these values.
The lowest value is
selected for the PNEC
derivation without the use of
any additional extrapolation
factor. The rationale for not
to apply an additional
extrapolation factor is
acceptable. In fact the CSTEE
has already recommended in a
previous opinion on the
derivation of environmental
thresholds for ozone that no
additional extrapolation
factors are required for the
protection of vegetation when
the no-observed effect
concentration was obtained from
ecologically relevant effects
observed for the most sensitive
plant species in long-term
field studies. However, because
the criteria for the selection
of these NOEC values are not
presented, the CSTEE cannot
produce an opinion on the
acceptability of this PNEC.
Secondary poisoning
The effect assessment
for secondary poisoning (non
compartment specific effects
relevant to the food chain)
includes assessments for
invertebrates, livestock and
wild terrestrial vertebrates
(mammals and birds).
On pg. 10 the report
cites a RIVM study that
concluded on "limited data"
that "accumulation through food
chains (biomagnification) is of
little significance in the
aquatic environment". In
contrast to the above statement
biomagnification has been found
in terrestrial invertebrates in
another RIVM report. However,
it is not possible to assess
the validity of these studies.
Several effects are
described but in reality these
effects are not considered in
the final risk characterisation
that is done on the basis of
the RIVM NOECs for livestock
calculated for the grazing and
winter seasons assuming a
maximum acceptable intake of 55
mg F/kg dry weight in feed.
The basis for the
derivation of the Dutch
drinking water limit
concentration for ruminants and
poultry values has not been
included and therefore the
CSTEE cannot produce an opinion
on the scientific validity of
these proposals or on the
acceptability of these Dutch
values for the conditions on
other parts of Europe. However,
the final risk characterisation
concludes that a potential risk
for livestock has been
identified, and that the risk
can be even higher for wild
species.
The CSTEE recognises
that with the available
information it is difficult to
conduct a proper risk
assessment. Nevertheless, the
direct comparisons of
concentrations in food
producing fluorosis, and of
NOELs from experimental
bioassays corrected for daily
food intake with the levels of
fluoride reported near some
industries, suggest that the
expected concentrations should
be in the same range or only
slightly lower than those
producing toxicity on the
exposed animals. Therefore, the
conclusion of the risk
assessment report on a
potential risk for terrestrial
vertebrates feeding on
contaminated food items is
considered acceptable.
Human Health
1. Local effects of HF
(irritation)
The studies performed
regarding local effects of HF
(irritation) are of limited
quality and do not allow
setting of a reliable NOAEL.
Therefore, further valid
volunteer studies concerning
irritation after acute and
repeated exposure to HF are
needed and conclusion (i) for
workers should also address
this.
The LOAEL for local
effects after short-term
exposure was derived in the
Risk Assessment Report from a
volunteer study by Lund et al.
(1997) with 1-h exposure
duration. Scores for irritation
were increased significantly
only at the high concentration
range (>2.5 mg/m³). This
conclusion is justified.
However, due to the fact that
no volunteers were used as
controls, it is not possible to
evaluate whether the effects
observed in the lower exposure
groups are HF specific or
unspecific effects due to
exposure conditions. Therefore,
further valid studies are
needed to cover the
concentrations below 2.5 mg/m³.
The recent study by Lund
et al. (1999) should be added.
From this study, performed
under the same exposure
conditions as Lund et al.
(1997), subclinical effects
(e.g. increase in CD3 positive
cells) occur at a dose range
between 0.7 and 2.4 mg/m³ and
above. However, again no
control exposures were
performed and a further study
would be necessary to identify
the concentration of the onset
of this effect more precisely.
The LOAEL for local
effects after repeated exposure
was derived from a volunteer
study performed in the 1960s.
Because of "somewhat
contradictory results" reported
in these publications (Largent
1960: discomfort observed in
volunteers exposed from 0.74 to
1.64 mg/m³; Largent 1961: 1.16
mg/m³ tolerated without
noticeable effects) the
concentration of 1.16 mg/m³ was
used, for safety sake, as LOAEL
in the Risk Assessment Report.
However, Largent 1960 reported
no detailed information about
the onset of irritation effects
at specific exposure
concentrations. Only ranges
were presented. In the second
report, Largent 1961 stated
more precisely that 1.16 mg/m³
were tolerated without
noticeable effects. Therefore,
it is not to agree with the
Risk Assessment Report that the
results of the two studies are
"somewhat contradictory" and
there is no reason not to
accept that 1.16 mg/m³ were
without any noticeable effects.
However, it is very important
to address the limited validity
of this study and to conclude
the need for further valid well
performed studies.
2. Systemic effects
The NOEL for systemic
effects after repeated exposure
of 0.48 mg/m³ is derived by the
study of Chan-Yueng et al.
1983. This NOEL is not
sufficiently supported because
other relevant information are
either not or insufficiently
considered (see "Specific
remarks"). Considering those, a
NOEL of 2 mg/m³ is
scientifically supported (See
SEG/SUM/56F).
The following studies
have not been evaluated. They
provide additional information
for identifying NOAEL and
LOAEL:
Chan-Yueng et al.
(1983): NOAEL: 0.48 mg/m³
Ehrnebo and Ekstrand
(1986): NOAEL: 0.9 mg/m³ / 50
ng/ml plasma (post-shift)
Dinman et al. (1979a,b):
NOAEL: 2.0 mg/m³ / 8 mg/l urine
Kaltreider et al.
(1972): NOAEL: 3 mg/I urine
LOAEL: 2.4 mg/m³ / 8.7
mg/l urine
From these studies, the
SCOEL identified an NOAEL for
inhalation exposure and adopted
an OEL of 2.0 mg/m³ (1.5 mg/m3)
and an LOAEL 2.4 mg/m³. In the
Risk Assessment Report a
previous proposal for OEL is
cited.
There is indication that
humans are more susceptible to
fluorosis than animals.
Therefore, human studies should
be preferred over animal
studies: there is no need to
use the NOAEL of 0.72 mg/m³
derived from a 91-day
inhalation study with rats for
risk assessment purposes.
Furthermore, there is no
scientific agreement on the
factors used for duration
correction ("duration corrected
value", DCV).
For the onset of
systemic effects (e.g.
fluorosis), the internal
exposure resulting from
background and additional
exposures (e.g. occupational)
is critical. This internal
exposure can be monitored
easily by determining fluorine
in plasma or urine. Based on
such data a NOEL of <8 mg/l
urine has been proposed by the
SCOEL, which corresponds to the
OEL of 2 ppm (1.5 mg/m³). The
LOEL is about 9 mg fluorine/l
urine.
Final General Remarks
Assessment factors are
applied to derive conclusions
for workers risk assessment by
giving a "minimal MOS". This
approach is not generally
accepted. In addition, deriving
a "duration corrected value"
(DCV) from animal experiments
is not a generally accepted
approach (see also "specific
remarks: Animal studies"). The
use of assessment factors has
already been criticised by the
CSTEE in the Risk Assessment
Reports of the glycolethers
2-(2-methoxy- and
2-(2-butoxyethoxy)ethanol.
Although there is
sufficient information from
human studies, a factor of 100
was used in the Risk Assessment
Report to meet a sufficient MOS
for man exposed indirectly via
the environment. In this
situation it seems warranted to
use a factor of 10 to
extrapolate to the general
population.
References
Dinman et al. (1976a) A
15-year retrospective study of
fluoride excretion and bone
radio-opacity among aluminium
smelter workers - Pt.4. J Occup
Med 18: 21-23
Dinman et al. (1976b)
Absorption and excretion of
fluoride immediately after
exposure - Pt. 1. J Occup Med
18: 7-13
Ehrnebo and Ekstrand
(1986) Occupational fluoride
exposure and plasma fluoride
levels in man. Int Arch Occup
Environ Health 58: 179-190
Kaltreider et al. (1972)
Health survey of aluminium
workers with special reference
to fluoride exposure. J Occup
Med 14: 531-541
Lund et al. (1999) Increased
CD3 positive cells in
bronchoalveolar lavage fluid
after hydrogen fluoride
inhalation. Scand J Work
Environ Health 25:
326-334