The Long-term Effects of Water Fluoridation on the Human Skeleton

Abstract

Municipal water fluoridation has notably reduced the incidence of dental caries and is widely considered a public health success. However, ingested fluoride is sequestered into bone, as well as teeth, and data on the long-term effect of exposure to these very low doses of fluoride remain inconclusive. Epidemiological studies suggest that effects of fluoride on bone are minimal. We hypothesized that the direct measurement of bone tissue from individuals residing in municipalities with and without fluoridated water would reveal a relationship between fluoride content and structural or mechanical properties of bone. However, consonant with the epidemiological data, only a weak relationship among fluoride exposure, accumulated fluoride, and the physical characteristics of bone was observed. Analysis of our data suggests that the variability in heterogenous urban populations may be too high for the effects, if any, of low-level fluoride administration on skeletal tissue to be discerned.

Full text

1219
RESEARCH REPORTS
Clinical
DOI: 10.1177/0022034510376070
Received April 20, 2009; Last revision May 13, 2010;
Accepted May 19, 2010
A supplemental appendix to this article is published elec-
tronically only at http://jdr.sagepub.com/supplemental.
© International & American Associations for Dental Research
D. Chachra6, H. Limeback5,
T.L. Willett3, and M.D. Grynpas1—4*
1Institute of Biomaterials and Biomedical Engineering,
University of Toronto, ON, Canada; 2Department of Materials
Science and Engineering, University of Toronto, ON, Canada;
3Samuel Lunenfeld Research Institute, Mount Sinai Hospital,
Room 840, 600 University Avenue, Toronto, ON, Canada
M5G 1X5; 4Department of Laboratory Medicine and
Pathobiology, University of Toronto, ON, Canada; 5Department
of Preventive Dentistry, University of Toronto, ON, Canada;
and 6Franklin W. Olin College of Engineering, Needham, MA,
USA; *corresponding author, grynpas@mshri.on.ca
J Dent Res 89(11):1219-1223, 2010
ABSTRACT
Municipal water fluoridation has notably reduced
the incidence of dental caries and is widely consid-
ered a public health success. However, ingested
fluoride is sequestered into bone, as well as teeth,
and data on the long-term effect of exposure to
these very low doses of fluoride remain inconclu-
sive. Epidemiological studies suggest that effects
of fluoride on bone are minimal. We hypothesized
that the direct measurement of bone tissue from
individuals residing in municipalities with and
without fluoridated water would reveal a relation-
ship between fluoride content and structural or
mechanical properties of bone. However, conso-
nant with the epidemiological data, only a weak
relationship among fluoride exposure, accumu-
lated fluoride, and the physical characteristics of
bone was observed. Analysis of our data suggests
that the variability in heterogenous urban popula-
tions may be too high for the effects, if any, of
low-level fluoride administration on skeletal tissue
to be discerned.
KEY WORDS: bone, fluoride, biomechanics,
mineralization, public health.
INTRODUCTION
The reduction of dental caries via the fluoridation of water supplies at 1
ppm is widely considered to be a public health success (McDonagh
et al., 2000). However, ingested fluoride is also incorporated into bone during
bone formation and remodeling (Whitford, 1989). One concern is that it may
alter bone mechanical properties (Mousny et al., 2006), which may present
clinically an altered risk of bone fracture. The mechanical properties of bone
result from the composition and properties of the bone material as well as
the amount and structure of the bone present, all of which can be affected by
fluoride. The response of bone to fluoride is complex and dose-dependent,
engaging different mechanisms at low, medium, and high doses (Boivin and
Meunier, 1990; Grynpas, 1990; Turner et al., 1993; Yan et al., 2007; Mousny
et al., 2008). Multi-decade exposure to environmental fluoride (~1 mg/day)
therefore cannot be modeled in animals or humans by using higher doses and
shorter times. Epidemiological techniques have been used to investigate the
association of fracture risks with fluoride exposure at these low levels (Allolio
and Lehmann, 1999); the results are generally inconclusive, with the excep-
tion of a study which found increased rates of fracture only for very low or
very high exposure (Li et al., 2001).
Here, we compare tissue-level data from bone specimens from a fluori-
dated region (Toronto) vs. a non-fluoridated region (Montreal). Compressive
mechanical testing of specimens was used as a proxy for fracture risk (Turner
and Burr, 1993). We sought to complement existing epidemiological findings
by examining bone samples from these populations directly, with the aim of
determining if a relationship existed between the physical properties of the
bone and the fluoride content. Data for samples from the two regions, as well
as bone samples in the highest and lowest quartiles of fluoride content, were
also compared. We hypothesized that the direct measurement of bone tissue
from individuals residing in municipalities with and without fluoridated water
would reveal a relationship between fluoride content and structural or
mechanical properties of bone.
MATERIALS & METHODS
Specimen Collection
Femoral heads were collected from patients undergoing total hip arthroplasty
at two hospitals in Canada, Mount Sinai Hospital in Toronto and the Jewish
General Hospital in Montreal, between September 1996 and August 2000.
The Long-term Effects of
Water Fluoridation on
the Human Skeleton

1220 Chachra et al. J Dent Res 89(11) 2010
Municipal water supplies in Toronto have been fluoridated at 1
ppm for more than four decades; Montreal has never had fluori-
dated water. The use of human study participants was approved
by the institutional review board of the University of Toronto.
Informed consent was obtained from all patients.
The femoral heads were stored at -70°C prior to being tested.
A cylinder of bone, approximately 6 mm in diameter and 6 mm
long, was excised from the center of each femoral head, cleaned
and weighed, and tested in compression (see below), after which
its fluoride content was determined by neutron activation analy-
sis (Mernagh et al., 1977). Three blocks of bone were excised
from the inferior (non-loaded) surface of each femoral head and
embedded in Spurr resin. One block was approximately 15 x 15
x 5 mm and was used for determination of mineralization by
backscattered electron imaging. Two smaller cubes (approxi-
mately 5 mm/side) were excised from a point near the apex of
the head (inferoproximal) and toward the shaft of the femur
(inferodistal) and used for microhardness testing. The exposed
faces were polished to a 0.01-µm finish.
Mechanical Testing
The dimensions of each cancellous core were measured with a
micrometer (together with the mass, these numbers were used to
determine the density), and the sample was then tested in uncon-
fined compression at 1 mm/min in a universal testing machine
(Instron 1011 or 4465, Instron Corp., Canton, MA, USA;
LabVIEW, National Instruments Corp., Austin, TX, USA) until
failure occurred. The compressive modulus, yield stress, ulti-
mate compressive stress, strain at ultimate compressive stress,
energy to failure, and energy to yield were determined (Turner
and Burr, 1993).
Microhardness Testing
Microhardness measurements were conducted on the embedded
bone samples by means of a hardness tester equipped with a
Knoop diamond indenter (HM-122, Mitutoya, Aurora, IL, USA).
Each indentation was made under a load of 25 g with a duration
Table 1. Information on Patients and Bone Samples, by Region
Toronto (fluoridated)
Montreal (non-
fluoridated)
Fluoride content (ppm)
Mean ± SD 1030 ± 60* 643 ± 35*
Range 192–2264 270–1200
Age of donors (yrs)
Mean ± SD 66 ± 11* 70 ± 13*
Gender
Male 26 15
Female 27 24
Disease state 47 osteoarthritis 28 osteoarthritis
2 osteoporosis 7 osteoporosis
1 rheumatoid arthritis 2 rheumatoid arthritis
2 avascular necrosis 1 ankylosing necrosis
1 osteonecrosis 1 psoriatric arthritis
of 10 sec. Ten indentations were made in the subchondral bone,
equally spaced along the width of the specimen. A further 10
indentations were made at random locations in the trabecular
bone of the specimen. The Knoop hardness (KH) was calculated
from the length of the indentations by software in the test system.
Backscattered Electron Imaging
We used backscattered electron imaging (Grynpas et al., 1994) to
quantify bone mineralization on the embedded bone samples
(coronal face). The samples were imaged by scanning electron
microscopy (Hitachi S-2500, Nissei Sangyo America Ltd.,
Mountain View, CA, USA) and a backscattered electron detector
(Link Tetra, Oxford Instruments, Abingdon, UK). We analyzed
the image of the bone by dividing the grayscale range of the bone
image into ‘bins’ (7 for the cancellous bone and 8 for the subchon-
dral bone) and determined the percentage of the image that was at
each level, producing a profile of the mineralization of the bone.
A weighted average of the mineralization was then calculated:
Statistical Analysis
Statistical tests (t tests and linear regressions) were performed
with Sigmastat (Systat Software Inc., San Jose, CA, USA). We
used heteroscedastic or homoscedastic t tests, as appropriate, to
identify differences between groups. Statistical significance is
reported if p < 0.05.
RESULTS
Sample Information
Information about patients and sample characteristics can be
found in Table 1. In total, 92 femoral heads were collected: 53
samples from patients residing in the Toronto area (mean age ±
SD: 66 ± 11 yrs), and the remaining 39 from Montreal residents
(70 ± 13 yrs).
Fluoridated vs. Non-fluoridated Region
Fluoride Content
The fluoride content of bone from individuals residing in
Toronto was significantly higher (p < 0.0001) than that of those
from Montreal. Note, however, that the range for the Toronto
bones fully subsumed the range of the Montreal bones (Fig. 1).
Compressive Mechanical Properties
The mean density of cancellous cores from the Toronto speci-
mens was significantly greater than that of those from Montreal
(p < 0.05). However, the density of cancellous cores in this
study did not correlate closely with either the fluoride content or
the age (data not shown). The mean strain at ultimate compres-
sive stress (UCS) of bone from the Toronto donors was greater
than that of their Montreal counterparts, as was the energy
absorbed to failure (p < 0.05) (Table 2).
WA nxn
n
or
=
=
∑
1
7 8
.

J Dent Res 89(11) 2010 Water Fluoridation and Bone Health 1221
Mineralization
No significant differences were observed in the degree of min-
eralization of the bone between the two regions, by BSE imag-
ing (data not shown). At the inferoproximal (apex) site, the
microhardness of the subchondral and of the cancellous bone
was greater for specimens from Toronto than from Montreal
(subchondral KH, 43.7 ± 1.1 vs. 38.8 ± 1.5; cancellous KH,
44.3 ± 1.1 vs. 39.8 ± 0.9; p < 0.05). No differences were
observed at the inferodistal site.
Comparison of Properties by Fluoride Content
Description of Quartiles
We used the fluoride content to identify bone samples in the top
and bottom quartiles (23 samples each), with mean fluoride
contents of 1434 ± 70 and 449 ± 25 ppm, respectively (ranges:
1082–2264 and 192–582 ppm). In the top quartile, 21/23 sam-
ples were from the fluoridated region. However, more than a
quarter (6/23) of the samples in the bottom quartile were also
from the fluoridated region. The patients in the top quartile were
older than those in the bottom (70 ± 11 vs. 62 ± 14 yrs of age;
p < 0.05), consistent with an increase in fluoride accumulation
with age (Richards et al., 1994; Chachra, 2001).
Compressive Mechanical Properties
In contrast to the comparisons by city, the density of the can-
cellous core was unchanged between the quartiles. Despite
this, the yield stress and the ultimate compressive stress were
greater for the bottom quartile than for the top quartile (yield
stress, 5.4 ± 0.8 vs. 7.5 ± 0.6 MPa; UCS, 6.0 ± 0.9 MPa vs. 8.4
± 0.6 MPa). No other differences were observed between the
quartiles.
Figure 1. Fluoride content of bone samples from Toronto and Montreal.
The error bars indicate standard deviations. The mean fluoride content
of bone samples from Toronto (n = 53) residents was higher (p <
0.0001) than that of those from Montreal (n = 39) residents. Note,
however, that the range of fluoride contents in the non-fluoridated
region is completely subsumed by those in the fluoridated region.
Table 2. Mechanical Properties of Bone Samples, by Region (mean ±
SEM; *p < 0.05)
Toronto
(fluoridated)
Montreal (non-
fluoridated)
Density (g/cm3) 0.90 ± 0.04* 0.75 ± 0.05*
Compressive modulus (MPa) 266 ± 21 232 ± 21
Yield stress (MPa) 7.3 ± 0.6 6.6 ± 0.5
Energy to yield (MJ/m3) 0.14 ± 0.02 0.14 ± 0.02
Ultimate compressive stress
(MPa)
8.3 ± 0.7 7.3 ± 0.6
Energy to failure (MJ/m3) 0.33 ± 0.06* 0.21 ± 0.02*
Strain at ultimate compressive
stress (%)
7.9 ± 0.3* 6.9 ± 0.3*
The mean density of the cancellous cores was greater for the Toronto
(n = 53) specimens than for the Montreal (n = 39) specimens. In
compression, the strain at failure and the energy absorbed to
failure were significantly increased in the Toronto specimens com-
pared with their Montreal counterparts. The microhardness values
of both the subchondral and cancellous regions of bone were also
greater for the Toronto samples compared with the Montreal
samples.
Mineralization
There was no difference in the degree of mineralization between
the two quartiles, as measured by BSE imaging (data not
shown). However, the top quartile had consistently higher
microhardness than the bottom quartile, and significant differ-
ences were observed for two of the four sites: subchondral bone
at the inferoproximal site (KH 45.4 ± 1.7 vs. 36.9 ± 0.2; p <
0.05) and cancellous bone at the inferodistal site (KH 41.9 ± 0.5
vs. 38.4 ± 0.8; p < 0.05).
Variability in the Data
A plot of the ultimate compressive stress as a function of fluoride
content suggests that there is a weak negative relationship between
them (Fig. 2). Note the variability, however: The fluoride concen-
tration accounts for less than 5% of the scatter in the data. In addi-
tion, the fluoride content increases with age, and the ultimate
compressive stress decreases with age, which further suggests that
any relationship between ultimate compressive stress and fluoride
may be an artifact of these other relationships (Chachra, 2001).
DISCUSSION
Epidemiological studies have failed to observe an effect of
municipally fluoridated drinking water on bone (McDonagh
et al., 2000), but the safety of long-term water fluoridation remains
uncertain in public discussions. In this study, we measured the
physical properties and fluoride content of the bone samples
directly. We then assessed the effect of water fluoridation in
three different ways: (i) a comparison of samples from residents
of municipalities with fluoridated (Toronto) and non-fluoridated
(Montreal) water (this is analogous to a retrospective cross-
sectional epidemiological study); (ii) a comparison of bone

1222 Chachra et al. J Dent Res 89(11) 2010
samples from the upper and lower quartiles of fluoride content;
and (iii) a comparison of the physical properties with the fluo-
ride content itself, treated as an independent and continuous
variable.
A striking finding of this study was the lack of a strong rela-
tionship between fluoride exposure and bone fluoride content.
This manifested as the wide range of bone fluoride content in
the specimens from the fluoridated municipality which, in turn,
entirely subsumed the observed range for samples from the non-
fluoridated municipality. This approach of comparing samples
(or, in the case of epidemiological studies, the fracture rates)
from two cities may therefore not be able to differentiate
between two populations on the basis of fluoride exposure,
whether a result of different patient histories (residency, diet) or
due to the wide variability in responses to fluoride ingestion (see
below). These ambiguous findings from the comparison of
municipalities suggested the more direct approach of comparing
the upper and lower quartiles by fluoride content.
The differences observed between quartiles are in contrast to
the differences observed between cities. Between quartiles, the
density is unaltered, but the strength of the bone is lower for the
more fluoridated group, which is consistent with some previous
animal studies (Mousny et al., 2006). Between cities, the density
is greater for the bones from the region with municipal fluorida-
tion, but the strength of the bone is unchanged, while the strain
at the ultimate compressive stress (UCS) and the energy
absorbed to failure are greater. Because the energy absorbed to
yield was identical in the two groups, this suggests that the dif-
ference in energy absorption is a consequence of the post-yield
behavior; the greater strain at UCS from the fluoridated samples
Figure 2. Relationship between ultimate compressive stress and
fluoride content. The ultimate compressive stress, as well as the yield
stress, declined with increasing fluoride content of the cancellous
core (n = 92; R2 = 0.048, p < 0.05). However, less than 5% of
the variation can be attributed to the fluoride concentration. As well,
the fluoride content also increases with age (see Appendix), so this
observed decline is likely to be at least partially attributable to
increasing age.
results in a greater energy absorption to failure, which means
that these bones may be more ductile and tough. This may be a
consequence of an effect of fluoride on the interface between the
mineral and organic phases (Kindt et al., 2008; Mousny et al.,
2008; Thurner et al., 2009).
Most importantly, the extremely wide variability in proper-
ties makes it difficult to point definitively to a fluoride-related
effect. The data presented here show a wide variation in fluoride
content, mineralization, structure, and mechanical properties.
Fluoride incorporation into bone depends on many factors,
including ingestion from sources in addition to water (Burt,
1992), age, duration of residency (Richards et al., 1994), renal
function and other disease states (Ekstrand and Spak, 1990),
remodeling rate (Ishiguro et al., 1993), and genetic susceptibil-
ity (Dequeker and Declerck, 1993; Mousny et al., 2006). About
40% of the population in areas with water supplies naturally
fluoridated at very high levels are unaffected by skeletal fluoro-
sis (Choubisa, 2001), and about a third of patients who receive
fluoride as a therapy for osteoporosis are described as ‘non-
responders’ (Dequeker and Declerck, 1993), indicating that
intrinsic susceptibility to fluoride varies with the individual. A
genetic basis for these differences is supported by research with
different strains of mice (Mousny et al., 2006, 2008). In a large,
diverse urban center like Toronto, therefore, one would expect
that the population would display a range of genetic suscepti-
bilities to fluoride, which may in turn explain the broad range in
fluoride content measured for Toronto specimens. This may also
be part of the explanation for the contrasting pattern of differ-
ences between cities and quartiles.
Because the bone samples for this study were obtained from
patients undergoing surgery, the patients were generally older; it
is possible that they were not representative of the larger popula-
tion. However, aged populations are likely to be the most vulner-
able to any negative effects of municipal fluoride administration
because of both fluoride accumulation in bone over time (Richards
et al., 1994) and age-related declines in the mechanical properties
of bone (Mosekilde and Danielsen, 1987).
Many decades of epidemiological studies have shown min-
imal evidence of any effects of fluoride administration on
bone, and it is therefore very unlikely that municipally fluori-
dated water affects adults with healthy bone. In this study, no
effects of fluoride on mineralization (by BSE) and no substan-
tive negative effects of fluoride administration on bone
mechanical properties were observed. Our analysis of samples
at the tissue level, rather than the population level, reveals high
levels of variability in response to water fluoridation, which
may account for the lack of differences observed in epidemio-
logical studies (McDonagh et al., 2000). While we cannot
definitively rule out an effect of low-level fluoride accumula-
tion over long periods of time, especially if specific individu-
als have a genetic or disease background that renders them
unusually susceptible to fluoride, it nevertheless appears that
the contributors to bone health are too many and varied, and
any possible effect of municipal fluoride ingestion is too small,
for municipal water fluoridation to be a significant determi-
nant of bone health within the general public.

J Dent Res 89(11) 2010 Water Fluoridation and Bone Health 1223
ACKNOWLEDGMENTS
This work was funded by a grant from the Canadian Institutes of
Health Research. The authors acknowledge the technical assis-
tance of Adeline Ng, Maria Mendes, and Douglas Holmyard,
and thank the participating surgeons: Drs. Carol Hutchison and
Allan E. Gross at Mount Sinai Hospital, and Drs. David Zukor
and Olga Huk at the Jewish General Hospital. This paper is
based on a thesis submitted to the School of Graduate Studies,
University of Toronto, in partial fulfillment of the requirements
for the PhD degree of Debbie Chachra.
REFERENCES
Allolio B, Lehmann R (1999). Drinking water fluoridation and bone. Exp
Clin Endocrinol Diabetes 107:12-20.
Boivin G, Meunier PJ (1990). Fluoride and bone: toxicological and thera-
peutic aspects. In: The metabolic and molecular basis of acquired dis-
ease. 2nd ed. Cohen RD, Lewis B, Alberti KGMM, Denman AM,
editors. London, UK: Baillière Tindall, pp. 1805-1823.
Burt BA (1992). The changing patterns of systemic fluoride intake. J Dent
Res 71:1228-1237.
Chachra D (2001). The influence of lifelong exposure to environmental fluo-
ride on bone quality in humans [thesis]. Toronto: University of Toronto.
Choubisa SL (2001). Endemic fluorosis in southern Rajasthan, India.
Fluoride 34:61-70.
Dequeker J, Declerck K (1993). Fluor in the treatment of osteoporosis. An
overview of thirty years clinical research. Schweiz Med Wochenschr
123:2228-2234.
Ekstrand J, Spak CJ (1990). Fluoride pharmacokinetics: its implications in the
fluoride treatment of osteoporosis. J Bone Miner Res 5(Suppl 1):53-61.
Grynpas MD (1990). Fluoride effects on bone crystals. J Bone Miner Res
5(Suppl 1):169-175.
Grynpas MD, Holmyard DP, Pritzker KP (1994). Bone mineralization and
histomorphometry in biopsies of osteoporotic patients treated with fluo-
ride. Cells Mater 4:287-297.
Ishiguro K, Nakagaki H, Tsuboi S, Narita N, Kato K, Li J, et al. (1993).
Distribution of fluoride in cortical bone of human rib. Calcif Tissue Int
52:278-282.
Kindt JH, Thurner PJ, Lauer ME, Bosma BL, Schitter G, Fantner GE, et al.
(2008). In situ observation of fluoride-ion-induced hydroxyapatite-
collagen detachment on bone fracture surfaces by atomic force micros-
copy. Nanotech 18:135102.
Li Y, Liang C, Slemenda CW, Ji R, Sun S, Cao J, et al. (2001). Effect of
long-term exposure to fluoride in drinking water on risks of bone frac-
tures. J Bone Miner Res 16:932-939.
McDonagh MS, Whiting PF, Wilson PM, Sutton AJ, Chestnutt I, Cooper
J, et al. (2000). Systematic review of water fluoridation. BMJ 321:
855-859.
Mernagh JR, Harrison JE, Hancock R, McNeill KG (1977). Measurement of
fluoride in bone. Int J Appl Radiat Isot 28:581-583.
Mosekilde L, Danielsen CC (1987). Biomechanical competence of vertebral
trabecular bone in relation to ash density and age in normal individuals.
Bone 8:79-85.
Mousny M, Banse X, Wise L, Everett ET, Hancock R, Vieth R, et al. (2006).
The genetic influence on bone susceptibility to fluoride. Bone 39:
1283-1289.
Mousny M, Omelon S, Wise L, Everett ET, Dumitriu M, Holmyard DP,
et al. (2008). Fluoride effects on bone formation and mineralization are
influenced by genetics. Bone 43:1067-1074.
Richards A, Mosekilde L, Sogaard CH (1994). Normal age-related changes
in fluoride content of vertebral trabecular bone—relation to bone qual-
ity. Bone 15:21-26.
Thurner PJ, Erickson B, Turner P, Jungmann R, Lelujian J, Proctor A,
et al. (2009). The effect of NaF in vitro on the mechanical and
material properties of trabecular and cortical bone. Adv Mater 21:451-
457.
Turner CH, Burr DB (1993). Basic biomechanical measurements of bone: a
tutorial. Bone 14:595-608.
Turner CH, Boivin G, Meunier PJ (1993). A mathematical model for fluo-
ride uptake by the skeleton. Calcif Tissue Int 52:130-138.
Whitford GM (1989). The metabolism and toxicity of fluoride. Monographs
in oral science. Vol. 13. Basel: Karger AG, pp. 1-160.
Yan J, Mecholsky JJ, Clifton KB (2007). How tough is bone? Application of
elastic-plastic fracture mechanics to bone. Bone 40:479-484.

Reproducedwithpermissionofthecopyrightowner.Further reproduction prohibited withoutpermission.