Printer Friendly

Pleural mesothelioma in New Caledonia: Associations with environmental risk factors.

BACKGROUND: High incidences of malignant mesothelioma (MM) have been observed in New Caledonia. Previous work has shown an association between MM and soil containing serpentinite.

OBJECTIVES: We studied the spatial and temporal variation of MM and its association with environmental factors.

METHODS: We investigated the 109 MM cases recorded in the Cancer Registry of New Caledonia between 1984 and 2008 and performed spatial, temporal, and space-time cluster analyses. We conducted an ecological analysis involving 100 tribes over a large area including those with the highest incidence rates. Associations with environmental factors were assessed using logistic and Poisson regression analyses.

RESULTS: The highest incidence was observed in the Houailou area with a world age-standardized rate of 128.7 per 100,000 person-years [95% confidence interval (CI), 70.41-137.84]. A significant spatial cluster grouped 18 tribes (31 observed cases vs. 8 expected cases; p = 0.001), but no significant temporal clusters were identified. The ecological analyses identified serpentinite on roads as the greatest environmental risk factor (odds ratio = 495.0; 95% CI, 46.2-4679-7; multivariate incidence rate ratio = 13.0; 95% CI, 10.2-16.6). The risk increased with serpentinite surface, proximity to serpentinite quarries and distance to the peridotite massif. The association with serpentines was stronger than with amphiboles. Living on a slope and close to dense vegetation appeared protective. The use of whitewash, previously suggested to be a risk factor, was not associated with MM incidence.

CONCLUSIONS: Presence of serpentinite on roads is a major environmental risk factor for mesothelioma in New Caledonia.

KEY WORDS: antigorite, asbestos, cluster analysis, ecological study, environment, mesothelioma, New Caledonia, serpentinite. Environ Health Perspect 119:695-700 (2011). doi:10.1289/ehp.1002862 [Online 30 December 2010]

Malignant mesothelioma (MM) is a rare and fatal tumor of the pleura, with a worldwide annual incidence of one or two cases per million inhabitants. The incidence has increased since the 1950s among men in regions where asbestos was mined or industrially used (MacDonald and MacDonald 1986). Inhalation of asbestos fibers is the only established causal factor for MM (Merler and Chellini 1992). MM may also occur after exposure to erionite fibers (Baris et al. 1979), or fluoroedcnite fibers (Putzu et al. 2006). The risk of MM increases with level and duration of exposure (Cugell and Kamp 2004), with a median latency period of 30-40 years (Montanaro et al. 2003). Incidence fates have been reported to be 1.2-9 times higher among men than among women because of occupational exposure (Ross and McDonald 1995).

MM has been associated with domestic exposure in family members of asbestos workers (Ferrante et al. 2007) and with environmental exposure (Hillerdal 1999; Maule et al. 2007) related to asbestos factories, mines, or naturally occurring asbestos in Greece (Constantopoulos et al. 1987), Cyprus (McConnochie et al. 1987), Turkey (Carbone et al. 2007), Corsica (Rey et al. 1993), Italy (Magnani et al. 2001), and the United States [California (Pan et al. 2005)].

There are two main families of asbestos: amphiboles (tremolite, actinolite, crocidolite) and serpentines (chrysorile, antigorite, lizard-ite). Asbestos fibers vary in length and shape: chrysotile has long, flexible fibers, whereas amphiboles fibers are brittle. Although all types of asbestos fibers can be associated with MM, the highest risks have been reported for amphiboles (Hodgson and Darnton 2000). Health problems associated with exposure to airborne asbestos particles have resulted in severe restrictions on the use of asbestos from the 1980s. In natural environments, asbestos is commonly associated with ultramafic rocks (rich in iron and magnesium minerals), mostly peridotites, of which hydration produces serpentinites (Schreier 1989).

A third of the main island of New Caledonia, called the Grande Terre, consists of ultramafic rocks, mainly peridotites and serpentinites (Birrell and Wright 1945). Three geological areas may contain asbestos: the Boghen unit, the Peridotite massifs, and the Northern Caledonian metamorphic complex (Figure 1). The territory is divided into 33 communes that are administrative areas centered in a city or village. These communes include a few Melanesian tribes or several dozen. Nickel mining and smelting is the main economic activity of the country, and nickel ores have been shown to contain asbestos (Langer et al. 1980).

Very high incidences of MM and lung cancer have been observed in New Caledonia. For the 1996-2005 period, die age-standardized incidence rate (ASR) of MM was 4.6 per 100,000 person-years among men and 3.1 among women; lung ASR in Melanesians was 92.8 per 100,000 person-years among men and 27.8 among women (Baumann and Rougier 2005). A previous study showed an association between MM and the use of a whitewash made with local soils, called "Po" (Goldberg et al. 1995). Analysis of some Po samples confirmed that this whitewash could contain tremolite.

We conducted a case-control study of MM using the 68 cases recorded in the Cancer Registry of New Caledonia for the 1984-2002 period (Baumann et al. 2007b). The world ASR of MM was 7.52 per 100,000 persons among Melanesians and 0.57 among Caucasians. The early onset of the disease, beginning at the age of 30 years, and the sex ratio of 1.03, suggested a nonoccupational cause for Melanesians. Thirty percent of the cases lived in Houailou, an area where mining activity was important during the 1960s-1970s. Geographical analysis demonstrated a significant association between soil containing serpentinite and MM.

In this article, we present the results of spatial, temporal, and space-time cluster analyses on all MM cases recorded in New Caledonia between 1984 and 2008, and an ecological analysis conducted in a specific area that included the places with the highest MM incidences. The main objectives of this ecological study were to analyze the clusters of MM, and the associations between MM and exposure to natural sources of asbestos.

Materials and Methods

We identified MM cases from the Cancer Registry of New Caledonia. Eligible cases were malignant pleural mesotheliomas that were histologically confirmed and diagnosed from 1984 to 2008.

We conducted epidemiological investigations on all 109 eligible MM cases. Using a standard questionnaire, we conducted interviews with the living cases (12%) or two of their closest relatives to collect demographic information: sex; date of birth; race/ethnicity; date of diagnosis; residential histories, including location and description of each location; occupational history; and school history. Of the cases interviewed, we excluded five because they had lived in New Caledonia for < 20 years before the date of diagnosis. All participants gave their informed consent. These investigations were conducted within the framework of the cancer registry and were approved by the French and the local authorities.

We identified a study area that included the communes with the highest MM incidences and the three geological units that may contain asbestos (Figure 1). We conducted an ecological analysis of the 100 tribes living in this area by comparing environmental exposure between tribes with and without MM cases. Environmental determinants, such as presence of serpentinite on roads, mining activity in close proximity, and vegetation cover, were assessed at the tribal level and will be described briefly below.

Environmental and geological investigations. To locate the known bodies of ulttamafic rocks and serpentinite and peri-dotite massifs--the potential sources of natural asbestos--we used a digital version of the BD SIGEOL 1/50,000 geological maps (Gouvetnement de la Nouvelle-Caledonie 2010). We conducted geological and environmental investigations in the environment of each of the 100 tribes. After our first investigations (Baumann et al. 2007a, 2008), a census of setpentinite quarries was carried out by the service of Mines and Quarries (Gouvernement de la Nouvelle-Caledonie 2010), which we also used. Serpentinite was widely used to cover roads and may thus be a potential source of exposure. The territorial and the provincial toads, which ate the most used, began to be sealed in the 1970s in the south and in the 1990s in the north.

In each tribe, we questioned elders about the material used to whitewash the houses, the history of this practice, and the localities of the quarries of Po. We collected samples of indoor and outdoor Pb on the walls of existing whitewashed dwellings and in the quarries. We detected three categories of Po: tremolitic (containing asbestos fibers), non-tremolitic, and partly tremolitic (some families used tremolitic Po; others used a material without fibers). We collected samples of all natural materials suspected to contain asbestos in die close vicinity of the tribe including outcrops, road cuts, and quarries.

We carried out mineralogical analyses on die 486 samples collected to assess the presence of asbestos. We first analyzed fibers using polarized light microscopy (PLM). Where required, we used CM20 200 kV analytical transmission electron microscope X-ray powder diffraction and transmission electron microscopy (Philips, Amsterdam, the Netherlands) coupled with an EDAX energy-dispersive X-ray spectrometer (Ametek, Mahwah, NJ, USA) to determine the nature of the asbestos. For cost reasons, we analyzed only 94 samples using both approaches. Of the 392 remaining samples, 97 were negative, and the PLM was sufficiently reliable to determine the nature of the asbestos for the remaining samples.

We constructed an index of mining activity influence as follows:

T x W/D/1,000,

where T is the tonnage of nickel ore extracted between 1940 and 1980; W is a coefficient depending on the orientation of the mining massif with respect to the tribe and the prevailing wind; and D is the distance between the center of the tribe and the nearest massif.

We integrated all data into the geographic information system (GIS) ArcGIS (version 9.02; ESRI, Redlands, CA, USA). Mean slope and mean curvature over 1 [km.sup.2] around the center of the tribe, surfaces areas, and all distances were automatically calculated using ArcGIS on a 50-m cell-size digital terrain model (MNT 50, Gouvernement de la Nouvelle-Caledonie, Noumea, Nouvelle-Caledonie).

Statistical analyses. We calculated the 95% confidence intervals (CIs) of incidences on die assumption that the number of cases observed followed a Poisson distribution (after testing the compliance with a Poisson distribution). Because of the mean latency of 30 years--minimum 20 years--we used two methods to study the geographical distribution of MM cases. The first involved the longest place of residence, provided that the subject had lived there for at least 20 years before diagnosis. The second method involved the place where the subject was residing in the 30th year before the date of diagnosis. For each commune and each tribe, we calculated crude and world ASRs of MM with their 95% CI by both analysis methods. We used the population data from the 1983, 1989, 1996, and 2004 census.

We performed a spatial, temporal, and space-time disease clustering analysis to detect possible areas or periods associated with a significantly higher risk of MM. We used the Kulldorffs spatial scan statistical method that adopts maximum likelihood techniques to test for the presence of clusters and identify their location. The spatial scan analysis was implemented in SaTScan software (version 8.2.1; SaTScan, Boston, MA, USA). Sex and age were considered as covariates. We evaluated clusters using 999 Monte Carlo simulations.

For the univariate and multivariate analyses, we grouped each tribe with < 50 inhabitants (1989 census) with the closest tribe, provided that it was located in the same valley or no further than 5 km away from the coast line. We analyzed the associations between exposure estimates and MM by two methods: we first compared tribes with and without MM cases and calculated odds ratios (ORs), then Poisson regression was applied to calculate incidence rate ratios (IRR). We used the first approach because it is statistically more effective in detecting a significant factor, and the second to take into account the extent of MM incidence and to estimate the linear influence of the factor studied on this incidence. To show how the risk of incidence changed with the categorized variables, ASRs are repotted using the lowest exposure category as reference groups for risk factors, and the highest exposure category for protective factors. Continuous variables were first categorized by tertiles. If two contiguous classes were comparable, the variable was then categorized into two classes. To enhance the strength of the results, we carried out multivariate analyses on bicategorized dependent variables.

We compared means with a t test, Welch variant when the Bartlett's test for equal variances was significant, and Kruskal-Wallis test when more than two groups were compared. Percentages were compared with a Pearson chi-square or a Fisher's exact test when numbers were small. The statistical tests applied to compare incidence rates are based on die convergence of the Poisson distribution toward the normal distribution. Multivariate analyses were conducted using backward stepwise logistic and Poisson regressions, including all the variables having a p-value < 0.25 in the univariate analysis. All interactions were tested by using the likelihood-ratio test. The most significant interactions were included in the multivariate logistic model. When two variables were strongly correlated (r > 0.7), only one of them was included in the multiple Poisson regression model. ORs and IRR are given with their 95% CIs. All reported p-values are two-tailed.

The statistical calculations were performed with R (R Foundation for Statistical Computing, Vienna, Austria) and Stata SE software (version 8.0; StataCorp, College Station, Texas, USA).

Results

Among the 104 MM cases, 49 were men and 55 women. Ninety-one (87%) were Melanesians, eight (8%) Caucasians, and five (5%) Tahitians. Age varied from 31 to 81 years, with a mean of 60.0 years, comparable in both sexes. Mean age was higher for Caucasian men (71.9 years of age) than for Melanesian men (60.3 years of age). The sex ratio (male:female) was higher for non-Melanesians (2.25) compared with Melanesians (0.784). See Supplemental Material, Table 1 (doi:10.1289/ehp.1002862) for a more detailed overview of population characteristics.
Table 1. Univariate analyses of environmental factors influences on
MM ASR by tribe.

                       Comparison of tribes with MM (n = 34) versus
                          tribes without MM cases (n = 48)
Univariate analyses/   No. of  OR (95% CI)
Variables              tribes

Serpentinite on
track

Absence                46      1

Presence               34      465.00 (2.360-9,
                               161)

Antigorite

Absence                35      1

Presence               47      20.667
                               (4.028-106.03)

Serpentinite surface
(k[m.sup.2])

0-1.2                  27      1

1.21-3                 31      13.333
                               I2.683-66.256)

3.01-30                24      25.000
                               (4.698-133.03)

No. of quarries < 20
km

0-3                    32      1

4-6                    27      4.320
                               (1.276-14.623)

7-21                   23      15.300
                               (4.034-58.021)

Distance to massif
(km)

10.01-43               39      1

0-10                   43      7.716
                               (2.423-24.570)

Distance to quarry
(km)

9.1-31                 25      1

3.1-9                  29      2.550
                               (0.875-7.429)

0-3                    28      9.452
                               (2.526-35.361)

Chrysotile

Absence                54      1

Presence               28      4.275
                               (1.520-12.023)

Coast distance (km)

4.1-22                 35      1

0-4                    47      0.318(0.122-0.830)

Curvature (in
degrees)

> 0                    16      1

[less than or equal    66      0.343
to] 0                          (0.107-1.096)

No. of censed
dwellings

0                      37      1

1-54                   45      2.471
                               (0.989-6.174)

Dense vegetation

No                     38      1

Yes                    44      0.420
                               (0.166-1.059)

Distance to main road
(km)

1.1-36.5               43      1

0-1                    39      2.180
                               (0.871-5.458)

Mine index (tonne/m)

0                      47      1

> 0                    35      2.051
                               (0.821-5.126)

Tremolite-actinolite

Absence                30      1

Presence               52      1.714
                               (0.664-4.423)

Mean slope (in
degrees)

10.1-24.6              50      1

1.6-10                 32      1.167(0.472-2.884)

Whitewashing

Nonfibrous             18      1

Partly fibrous         16      1.333
                               (0.337-5.270)

Totally fibrous        48      0.330
                               (0.108-1.008)

No. of years of Po     1945
use after

0-10                   28      1

10.5-30                28      1.000
                               (0.351-2.851)

30.5-55                26      0.300
                               (0.093-0.972)

                        Comparison of mean MM ASR by covariate class
Univariate analyses/   MM ASR  p-Value             Poisson IRR (95%
Variables                                          CI)

Serpentinite on
track

Absence                1.6

Presence               97.0    < 0.0001            60.654
                               (a)                 (68.144-76.415)

Antigorite

Absence                3.3

Presence               71.2    < 0.0001            21.682
                               (a)                 (18.003-26.114)

Serpentinite surface
(k[m.sup.2])

0-1.2                  3.2

1.21-3                 52.4

3.01-30                72.9    0.0001              2.484
                               (a)                 (2.366-2.609)

No. of quarries < 20
km

0-3                    8.9

4-6                    44.0

7-21                   86.4    0.0004              2.665
                               (b)                 (2.541-2.795)

Distance to massif
(km)

10.01-43               7.9

0-10                   73.3    < 0.0001            9.309
                               (a)                 (8.281-10.466)

Distance to quarry
(km)

9.1-31                 29.9

3.1-9                  39.9

0-3                    55.6    0.0265              1.368
                               (b)                 (1.310-1.427)

Chrysotile

Absence                29.164

Presence               67.339  0.0289              2.309
                                                   (2.159-2.469)

Coast distance (km)

4.1-22                 60.4

0-4                    28.7    NS                  0.464
                                                   (0.444-0.508)

Curvature (in
degrees)

> 0                    55.0

[less than or equal    39.1    NS                  0.711
to] 0                                              (0.659-0.768)

No. of censed
dwellings

0                      28.9

1-54                   53.1    NS                  1.834
                                                   (1.706-1.971)

Dense vegetation

No                     58.0

Yes                    28.5    NS (a)              0.492
                                                   (0.456-0.527)

Distance to main road
(km)

1.1-36.5               39.6

0-1                    45.0    NS                  1.136
                                                   (1.062-1.214)

Mine index (tonne/m)

0                      26.5

> 0                    63.3    0.0438              2.392
                               (a)                 (2.232-2.564)

Tremolite-actinolite

Absence                38.7

Presence               44.2    NS                  1.142
                                                   (1.065-1.226)

Mean slope (in
degrees)

10.1-24.6              51.0

1.6-10                 28.4    NS                  0.557(0.517-0.601)

Whitewashing

Nonfibrous             43.9

Partly fibrous         88.5

Totally fibrous        26.1    0.0451              0.727
                                                   (0.670-0.755)

No. of years of Po
use after

0-10                   48.3

10.5-30                57.2

30.5-55                19.5    NS (b)              0.709
                                                   (0.680-0.740)

Abbreviations: ASR, world age-standardized incidence rate per 100,000
person-years; NS, nonsignificant.
(a) Welch test. (b) Kruskal-Wallis test.


A large proportion of men had many different jobs for very short periods (i.e., a few months). Therefore, detailed analyses of occupation exposure were not feasible.

Both methods used to analyze place of residence showed no significant difference [Supplemental Material, Figure 1 (doi:10.1289/ehp.1002862)]. The results are presented according to the place of residence 30 years before the date of diagnosis.

MM incidence. Crude and standardized MM incidences were high in three communes (Figure 1). Compared with the main city of Noumea (ASR = 0.64; 95% CI, 0.13-1.88), we found that the MM incidence was 200 times higher in Houailou (ASR = 128.66; 95% CI, 70.41-137.84), 40 times higher in Kone (ASR = 25.46; 95% CI, 14.25-41.98), and 25 times higher in Poindimie (ASR = 15.37; 95% CI, 3.17-44.92). The study area included 7 of the 10 communes with the highest MM incidence.

The number of cases by tribe varied from 0 to 7. The logarithm of ASR showed a bimodal distribution, that is, 62 tribes had no incidence, and the 38 odiers presented a symmetric distribution centered on the median of 4. The chi-square test of compliance of the observed distribution among these 38 tribes with a Poisson distribution did not show any significant deviation.

Seventy-eight MM cases lived in the study area 30 years before diagnosis. One of them had never lived in a tribe and was excluded from the ecological analysis. The other 77 cases were Melanesians, except for one Polynesian. The sex ratio M:F of < 1 and the same mean age at diagnosis for both sexes (< 60 years of age) confirmed the high probability of an environmental causal exposure.

Cluster analyses. The spatial analysis highlighted one cluster grouping 18 central tribes in Houailou-Bourail area (Figure 2). Thirty-one cases were observed versus 8.12 expected cases, with a relative risk of 5-76 and a log-likelihood ratio of 23.02; p = 0.001. Two secondary clusters were observed in Kone and Poindimie, but these were not significant.

No significant temporal cluster was found. The spatiotemporal analysis showed one cluster grouping of 14 tribes in Houailou area for the 1991-2003 period: 18 observed cases, 3.10 expected cases, p = 0.001.

Description of environmental determinants. The natural sources of asbestos. Based on geological maps and environmental sampling, we identified the following natural sources of asbestos: in the Boghen unit, serpentinite slivers were the host of tremolitic occurrences that, by weathering at the base of the soil horizon, evolve into white aggregates [Supplemental Material, Figure 2 (doi:10.1289/ehp.1002862)]. This white soil was doughlike and similar to clay. In the peridotite massifs, three kinds of asbestos were observed: mainly anti-gorite having an appearance of friable rock, which splits up into smaller fragments, chrysotile occurring as small (in the millimeter to centimeter range) veins in fresh rocks, and sometimes tremolite [Supplemental Material, Figure 3 (doi:10.1289/ehp.1002862)]. The majority of asbestos occurrences in the Northern Caledonian Metamorphic complex were represented by tremolite-actinolite white aggregates.

We found three categories of fibers: short (< 20 [micro]m) and thin (< 1 [micro]m) tremolite fibers; long (> 100 [micro]m) and flexible chrysotile fibers; and sheaves of fine fibers (< 1 [micro]m), splintering from the edges of antigorite laths.

Serpentinite quarries. Within a 5-km radius around the center of each tribe, the surface of serpentinite outcrops exceeded 7 [km.sup.2] in 12 tribes [Supplemental Material, Figure 4 (doi:10.1289/ehp.1002862)]. Forty-seven tribes were situated < 5 km from a serpentinite quarry. Seventeen tribes lived within a 20-km distance by road from > 10 serpentinite quarries.

Roads. Eighty-nine tribes lived < 10 km from the main toad. The presence of serpentinite on roads (surface or road cut) was identified in 44 tribes. According to the local authorities, quarries and roads in our study area were in use at least at the end of the 1960s.

Whitewash of dwelling walls. The material used for whitewashing varied according to the geological environment of the tribe. Most of the tribes living near the sea used a soup made with fired coral. Others used white clay or crushed white rocks diluted with water, including silica and tremolite. Since the 1950s, the cob dwellings have been replaced with corrugated iron or concrete houses. The last year for whitewash use varied from 1950 to 2000. After 1970, fibrous Po was still used in 41 ttibes [Supplemental Material, Figutes 5 and 6 (doi:10.1289/ehp.1002862)].

Mining activity. Fifty-five tribes lived < 10 km from the peridotite massif. The tonnage of nickel ore mined between 1940 and 1980 was substantial in the East of Houailou. About 12,000,000 tons were extracted from the mines of Poro and 14,300,000 tons from the mines of Kouaoua [Supplemental Material, Figure 7 (doi:10.1289/ehp.1002862)].

Vegetation and situation. Four types of vegetation were observed in the environment of the tribes (data before 1996): bush, savannah, forest, and dense mining maquis. The plant cover was dense in 51 tribes (50 forest and 1 dense mining maquis) and nondense in 49 tribes (48 bush and one savannah).

The mean curvature was positive, indicating a convex shape, for only 22 tribes. Distance to the coasdine was < 4 km for 51 tribes.

Analyses of exposure and environmental factors. After grouping the smallest tribes, univariate and multivariate analyses were carried out on 82 geographical units.

Comparison of tribes with and without MM cases. The most significant risk factor highlighted by the univariate analyses (Table 1) is the presence of serpentinite on roads: OR = 465.0; 95% CI, 2.360-9161. This factor was present among 97% of tribes with MM cases versus 6% of tribes without MM cases (p < 0.001).

We also found significant increased risks (in descending order) for the presence of antigorite, surface of serpentinite outcrops, the number of serpentinite quarries in a 20-km radius, proximity to the peridotite, proximity to the nearest quarry of serpentinite, and presence of chrysotile. Proximity to the coast constituted a significantly protective factor.

Univariate Poisson regression (Table 1) showed similar results, with the highest IRR observed for serpentinite on roads (IRR = 60.654; 95% CI, 68.144-76.415). Statistically significant associations were also observed for the presence of antigorite, the distance to the peridotite massif, the number of serpentinite quarries, the surface area of serpentinite, the mine index, and the presence of chrysotile. Proximity to the coast line, dense vegetation, and mean slope >10[degrees] were inversely associated.

Interactions. The mine index was strongly correlated with the distance to the peridotite massif (r = 0.77; p < 0.0001). Thus, to avoid coiinearity, we omitted the mine index in the multivariate analyses. Other correlations were [less than or equal to] 0.7.

Significant interactions were observed between the number of quarries and the presence of antigorite (p = 0.0159), the presence of chrysotile (p = 0.0025) and serpentinite on roads (p = 0.0026), and between the distance to the nearest quarry and serpentinite surface (p = 0.0218). These interactions were included in the multivariate logistic regression.

Multivariate analyses (Table 2). Logistic regression resulted in one highly predictive factor: the presence of serpentinite on roads (OR -465.0; 95% CI, 46.2-4679.7), which explained 72% of the total variance. We carried out a second logistic regression with the same variables except serpentinite on roads. This resulted in five predictive factors in die final model: presence of antigorite, distance to peridotite massif [less than or equal to] 10 km, serpentinite surface > 2 [km.sup.2], presence of chryso-tile, and negative curvature, which explained 52% of the total variance.
Table 2. Multivariate analyses of environmental factors influences
on MM ASR by tribe.

Tribes with MM (n = 34) versus      Logistic regression (pseudo
tribes without MM cases (n = 48)    [R.sup.2] = 0.7221)

Variable                            OR (95% CI)                p-Value

Serpentinite on roads:              465.0 46.2-4679.7          < 0.001
presence/absence

Tribes with MM (n = 34) versus      Logistic regression without
tribes without MM cases (n = 48)    serpentinite on roads
                                    (pseudo [R.sup.2] = 0.5334)

Variable                            OR (95% CI)                p-Value

Antigorite: presence/absence        35.823 (4.053-316.6)       0.001

Distance to massif: 0-10 km/> 10    19.825 (2.935-134.05)      0.002
km

Serpentinite surface: 2.01-30/0-2   9.773 (1.871-51.04)        0.007
k[m.sup.2]

Chrysotile: presence/absence        9.652 (1.638-56.860)       0.012

Curvature: [less than or equal to]  0.157 (0.0074-0.503)       0.009
0/> 0

Multivariate analysis of MM ASR (n  Poisson regression ([R.sup.2]
= 80)                               = 0.7615)

Variable                            IRR (95% CI)               p-Value

Serpentinite on roads:              13.052 (10.225-16.661)     < 0.001
presence/absence

Antigorite: presence/absence        4.799 (3.880-5.935)        < 0.001

Distance to massif: 0-10 km/> 10    3.241 (2.812-3.736)        < 0.001
km

Serpentinite surface: 2.01-30/0-2   2.430 (2.158-2.728)        < 0.001
k[m.sup.2]

Chrysotile: presence/absence        1.638 (1.547-1.832)        < 0.001

Distance to quarry: 0-5 km/> 5 km   1.205 (1.104-1.315)        < 0.001

Mean slope: >10[degrees]/<          0.679 (0.599-0.701)        < 0.001
10[degrees]

Curvature: [less than or equal to]  0.844 (0.775-0.919)        < 0.001
0/> 0

Coast distance: 0-4 km/> 4 km       0.849 (0.784-0.818)        < 0.001


The multivariate Poisson regression model involved six environmental determinants: serpentinite on roads, antigorite, distance to peridorite massif [less than or equal to] 10 km, serpentinite surface > 2 [km.sup.2], chrysotile, distance to nearest quarry [less than or equal to] 5 km, and three protective factors: mean slope < 10[degrees], negative curvature, and distance to coast line [less than or equal to] 4 km, explaining 76% of the total variance.

Discussion

Our study has confirmed the high risk of MM in New Caledonia and has identified geographical clusters. Living near roads covered by serpentinite was associated with the greatest risk of MM. Other significant risk factors included the number and the proximity of serpentinite quarries and the size of the surface area containing serpentinites.

Because of the good mechanical properties of serpentinite, its stability under rain, and the ease of crushing it, this material was widely used to cover roads, with the potential of asbestos exposure for those living closest to these roads. The large number of quarries along the road between Kone and Poindimie provides evidence of this extensive practice until the completion of the present sealed road. Serpentinite quarries are numerous in the Houailou area, and some of them were still in operation during our investigations.

The second interesting result is the difference in risk between serpentines and amphiboles. In particular, a much stronger association with MM risk was found for antigorite (OR = 20.667; 95% CI, 4.028-106.03) compared with tremolite-actinolite (OR = 1.714; 95% CI, 0.664-4.423). This difference may be explained by the shape or the use of these minerals. Tremolite commonly comes in the form of doughy and aggregated white soil under the vegetable layer, whereas the fibrous form of antigorite comes from the weathering of the rocks, which produces increasingly small pieces down to very fine dust.

Unofficial serpentinite qualities containing mostly antigorite were abundant, and this matetial was very commonly used to cover the roads. The two serpentines chrysotile and antigorite have very similar chemical composition but different structures. Antigorite presents lathes that splinter, whereas chrysotile has a coiled structure. Recent works showed the cytotoxicity of antigorite fibers in vivo (Pugnaloni et al. 2010). However, antigorite does not belong to industrial asbestos and is therefore not recognized in asbestos regulation.

In the 1930s, Melanesians were encouraged to whitewash the walls of their dwellings. This Po could be made with different material, depending on the environment. Tremolite was used by melting this paste with hot water. This practice was widespread until the 1990s in areas where MM incidence rates were low; census data showed 95 tremolitic dwellings in Poya (three MM cases, two living in tribes) and 65 in Hienghene (one MM case) compared with only 26 dwellings in Houailou and 17 in Poindimie, where the MM incidences were the highest. Consistent with this, the use of whitewash did not emerge as a significant risk factor for MM in our study. The different indicators (number of tremolitic dwellings, use of tremolitic Po after 1970, last year of Po use) confirm this result and contradict previous studies (Luce et al. 2000). Thus, either natural tremolite fibers are less hazardous, or tremolite aggregates as used in whitewash retain the fibers and avoid their dispersal and subsequent exposure.

Mining activity produces dust that may contain asbestos fibers when peridotites are serpentinized. This dust can affect tribes downwind from the mines. However, our mine index showed only a very small effect; proximity to the peridotite massifs, on the other hand, was significantly associated with MM. Living on a slope > 10[degrees] and around dense vegetation (nonsignificant) appeared to be protective, probably because they prevent dust propagation.

Environmental exposure is usually related to low relative risks with a complex mixture of several low-level intercorrelated exposures (Pekkanen and Pearce 2001). Because of the long latency period and the involuntary nature of environmental exposure, it is difficult to reconstruct personal exposure accurately. Most environmental variables, such as presence of serpentinite on roads, mining activity, and plant cover, are measured at the population level, so they can be analyzed by comparing populations rather than individuals. We therefore chose an ecological study design to assess the association between environmental asbestos exposure and MM. To minimize the typical problems associated with ecological analyses, and because our environmental data are relative to the tribes, we studied spatial units at the tribal level. Environmental data were similar to only nearby tribes, which were grouped when the population of one tribe was very small.

To test whether some MM cases were not identified by the cancer registry, we checked all MM cases recorded in the death certificate files, questioned the public and private lung specialists, and visited the health centers. We did not find any new cases. We also studied die geographic distribution of all cancer incidence in New Caledonia for 1990-2005 (results not shown). The ASR of all cancers was in the same order in Hienghene and Poya as in the other communes. Hienghene showed the highest incidence rate among women (due to very high incidence of thyroid and breast cancer), indicating good quality of medical detection and registration of cancers in this area. Thus, the low MM incidence in Hienghene and Poya is unlikely to have arisen from a lack of registration, and the observed spatial MM distribution is unlikely to be due to selection effect.

Information on occupation was insufficient to allow the role of occupational exposure to be assessed. This is a limitation of our study. However, the sex ratio (M:F < 1) of MM cases and the mean age of < 60 years for both sexes suggest a high probability of a causal environmental exposure rather than occupational exposure. Moreover, occupational sources of exposure are unlikely to follow the same spatial distribution as environmental sources. Therefore, residual confounding by occupational exposure is not likely to explain our results.

Pan et al. (2005) showed an inverse association between mesothelioma and the distance that people lived from the nearest naturally occurring asbestos in California. Our study adds to the evidence that natural sources of asbestos increase the risk of mesothelioma, and a strength of our study is that it takes into account multiple exposure sources and protective environmental factors.

Conclusions

This study provides strong evidence that the use of serpentinite for road surfacing is a major environmental risk factor for mesothelioma in New Caledonia. We communicated our first results to health and local authorities, which resulted in the government establishing inventories of quarries and sites built on naturally occurring asbestos and in initiating rehabilitation studies and new laws to protect workers. The same measures have not yet been established for residents.

Most environmental asbestos fibers are short and thin. As for antigorite fibers, short fibers are not usually taken into account in analyses. Recent studies have shown that these fibers could be responsible for mesothelioma (Suzuki et al. 2005). Thus, to understand the hazards of natural sources of asbestos, studies are now needed on the ability of these sources to release fibers into the air, and the types and sizes of fibers causally related to the disease.

REFERENCES

Baris YL, Artvinli M, Sahin LAA. 1979. Environmental mesothelioma in Turkey. Ann MY Acad Sci 330:423-432.

Baumann F, Maurizot P, Ambrosi JP, Robineau B. 2007a. Confrontation des donnees epidemiologiques et des donnees geologiques pour etudier I'etiologie du mesotheliome environnemental en Nouvelle-Caledonie: application dans les deux zones test des communes de Houailou et de Bourail [in French]. Noumea, New Caledonia:Rapport pour le Ministere de I'Outre-Mer, Institut Pasteur de NouvelleCaledonie.

Baumann F, Maurizot P, Robineau B, Ambrosi JP. 2008. L'amiante environnemental en Nouvelle-Caledonie: confrontation des donnees epidemiologiques et des donnees ecologiques dans le secteur de Kone-Touho-Poindimie [in French]. Noumea, New Caledonia:Rapport pour le Ministere de I'Outre-Mer, Institut Pasteur de Nouvelle-Caledonie.

Baumann F, Rougier Y. 2005. Registre du cancer de NouvelleCaledonie, bilan de 15 annees de surveillance [in French]. Bulletin epidemiologique hebdomadaire 34:169-171.

Baumann F, Rougier Y, Ambrosi JP, Robineau B. 2007b. Pleural mesothelioma in New Caledonia: an acute environmental concern. Cancer Detect Prev 31:70-76.

Birrell K, Wright A. 1945. A serpentine soil in New Caledonia. New Zealand J Sci Technol 27:72-76.

Carbone M, Emri S, Dogan A, Steele I, Tuncer M, Pass H, et al. 2007. A mesothelioma epidemic in Cappadocia: scientific developments and unexpected social outcomes. Nat Rev Cancer 7:147-154.

Constantopoulos SH, Malamou-Mitsi V, Goudevenos J, Papathanasiou MP, Pavlidis NA, Papadimitriou CS. 1987. High incidence of malignant pleural mesothelioma in neighbouring villages of North-Western Greece. Respiration 51:266-271.

Cugell D, Kamp D. 2004. Asbestos and the pleura. Chest 125:1103-1117.

Ferrante D, Bertolotti M, Todesco A, Mirabelli D, Terracini B, Magnani C. 2007, Cancer mortality and incidence of mesothelioma in a cohort of wives ot asbestos workers in Casale Monferrato, Italy. Environ Health Perspect 115:1401-1405.

Goldberg P, Luce D, Billon-Gailland MA, Quenel P, SalomonNekiriai C, Nicolau J, et al. 1995. Identification d'un exces de risque de cancer de la plevre en Nouvelle-Caledonie lie a I'exposition environnementale et domestique a la tremolite [in French]. Rev Epidemiol Sante Publique 43:444-450.

Gouvernement de la Nouvelle-Caledonie. 2010. GEOREP. Available: http://carto.gouv.nc/arcgis/rest/services/geologie/MapServer?f=jsapi [accessed 1 April 2010].

Hillerdal G. 1999. Mesothelioma: cases associated with nonoccupational and low dose exposures. Occup Environ Med 56:505-513.

Hodgson JT, Darnton A. 2000. The quantitative risks of mesothelioma and lung cancer in relation to asbestos exposure. Ann Occup Hyg 44:565-601.

Langer AM, Rohl AN, Slikoff IJ, Harlow GE, Prinz M. 1980. Asbestos as a cofactor in carcinogenesis among nickel processing workers. Science 209:420-422.

Luce D, Bugell, Goldberg P, Goldberg M, Salomon C, BillonGalland MA, et al. 2000. Environmental exposure to tremolite and respiratory cancer in New Caledonia: a case-control study. Am J Epidemiol 151:259-265.

MacDonald AD, MacDonald JC, 1986. Epidemiology of malignant mesothelioma. In: Asbestos-Related Malignancy (Antman K, Aisner J, eds). New York:Grune and Stratton, 31-35.

Magnani C, Dalmasso P, Biggeri A, Ivaldi C, Mirabelli D. 2001. Increased risk of malignant mesothelioma of the pleura after residential or domestic exposure to asbestos: a casecontrol study in Casale Monferrato, Italy. Environ Health Perspect 109(9):915-919.

Maule MM, Magnani C, Dalmasso P, Mirabelli D, Merletti F, Biggeri A. 2007. Modeling mesothelioma risk associated with environmental asbestos exposure. Environ Health Perspect 115:1066-1071.

McConnochie K, Simonato L, Mavrides P, Christofides P, Pooley FD, Wagner JC. 1987. Mesothelioma in Cyprus: the role of tremolite. Thorax 42:342-347.

Merler E, Chellini E. 1992. Epidemiology of primary tumors of the pleura [in Italian]. Ann 1st Super Sanita 28:133-146.

Montanaro F, Bray F, Gennaro V, Merler E, Tyczynski J, Parkin D. 2003. Pleural mesothelioma incidence in Europe: evidence of some deceleration in the increasing trends. Cancer Causes Control 14:791-803.

Pan XL, Day HW, Wang W, Beckett LA, Schenker MB. 2005. Residential proximity to naturally occurring asbestos and mesothelioma risk in California. Am J Respir Crit Care Med 172:1019-1025.

Pekkanen J, Pearce N. 2001. Environmental epidemiology: challenges and opportunities. Environ Health Perspect 109:1-5.

Pugnaloni A, Giantomassi F, Lucarini G, Capella S, Belmonte MM, Orciani M, et al. 2010. Effects of asbestiform antigorite on human alveolar epithelial A549 cells: a morphological and immunohistochemical study. Acta Histochem 112:133-146.

Putzu MG, Bruno C, Zona A, Massiccio M, Pasetto R, Piolatto PG, et al. 2006. Fluoro-edenitic fibres in the sputum of subjects from Biancavilla (Sicily): a pilot study. Environ Health 5:20; doi:10.1186/1476-069X-5-20 [Online 16 June 2006].

Rey F, Viallat JR, Boutin C, Farisse P, Billon-Galland MA, Hereng P, et al. 1993. Les mesotheliomes environnementaux en Corse du nord-est [in French] Rev Mal Respir 10:339-345.

Ross D, McDonald JC. 1995. Occupational and geographical factors in the epidemiology of malignant mesothelioma. Monaldi Arch Chest Dis 50:459-463.

Schreier H. 1989. Asbestos in the Natural Environment. Studies in Environmental Science 37, Amsterdam:Elsevier.

Suzuki Y, Yuen S, Ashle R. 2005. Short, thin asbestos fibers contribute to the development of human malignant mesothelioma: pathological evidence. Int J Hyg Environ Health 208:201-210.

Francine Baumann, (1) Pierre Maurizot, (2) Morgan Mangeas, (3) Jean-Paul Ambrosi, (4) Jeroen Douwes, (5) and Bernard Robineau (6)

(1) Universite de la Nouvelle-Caledonie, Equipe de Recherche en Informatique et Mathematiques, Noumea, New Caledonia; (2 Bureau de la Recherche Geologique et Miniere, Noumea, New Caledonia; (3) lnstitut de Recherche pour le Developpement, US140 Espace, Noumea, New Caledonia; (4) Centre Europeen de Recherche et d'Enseignement en Geosciences de I'Environnement, Aix Marseille University, CNRS, Aix-en-Provence, France; (5) Centre for Public Health Research, Massey University, Wellington, New Zealand; (6) Service Geologique, Direction des Mines et de I'Energie de Nouvelle-Caledonie, Noumea, New Caledonia

Address correspondence to F. Baumann, Institut de recherche pour le Developpement, US140 Espace BP A5, 98848 Noumea cedex, New Caledonia. Telephone: (687) 26 08 32. Fax: (687) 26 43 26. E-mail: fbaumann@live.fr

Supplemental Material is available online (doi:10.1289/ehp.1002862 via http://dx.doi.org/).

We thank pathologists Y. Rougier and V. Rouleau, the lung specialists, and the doctors of the dispensaries for their contributions; O. Grauby (Centre de Recherche en Matiere Condensee et Nanosciences, Centre National de la Recherche Scientifique) for providing scanning electron microscopy facilities; and J. Bell for his assistance in the English writing. We are grateful to the local and tribal authorities of Houailou, Bourail, Poya, Kone, Touho, Poindimie, and Hienghene for helping us to collect information.

This work was supported by grants from the French Ministry of Overseas Territories and from the Government of New Caledonia.

The funding sources had no role in the study design, data collection, data analysis, or interpretation of the findings.

The authors declare they have no actual or potential competing financial interests.

Received 15 August 2010; accepted 29 December 2010.
COPYRIGHT 2011 National Institute of Environmental Health Sciences
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2011 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Research
Author:Baumann, Francine; Maurizot, Pierre; Mangeas, Morgan; Ambrosi, Jean-Paul; Douwes, Jeroen; Robineau,
Publication:Environmental Health Perspectives
Article Type:Report
Geographic Code:8NEWC
Date:May 1, 2011
Words:6797
Previous Article:Exploration of the rapid effects of personal fine particulate matter exposure on arterial hemodynamics and vascular function during the same day.
Next Article:Toward a quantitative estimate of future heat wave mortality under global climate change.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters