close

Вход

Забыли?

вход по аккаунту

?

121

код для вставкиСкачать
Int. J. Cancer: 71, 917–923 (1997)
r 1997 Wiley-Liss, Inc.
Publication of the International Union Against Cancer
Publication de l’Union Internationale Contre le Cancer
OESOPHAGEAL CANCER IN FRANCE: POTENTIAL IMPORTANCE
OF HOT ALCOHOLIC DRINKS
Guy LAUNOY1,2, Chantal MILAN3, Nick E. DAY2, Jean FAIVRE3, Patrice PIENKOWSKI4 and Marc GIGNOUX1
1Registre des cancers digestifs du Calvados (CJF INSERM 9603), Caen, France
2MRC Biostatistics Unit, Cambridge, UK
3Registre des cancers digestifs de Côte d’Or (CRI INSERM 9505), Dijon, France
4Registre des cancers digestifs de Haute-Garonne, Toulouse, France
In France, major geographic variation exists in the incidence of oesophageal cancer, the highest incidence being
reported in Normandy and Brittany. The role of alcohol in the
risk of oesophageal cancer is well established in Western
countries. One possible explanation for geographical variation of incidence is that higher incidence of oesophageal
cancer is due to specific local alcoholic beverages. The aim of
this study was to determine whether different types of
alcoholic beverages exert different effects on the risk of
oesophageal cancer, and whether the variation of incidence in
France is due to variation in local drinking behaviour. We
conducted a multicentre case-control study in 3 regions of
France (Normandy, Burgundy and Midi-Pyrenées), among
which there is a 5-fold variation in incidence. We selected 208
cases and 399 controls, all males. During the interview, the
subject’s entire alcohol history was reconstituted, noting
each type of alcoholic beverage consumed throughout life.
The link between the risk of oesophageal cancer and alcohol
varies greatly according to the type of alcoholic beverage,
with aniseed aperitifs, hot spirits (especially hot Calvados)
and beer carrying the highest risk. Consumption of hot Calvados
appeared to explain about 2⁄3 of the inter-regional and urban/
rural differences in incidence, whereas total alcohol intake
explained less than 1⁄5. Even after adjustment for all other
alcoholic beverages, consumption of hot calvados explained
almost half of the peak in incidence of oesophageal cancer in
the Northwest of France, as well as half of the urban/rural
differences in incidence. Int. J. Cancer 71:917–923, 1997.
r 1997 Wiley-Liss, Inc.
In France, major geographic variations exist in the incidence of
oesophageal cancer. Large differences in incidence have been
reported among several areas, each covered by a cancer registry,
the highest incidence being reported in the West (Calvados,
Normandy), especially in the rural population (Benhamou et al.,
1990; Pottier et al., 1989). The role of alcohol in the risk of
oesophageal cancer is well established in Western countries.
However, in France, geographical variations in total alcohol
consumption accounts for little of the variations in the incidence of
oesophageal cancer. One possible explanation is that higher
incidence of oesophageal cancer is due to specific local alcoholic
beverages. Few studies have attempted to determine whether
different types of alcoholic beverages are associated with different
risks of oesophageal cancer, and conclusions have often been
limited by 2 factors (i) most studies were designed to study the
influence of total alcohol consumption and thus lack accurate data
on types of alcoholic beverages; (ii) many studies were based on a
single centre, so that the beverage associated most strongly with
cancer risk was often the one most frequently consumed [e.g., beer
in Soweto (South Africa) (Segal et al., 1988) or in a brewery
workers’ union (Adelhart et al., 1985), wine in Pordenone Province
and the Greater Milan area (Italy) (Barra et al., 1990)], so that it is
impossible to distinguish between a specific effect of a particular
beverage and the effect of total alcohol. In the west of France, cider
and apple brandy (Calvados) are major components of alcohol
consumption; however, studies in this area (Tuyns et al., 1979)
failed to demonstrate a specific effect for apple-based alcoholic
drinks, analysis suggesting that risk was largely determined by the
level of consumption of total alcohol (Breslow and Day, 1980).
Although these studies investigated type of beverage, they did not
examine drinking behaviour. In France, drinking habits vary
greatly from one area to another. We therefore conducted a
multicentre study in 3 French areas, each covered by a cancer
registry, with different drinking habits (Normandy, Burgundy and
Midi-Pyrenées), and with a 5-fold difference in oesophageal-cancer
incidence rates (Launoy et al., 1994).
SUBJECTS AND METHODS
The study was conducted between 1991 and 1994 in the
university hospitals of Caen (Normandy, department of Calvados),
Dijon (Burgundy, department of Côte d’Or) and Toulouse (MidiPyrenées, department of Haute-Garonne). Eligible cases were
males aged less than 85 years admitted to one of these hospitals
between January 1991 and April 1994 with histologically proven
squamous-cell carcinoma of the oesophagus. Adenocarcinoma of
the oesophagus was excluded from the study, due to expected
differences in risk factors. During this period, 223 cases were
identified; 4 patients left the hospital before the interview, 6 were
physically or mentally incapable of being interviewed, and 5
refused to be interviewed. Finally, 208 cases were included.
The control group consisted of 399 males admitted to the same
hospitals during the same period, in the rheumatology or orthopaedics unit for osteoarthritis (n 5 229), lumbago or sciatica (n 5 127),
or in the eye unit (n 5 43). Patients hospitalized for trauma were
excluded. Controls were matched for the hospital and for age
(Table I).
Data regarding alcohol, tobacco and diet were collected from
cases and from controls in a face-to-face, 2-hour interview.
Interviews were conducted by specially trained dieticians (4 in
Caen, 2 in Dijon and 1 in Toulouse) in a special separate room, with
no family members present. During the interview, the subject’s
entire smoking and alcohol histories were reconstituted, noting
each brand of tobacco and each type of alcoholic beverage
consumed throughout life. Up to 4 distinctive periods could be
recorded for each kind of alcoholic beverage. Within each consumption period, mean weekly intake and the subject’s age at the
beginning and end were recorded. Mean weekly lifetime intake
could thus be calculated for each kind of alcohol as total life-time
intake divided by number of years of consumption. Each beverage
consumption was assessed as mean weekly lifetime intake, as
defined above, since we have shown (Launoy et al., 1997) that
duration of alcohol consumption was not associated with risk of
oesophageal cancer and that mean weekly lifetime intake was the
most appropriate exposure measure. Similarly, duration of smoking
has been shown to be the most appropriate (Launoy et al., 1997).
The different alcoholic beverages were grouped into the following
categories: red wine, white wine, beer, cider, aniseed aperitifs,
other aperitifs (vermouth, quinine, tonic wine, etc.), whisky,
*To whom correspondence and reprint requests should be sent, at MRC
Biostatistics Unit, Institute of Public Health, University Forvie Site,
Robinson Way, Cambridge CB2 2SR, UK Fax: 01223330388. E mail
[email protected] (until August 1997), then: CJF INSERM
9603, Faculté de Médecine, CHU, Caen, France. Fax: 0231530852.
Received 16 January 1996; revised 22 February 1997
LAUNOY ET AL.
918
TABLE I – DISTRIBUTION OF CASES AND CONTROLS ACCORDING TO CENTRE, INTERVIEWER, SOCIODEMOGRAPHIC
VARIABLES, SMOKING HABITS AND TOTAL ALCOHOL INTAKE
Center
Caen (4 interviewers)
Interviewer 1
Interviewer 2
Interviewer 3
Interviewer 4
Dijon (2 interviewers)
Interviewer 5
Interviewer 6
Toulouse (1 interviewer)
Age
#50
51–60
61–70
.70
Place of residence
Urban
Rural
Occupation
Farmers
Workers and employees
Others
Level of education1
No certificate
Low
High
Marital status2
Living alone
Living with partner
Duration of tobacco consumption (years)
Non-smoker
1–14
15–29
30–44
45 and more
Total alcohol intake (g/week)
0–150
151–300
301–450
451–600
601 and more
Cases
(n 5 208)
N (%)
Controls
(n 5 399)
N (%)
Total
(N 5 607)
N (%)
112 (53.8)
30
44
5
33
45 (21.6)
40
5
51 (24.6)
203 (50.9)
78
79
8
38
90 (22.6)
79
11
106 (26.5)
315 (51.9)
108
123
13
71
135 (22.2)
119
16
157 (25.9)
33 (15.9)
62 (29.8)
82 (39.4)
31 (14.9)
79 (19.8)
119 (29.8)
146 (36.6)
55 (13.8)
112 (18.5)
181 (29.8)
228 (37.5)
86 (14.2)
140 (67.3)
68 (32.8)
301 (75.4)
98 (24.6)
441 (72.6)
166 (27.3)
36 (17.3)
117 (56.3)
55 (26.4)
54 (13.5)
175 (43.9)
170 (42.6)
90 (14.8)
292 (48.1)
225 (37.1)
103 (49.6)
93 (44.7)
12 (5.7)
144 (36.1)
195 (48.9)
60 (15.0)
247 (40.7)
288 (47.4)
72 (11.9)
49 (23.6)
159 (76.4)
68 (17.0)
331 (83.0)
117 (19.3)
490 (80.7)
10 (4.8)
9 (4.3)
40 (19.2)
92 (44.2)
57 (27.4)
77 (19.3)
55 (13.8)
110 (27.6)
113 (28.3)
44 (11.0)
87 (14.3)
64 (10.5)
150 (24.7)
205 (33.8)
101 (16.6)
14 (6.7)
41 (19.7)
40 (19.2)
45 (21.6)
68 (32.7)
107 (26.8)
100 (25.1)
99 (24.8)
53 (13.3)
40 (10.0)
121 (19.9)
141 (23.2)
139 (23.0)
98 (16.1)
108 (17.8)
x2
p 5 0.30
p 5 0.67
p , 0.05
p , 1023
p , 1023
p 5 0.05
p , 1024
p , 1024
1Low, no university-entrance qualification; high, at least university-entrance qualification.–2Living with
partner, married or cohabitating; living alone, unmarried, divorced or widowed.
Calvados, other spirits (rum, cognac, kirsch, fruit brandies, etc.).
For Calvados and other spirits, we distinguished the amount
consumed together with a hot beverage (coffee or grog) and the
amount consumed alone and cold. Intake of alcoholic beverages
was later transformed by computer into grams of ethanol. A specific
ethanol concentration was estimated for each of type of alcoholic
beverage, from 80 to 110 g/l for wines, 40 g/l for beer, 40 g/l for
cider, 360 g/l for aniseed aperitifs, 140 g/l for other aperitifs, 320
g/l for whisky, 360 g/l for rum, 400 g/l for home-made Calvados
and other spirits and 320 g/l for factory-made Calvados and other
spirits.
Unconditional logistic regression was used to estimate odds
ratios (OR) and 95% confidence intervals (CI) for each type of
alcoholic beverage (Breslow and Day, 1980). As the consumption
of one alcoholic beverage is often correlated with another, forward
step-wise procedures were used to construct multivariate models of
risk, eliminating variables which no longer had any effect when
adjusted for the others. In these analyses, variables were treated as
categorical. The effect of each alcoholic beverage was studied after
adjustment for total alcohol consumption and also as a proportion
of total alcohol consumption. As the place of residence, occupation,
level of education, marital status and smoking habits differed
significantly between cases and controls (Table I), all OR were
adjusted for age, interviewer, place of residence, occupation, level
of education, marital status and smoking habits.
The extent to which the relative difference in incidence between
2 areas could be explained by the difference in pattern of exposure
to a particular alcoholic beverage was derived from the relative
attributable risk (RAR) calculated as suggested by Breslow and
Day (1980). Briefly,
RAR 5 R*(1 2 w)/(R 2 1), with w
5 (P11r11 1 P12r11 1 . . . Plirli)/(P21r21 1 P22r22 1 . . . P2ir2i),
where Pji and rji are respectively the proportion of population j and
the relative risk at the ith level of exposure, and R is the ratio of
overall incidence between the 2 populations.
Defined as above, RAR reflects the extent to which excess
incidence between two populations would be reduced if population
2 had the same distribution of risk factor as population 1. The
incidence ratios R used were those provided by data from cancer
registries (Launoy et al., 1994), i.e., 2.5 between Normandy (Caen)
and Burgundy (Dijon) and 4.8 between Normandy and MidiPyrenées (Toulouse). Similarly, the incidence ratio between the
ALCOHOL AND CANCER OF THE OESOPHAGUS
919
TABLE II – BEER, WINE, CIDER, TOTAL ALCOHOL CONSUMPTION AND RISK OF OESOPHAGEAL CANCER
Alcohol
consumption
(g/week)
Cases/
controls
Model 01
OR1 (95% CI)
and p for trend x2
Model 12
ORa (95% CI)
and p for trend x2
Model 23
ORa (95% CI)
and p for trend x2
Total alcohol
1–150
151–300
301–450
451–600
601 and more
14/107
41/100
40/99
45/53
68/40
1
2.85 (1.36–5.96)
2.39 (1.14–4.99)
4.42 (2.03–9.58)
10.21 (4.68–22.05)
p , 1024
not tested in
this model
1
2.33 (1.09–4.98)
1.55 (0.71–3.40)
2.68 (1.19–6.06)
5.07 (2.16–11.92)
NS
Beer
Non-drinker
1–50
51–100
101–150
151–200
201 and more
69/176
56/137
23/34
17/19
10/8
33/25
1
0.98 (0.60–1.59)
1.61 (0.82–3.17)
2.57 (1.16–5.72)
2.33 (0.79–6.91)
3.34 (1.69–6.57)
p , 1024
1
1.06 (0.62–1.82)
1.45 (0.69–3.05)
2.77 (1.16–6.61)
1.34 (0.38–4.73)
2.95 (1.39–6.30)
p , 1022
1
1.05 (0.62–1.78)
1.62 (0.78–3.37)
2.66 (1.11–6.39)
1.40 (0.41–4.81)
2.38 (1.11–5.08)
p , 1022
Red wine
Non-drinker
1–100
101–200
201–300
301–400
401–500
501 and more
10/30
35/125
27/74
35/54
24/39
28/30
49/47
1
0.82 (0.33–1.99)
1.48 (0.56–3.81)
2.11 (0.84–5.32)
2.38 (0.89–6.36)
3.23 (1.19–8.77)
3.31 (1.26–7.86)
p , 1024
1
0.87 (0.33–2.32)
1.42 (0.51–3.93)
2.27 (0.82–6.27)
2.41 (0.83–6.98)
2.79 (0.95–8.20)
2.87 (1.03–7.21)
p , 1023
1
0.96 (0.34–2.71)
1.43 (0.48–4.26)
2.08 (0.68–6.35)
2.75 (0.80–9.42)
2.31 (0.65–8.23)
1.87 (0.59–5.17)
NS
White wine
Non-drinker
1–20
21–40
41 and more
144/268
30/84
11/18
23/29
1
0.54 (0.31–1.01)
0.98 (0.40–2.83)
1.08 (0.54–2.18)
NS
1
0.64 (0.37–1.26)
1.02 (0.35–3.71)
0.70 (0.32–1.55)
NS
1
0.58 (0.34–1.09)
1.19 (0.38–4.17)
0.58 (0.27–1.27)
NS
Cider
Non-drinker
1–100
101–200
201–300
301–400
401 and more
95/204
31/73
25/42
29/48
8/10
20/22
1
1.25 (0.62–2.50)
1.53 (0.69–3.39)
1.52 (0.68–3.40)
1.89 (0.55–6.51)
1.92 (0.74–4.99)
NS
1
1.39 (0.64–3.04)
1.52 (0.63–3.67)
1.51 (0.60–3.78)
2.01 (0.48–8.40)
1.72 (0.58–5.17)
NS
1
1.69 (0.79–3.61)
1.59 (0.66–3.81)
1.42 (0.57–3.52)
1.37 (0.34–5.57)
0.89 (0.29–2.76)
NS
1Model 0 includes interviewer, age, place of residence, occupation, level of education, marital status and
smoking habits.–2Model 1 includes interviewer, age, place of residence, occupation, level of education,
marital status, smoking habits, beer, red wine, aniseed aperitifs, whisky, hot Calvados and other hot
spirits.–3Model 2 includes interviewer, age, place of residence, occupation, level of education, marital
status smoking habits, total alcohol intake (forced into the model), beer, aniseed aperitifs, whisky and hot
Calvados.
rural and the urban populations in Normandy was taken to be 1.5
(Pottier et al., 1989).
RESULTS
Tables II and III show the effect of total alcohol consumption and
of each type of alcoholic beverage on the risk of oesophageal
cancer in 3 different models. In model 0, each alcoholic-beverage
effect was studied after adjustment for age, interviewer, place of
residence, occupation, level of education, marital status and
smoking habits. The OR was increased over 2 for mean weekly
intake of alcohol over 200 g/week for red wine, 100 g/week for
beer, 40 g/week for hot Calvados, 20 g/week for aniseed aperitifs
and 10 g/week for other hot spirits. Whisky seemed to be protective
for relatively low mean weekly intake. There was no significant
increase in risk with increasing consumption of cider or of white
wine.
Multivariate analysis was conducted to test whether the effect of
a given beverage was due to correlation with another beverage or
with total alcohol intake. The effect of each alcoholic beverage was
studied after adjustment for other alcoholic beverages, without
(model 1) or with (model 2) adjustment for total alcohol consump-
tion. Aniseed aperitifs, hot Calvados and beer were significantly
and independently associated with increased risk in both models.
Whisky was significantly associated with lower risk in both
models. Red wine was associated with increased risk in model 1,
but no longer after adjustment for total alcohol intake (model 2). In
fact, since red wine represented the largest proportion of alcohol
(56.6% of total alcohol in controls), these 2 variables were highly
correlated. Hot spirits excluding Calvados were significantly
associated with increased risk in model 1 and in a borderline
significant way in model 2 ( p 5 0.13). Moreover, when all spirits,
excluding Calvados, taken as grog, i.e., with hot water, were
distinguished from those drunk with coffee, their association with
increased risk remained significant even in model 2 ( p , 0.05).
Cider, white wine, non-aniseed aperitifs and cold spirits (Calvados
and other) had no effect in either model. In summary, after
adjustment for age, interviewer, place of residence, occupation,
level of education, marital status, smoking habits and total alcohol
intake, only hot Calvados, aniseed aperitifs and beer were independently associated with increasing risk, while whisky was associated
with decreasing risk.
Similar analyses were conducted, with the consumption of each
beverage being expressed as the proportion of total alcohol
LAUNOY ET AL.
920
TABLE III – APERITIFS, SPIRITS AND RISK OF OESOPHAGEAL CANCER
Model 01
OR1 (95% CI)
and p for
trend x2
Model 12
ORa (95% CI)
and p for
trend x2
Model 23
ORa (95% CI)
and p for
trend x2
53/185
60/114
29/45
17/16
49/39
1
1.24 (0.76–2.03)
1.51 (0.81–2.82)
2.83 (1.23–6.51)
3.74 (2.08–6.74)
p , 1024
1
1.36 (0.79–2.34)
1.12 (0.56–2.25)
2.07 (0.84–5.09)
2.74 (1.44–5.23)
p , 1022
1
1.34 (0.78–2.30)
1.17 (0.60–2.32)
2.24 (0.91–5.54)
2.68 (1.41–5.10)
p , 1022
Whisky
Non-drinker
1–10
11–20
21 and more
180/285
15/48
5/32
8/34
Other aperitifs 4
Non-drinker
1–5
6–10
11 and more
1
0.47 (0.24–0.94)
0.30 (0.11–0.85)
0.48 (0.20–1.13)
p , 1022
1
0.47 (0.22–0.99)
0.23 (0.07–0.76)
0.41 (0.15–1.08)
p , 1022
1
0.48 (0.23–1.02)
0.23 (0.07–0.75)
0.36 (0.14–0.95)
p , 1022
179/311
10/58
10/16
9/14
1
0.31 (0.14–0.68)
1.19 (0.48–2.94)
1.03 (0.40–2.67)
NS
1
0.28 (0.12–0.69)
1.16 (0.41–3.27)
0.81 (0.27–2.47)
NS
1
0.28 (0.12–0.66)
1.06 (0.37–3.04)
1.02 (0.35–2.95)
NS
Cold calvados5
Non-drinker
1–5
6 and more
195/369
9/21
4/9
1
0.82 (0.33–2.01)
1.02 (0.26–3.99)
NS
1
1.15 (0.42–3.12)
0.85 (0.20–3.62)
NS
1
1.01 (0.37–2.74)
0.86 (0.20–3.78)
NS
Hot Calvados6
Non-drinker
1–20
21–40
41 and more
124/303
24/46
8/9
52/41
Cold spirits7
Non-drinker
1–5
6–10
11 and more
1
1.45 (0.73–2.90)
1.77 (0.56–5.63)
3.64 (1.87–7.07)
p , 1023
1
1.50 (0.71–3.16)
1.25 (0.35–4.40)
3.24 (1.56–6.73)
p , 1022
1
1.40 (0.67–3.92)
1.40 (0.39–5.08)
2.33 (1.12–4.87)
p , 0.05
130/212
44/126
13/28
21/33
1
0.62 (0.39–0.99)
0.65 (0.30–1.38)
0.91 (0.47–1.80)
NS
1
0.70 (0.41–1.23)
0.57 (0.23–1.42)
0.70 (0.32–1.53)
NS
1
0.73 (0.42–1.27)
0.68 (0.28–1.62)
0.76 (0.36–1.61)
NS
Hot spirits8
Non-drinker
1–5
6–10
11 and more
163/345
14/29
5/8
26/17
1
1.02 (0.46–2.26)
1.79 (0.48–6.60)
2.83 (1.38–5.80)
p , 1022
1
0.81 (0.33–1.97)
1.40 (0.31–6.39)
2.31 (1.04–5.12)
p , 0.10
1
0.80 (0.33–1.94)
1.65 (0.37–7.33)
1.83 (0.91–4.31)
NS
Alcohol
consumption
(g/week)
Aniseed aperitifs
Non-drinker
1–10
11–20
21–30
31 and more
Cases/
controls
1Model 0 includes interviewer, age, place of residence, occupation, level of education, marital status and
smoking habits.–2Model 1 includes interviewer, age, place of residence, occupation, level of education,
marital status, smoking habits, beer, red wine, aniseed aperitifs, whisky, hot Calvados and other hot
spirits.–3Model 2 includes interviewer, age, place of residence, occupation, level of education, marital
status smoking habits, total alcohol intake (forced into the model), beer, aniseed aperitifs, whisky and hot
Calvados.–4All aperitifs excluding aniseed and whisky.–5Calvados drunk alone.–6Calvados drunk with
coffee.–7All spirits (excluding Calvados) drunk alone.–8All spirits (excluding Calvados) with hot water or
coffee.
consumption for that individual. Results were similar, the beverages still independently influencing risk in models 1 and 2 being
hot Calvados ( p , 0.05 in both models), beer (respectively p , 0.05
and p , 0.10), aniseed aperitifs ( p , 0.10 and p , 0.05 respectively), while whisky had a protective influence ( p , 0.10 in both
models). Expressed as a proportion of total alcohol, red wine was
not significant in either model.
Whisky showed a significantly decreased relative risk even at
consumption of less than 10 g a week. High whisky consumption
was linked to low tobacco consumption ( p , 0.05), high socioeconomic status ( p , 1023 ), urban residence ( p , 1022 ), high
level of education ( p , 1024 ), living with a partner ( p , 0.05) and
low consumption of hot Calvados. However, as shown in Table III,
adjustment for smoking, sociodemographic characteristics and
other alcoholic beverages had almost no effect on the OR associated with whisky consumption.
In order to explain the large geographical variations in incidence
of oesophageal cancer, the population attributable risk for total
alcohol consumption and for each of the alcoholic beverages
included in the final model 2 (hot Calvados, aniseed aperitifs, beer
and whisky) was calculated for each centre, together with the
corresponding relative attributable risk (RAR) for Caen vs. Dijon
and for Caen vs. Toulouse. The differences in type of alcoholic
beverages consumption in controls between the 3 centres are
shown in Tables IV and V.
Attributable risk and relative attributable risk are shown in Table
VI. They were estimated using relative risk adjusted for age,
interviewer and tobacco (adjustment A) and using relative risk
previously provided by model 2 (adjustment B). The population
attributable risk for total alcohol intake for each centre was close to
each other whatever the adjustment and, as a consequence, the
corresponding RAR was moderate for Caen/Dijon (19% or 12% for
adjustment A and B respectively) or nil for Caen/Toulouse.
Consumption of hot Calvados among controls was virtually limited
to Caen, so that the highest difference in attributable risk between
Caen and other centres was related to hot Calvados from 0% in
ALCOHOL AND CANCER OF THE OESOPHAGUS
TABLE IV – BEER, WINE, CIDER AND TOTAL ALCOHOL CONSUMPTION AMONG
CONTROLS IN THE 3 CENTRES
Alcohol
consumption
(g/week)
Total alcohol
0
1–150
151–300
301–450
451–600
601 and more
Beer
Non-drinker
1–50
51–100
101–150
151–200
201 and more
Red wine
Non-drinker
1–100
101–200
201–300
301–400
401–500
501 and more
White wine
Non-drinker
1–20
21–40
41 and more
Cider
Non-drinker
1–100
101–200
201–300
301–400
401 and more
Caen
N (%)
(n 5 203)
Dijon
N (%)
(n 5 90)
Toulouse
N (%)
(n 5 106)
0 (0)
51 (25.1)
56 (27.6)
51 (25.1)
22 (10.8)
23 (11.3)
2 (2.2)
28 (31.1)
27 (30.0)
16 (17.8)
9 (10.0)
8 (8.9)
4 (3.8)
22 (20.8)
17 (16.0)
32 (30.2)
22 (20.8)
9 (8.5)
72 (35.5)
79 (38.9)
17 (8.4)
12 (5.9)
5 (2.5)
18 (8.9)
30 (33.3)
43 (47.8)
9 (10.0)
3 (3.3)
2 (2.2)
3 (3.3)
74 (69.8)
15 (14.2)
8 (7.5)
4 (3.8)
1 (0.9)
4 (3.8)
17 (8.3)
86 (42.3)
38 (18.7)
24 (11.8)
14 (6.9)
3 (1.5)
21 (10.3)
4 (4.4)
26 (28.9)
19 (2.1)
16 (17.8)
8 (8.9)
8 (8.9)
9 (10.0)
9 (8.5)
13 (12.3)
17 (16.0)
14 (13.2)
17 (16.0)
19 (17.9)
17 (16.0)
156 (76.8)
28 (13.8)
3 (1.5)
16 (7.9)
15 (16.7)
48 (53.3)
14 (15.6)
13 (14.4)
97 (91.5)
8 (7.5)
1 (0.9)
0 (0)
31 (15.3)
56 (27.6)
39 (19.2)
45 (22.2)
10 (4.9)
22 (10.8)
68 (75.6)
16 (17.8)
3 (3.3)
3 (3.3)
0 (0)
0 (0)
105 (99.1)
1 (0.9)
0 (0)
0 (0)
0 (0)
0 (0)
921
TABLE V – APERITIFS AND SPIRITS CONSUMPTION AMONG CONTROLS
IN THE 3 CENTRES
Alcohol
consumption
(g/week)
Aniseed aperitifs
Non-drinker
1–10
11–20
21–30
31 and more
Whisky
Non-drinker
1–10
11–20
21 and more
Other aperitifs1
Non-drinker
1–5
6–10
11 and more
Cold Calvados2
Non-drinker
1–5
6 and more
Hot Calvados3
Non-drinker
1–20
21–40
41 and more
Cold spirits4
Non-drinker
1–5
6–10
11 and more
Hot spirits5
Non-drinker
1–5
6–10
11 and more
Caen
N (%)
(n 5 203)
Dijon
N (%)
(n 5 90)
Toulouse
N (%)
(n 5 106)
93 (45.7)
60 (29.8)
19 (9.3)
18 (8.9)
13 (6.4)
34 (37.8)
31 (34.3)
11 (12.2)
8 (8.9)
6 (6.7)
58 (54.7)
23 (21.7)
15 (14.2)
5 (4.7)
5 (4.7)
143 (70.4)
26 (12.8)
16 (7.9)
18 (8.9)
58 (64.4)
18 (20.0)
8 (8.9)
6 (6.7)
84 (79.2)
4 (3.8)
8 (7.5)
10 (9.4)
170 (83.7)
20 (9.9)
10 (4.9)
3 (1.5)
45 (50.0)
33 (36.7)
4 (4.4)
8 (8.9)
96 (90.6)
5 (4.7)
2 (1.9)
3 (2.8)
186 (91.6)
11 (5.4)
6 (3.0)
79 (87.8)
8 (8.9)
3 (3.3)
104 (98.1)
2 (18.9)
0 (0)
109 (53.7)
45 (22.2)
9 (4.4)
40 (19.7)
88 (97.8)
1 (1.1)
0 (0)
1 (1.1)
106 (100.0)
0 (0)
0 (0)
0 (0)
116 (57.1)
67 (33.0)
8 (3.9)
12 (5.9)
27 (30.0)
40 (44.4)
11 (12.2)
12 (13.3)
69 (65.1)
19 (17.9)
9 (8.5)
9 (8.5)
181 (89.1)
9 (4.4)
4 (2.0)
9 (4.5)
61 (67.8)
20 (22.2)
4 (4.4)
5 (5.6)
103 (97.2)
0 (0)
0 (0)
3 (2.8)
1All aperitifs excluding aniseed and whisky.–2Calvados drunk alone.–
Toulouse and 4% in Dijon to 47% in Caen with adjustment A,
corresponding figures being 0, 2% and 27% with adjustment B.
RAR to hot Calvados was thus very high: using adjustment A, 74%
of the difference in incidence between Caen and Dijon and 59% of
the difference between Caen and Toulouse could be explained by
the drinking of hot Calvados. Even after adjustment for all
significant alcoholic beverages and total alcohol, these figures were
still high: 43% and 34% respectively. As regards beer and aniseed
aperitifs, attributable risk for each centre was not so far to each
other, so RAR was moderate. For whisky, attributable risk was
negative and RAR was very low.
Similarly, in order to explain the rural/urban variations in
incidence of oesophageal cancer in the northwest of France, the
risk attributable to each alcoholic beverage for urban and rural
populations of the centre of Caen was calculated, together with
corresponding RAR. Risk attributable to total alcohol intake was
the same for urban and rural populations (56% for adjustment A,
29% for adjustment B) and RAR rural/urban was 20.10 (adjustment A) or 20.03 (adjustment B), suggesting that difference in
patterns of exposure to total alcohol was unable to explain the
higher incidence in the rural population. As before, the highest
difference in attributable risk was found for hot Calvados, with
38% of the rural population and 20% of the urban population
drinking more than 40 g a week. Therefore the relative attributable
risk to hot Calvados for the rural population vs. the urban
population was 72% for adjustment A, 41%, for adjustment B.
DISCUSSION
Our results show that the well-established link between the risk
of oesophageal cancer and alcohol depends on the type of alcoholic
beverage, with aniseed aperitifs, hot spirits (especially hot Calva-
3Calvados
drunk with coffee.–4All spirits (excluding Calvados) drunk
alone.–5All spirits (excluding Calvados) drunk with hot water or coffee.
dos) and beer carrying the highest risk in this study. They strongly
suggest that the risk of oesophageal cancer associated with
alcoholic drinks may be strongly potentiated if the drinks are
consumed hot, the consumption of hot Calvados or other hot spirits
being markedly associated with risk, whilst consumption of the
same spirits cold had no influence. The role of hot drinks and hot
diet on the risk of oesophageal cancer has been suggested from
many parts of world: hot coffee in Puerto Rico (Martinez, 1969),
hot tea and chagayu in Asia (De Jong et al., 1974; Segi, 1975;
Ghadirian, 1987; Hu et al., 1994) and the former USSR (Kolicheva,
1980), while only one study showed no effect of hot carbonated
beverages, coffee or tea (Pottern et al., 1981). Several case-control
studies have established the role of hot maté in oesophageal cancer
in South America (Vassalo et al., 1985; Victora et al., 1987; De
Stefani et al., 1990). Maté drinkers also had a 2.2-fold higher risk
of developing oesophagitis than non-maté drinkers (Muñoz et al.,
1987). Thermal injury is acknowledged as a possible predisposing
factor, like chemical injury and physical abrasion (Pera et al.,
1987), and experimentally it has been shown to increase the risk of
induced oesophageal cancer (Yioris et al., 1984). Hot drinks could
also facilitate the transport of carcinogens across the epithelium
(Cheng, 1994).
Our study shows that aniseed aperitifs carry a very high risk for
oesophageal cancer, probably because this kind of aperitif is not
consumed in most countries. Moreover, other French studies did
not distinguish aniseed aperitifs from other aperitifs, and showed
that aperitifs appeared to have less effect than other beverages
(Tuyns et al., 1979; Breslow and Day, 1980). Our results point to
the possible responsibility of a component unique to aniseed
LAUNOY ET AL.
922
TABLE VI – ATTRIBUTABLE RISK FOR HOT CALVADOS, ANISEED APERITIFS,
BEER, WHISKY AND TOTAL ALCOHOL CONSUMPTION ACCORDING TO
CENTRE AND CORRESPONDING RELATIVE ATTRIBUTABLE RISK
Relative
attributable risk1
Attributable risk
Caen
Total alcohol
Adjustment A2
Adjustment B3
Hot Calvados
Adjustment A2
Adjustment B3
Aniseed aperitifs
Adjustment A2
Adjustment B3
Beer
Adjustment A2
Adjustment B3
Whisky
Adjustment A2
Adjustment B3
Dijon
Toulouse
Caen vs.
Dijon
Caen vs.
Toulouse
0.56
0.29
0.51
0.25
0.57
0.27
0.19
0.12
0.00
0.06
0.47
0.27
0.04
0.02
0
0
0.74
0.43
0.59
0.34
0.37
0.26
0.39
0.27
0.30
0.20
20.05
20.03
0.12
0.10
0.26
0.23
0.17
0.16
0.16
0.15
0.18
0.14
0.15
0.13
20.25
20.22
20.30
20.27
20.17
20.16
0.07
0.06
20.08
20.07
1Calculated using incidence ratio of 2.5 and 4.8 respectively for
Caen/Dijon and Caen/Toulouse.–2Calculated using relative risk adjusted for age, interviewer and smoking.–3Calculated using relative
risk from model 2 including interviewer, age, place of residence,
occupation, level of education, marital status, smoking habits, total
alcohol intake (forced into the model), beer, aniseed aperitifs, whisky
and hot Calvados.
aperitifs. The results of experimental studies are conflicting, but
trans-anethol, a flavouring substance present in the essential oils of
a variety of plants, is a candidate (Truhaut et al., 1989; Newberne et
al., 1989). The independent effect of beer is the most consistent
result in all studies, whether it is the most common alcoholic
beverage (Segal et al., 1988; Adelhart et al., 1985) or not (Pottern et
al., 1981; Mettlin et al., 1981; Wynder and Bross, 1961). In 3
counties in Western New York, Graham et al. (1990) found that
beer carried a higher risk than total alcohol ingestion. As regards
red wine, the adjustment on total alcohol completely removed its
effect on risk. To our knowledge, the specific effect of wine was not
significant in only one other study (Graham et al., 1990). In view of
the high correlation between red wine and total alcohol consumption in our study, and of conflicting results in several studies (Barra
et al., 1990; Pottern et al., 1981; Tuyns et al., 1979; Breslow and
Day, 1980; Wynder and Bross, 1961), a specific effect of red wine
on risk of oesophageal cancer cannot be ruled out. The protective
effect of whisky was puzzling. Although whisky was correlated
with high social status and low total alcohol intake, the lower risk
associated with whisky persisted after adjustment for sociodemographic characteristics, consumption of other beverages and total
alcohol intake. Whisky consumption is probably a marker for other
non-alcoholic variables, possibly dietary, which will be investigated in a future paper.
Since Normandy is the biggest French apple-production area,
apple-derived alcoholic beverages, i.e., cider and Calvados (apple
brandy) have been suspected of being responsible for the high
incidence of oesophageal cancer in this region. We found no
association between cider consumption and the risk of oesophageal
cancer even after adjustment for other alcoholic drinks. For
Calvados, an increase in risk was only observed when the Calvados
was drunk with a hot beverage, as other spirits. Only one
epidemiological study, conducted in the department of Calvados,
has suggested an additional risk related to consumption of cider,
with a higher risk among consumers of cider (5.74) than among
non-consumers (1.80) (Tuyns et al., 1979). This study also
suggested that Calvados carried a higher risk than other beverages,
but Calvados consumption was not distinguished from other spirits.
In the analysis of data from the Ille et Villaine region (Brittany),
which is also an area producing apple-derived beverages, no
difference was observed in the effect of different beverages
(Breslow and Day, 1980). Mutagenicity tests on apple-derived
beverages showed a weak effect that could be attributed neither to
nitrosamines nor to polycyclic aromatic hydrocarbons, while the
search for other specific carcinogens was not conclusive (Loquet et
al., 1981). Yamada et al. (1992) showed that formation of
06-methyldeoxyguanosine from N-nitrosomethylbenzylamine in
rats was higher after treatment with apple-brandy and red burgundy
than with other beverages. In our study, most Calvados consumed
was home-made. Only 15 cases and 15 controls drank hot
factory-made Calvados, and 4 cases and 14 controls cold factorymade Calvados. We were thus unable to compare the effects of
home-made and factory-made Calvados.
Our study suggests that the custom of consuming hot alcoholic
drinks may underlie the striking pattern of incidence of the disease
in France. Our results suggest that 2⁄3 of the high incidence of
oesophageal cancer in the west of France and in rural populations
could be due to the specific habit of drinking hot beverages,
especially hot Calvados, whereas variation in total alcohol consumption could not explain any difference between urban and rural
populations, and could explain less than 20% of the high incidence
in the northwest of France. Even after adjustment for all other
alcoholic beverages, consumption of hot Calvados explained
almost half of the peak in incidence of oesophageal cancer in the
northwest of France, as well as half of the urban/rural difference in
incidence. The high proportion of controls with a high intake of hot
Calvados in Caen may seem surprising. Nowadays, hot Calvados is
no longer drunk to such an extent, and a survey of current drinking
data would not produce the same results as our study, in which the
entire alcohol history of each person was reconstituted. In former
times, especially because a large part of the population came from a
rural background, drinking of hot Calvados was a very widespread
custom in Normandy. It is possible that the fall in oesophagealcancer incidence in the last 20 years in the west of France (Launoy
et al., 1994) is a result from a decline in this custom.
ACKNOWLEDGEMENTS
We are very grateful to Dr. J. Estève for his very helpful advice
on data analysis, to Dr. C. Milan for his valuable assistance in data
processing, and to the dieticians Mrs. B. Coudray, Mrs. C.
Desprat-Belghiti, Ms. I. Duvelleroy, Ms. Z. El Yaagoubi, Mrs. N.
Hue, Mrs. A. Leclerc and Mrs. M. Remy.
REFERENCES
ADELHARDT, M., MOLLER, J.O. and SAND, H.H., Cancer of the larynx,
pharynx, and oesophagus in relation to alcohol and tobacco consumption
among Danish brewery workers. Dan. Med. Bull., 32, 119–123 (1985).
BRESLOW, N.E. and DAY, N.E., Statistical methods in cancer research. IARC
Scientific Publication No 32. Vol 1, IARC, Lyon (1980).
CHENG, K.K., The etiology of esophageal cancer in Chinese. Semin. Oncol.,
21, 411–415 (1994).
BARRA, S., FRANCESCHI, S., NEGRI, E., TALAMINI, R. and LA VECCHIA, C.,
Type of alcoholic beverage and cancer of the oral cavity, pharynx and
oesophagus in an Italian area with high wine consumption. Int. J. Cancer,
46, 1017–1020 (1990).
DE JONG, U.W., BRESLOW, N., GOTH EWE, H., SRIDHARAN, M. and
SHANMUGARATNAM, K., Aetiological factors in oesophageal cancer in
Singapore Chinese. Int. J. Cancer, 13, 291–303 (1974).
BENHAMOU, E., LAPLANCHE, A., WARTELLE, M., SHAFFER, P., GIGNOUX, M.,
ROBILLARD, J., FAIVRE, J. and MENEGOZ, F., Incidence des cancers en France
1978–1982. INSERM, Paris (1990).
DE STEFANI, E., MUÑOZ, N., ESTÈVE, J., VASALLO, A., VICTORA, C.G. and
TEUCHMANN, S. Maté drinking, alcohol, tobacco, diet, and esophageal
cancer in Uruguay. Cancer Res., 50, 426–431 (1990).
ALCOHOL AND CANCER OF THE OESOPHAGUS
GHADIRIAN, P., Thermal irritation and esophageal cancer in Northern Iran.
Cancer, 60, 1909–1914 (1987).
GRAHAM, S., MARSHALL, J., HAUGHEY, B., BRASURE, J., FREUDENHEIM, J.,
ZIELEZNY, M., WILKINSON, G. and NOLAN, J., Nutritional epidemiology of
cancer of the esophagus. Amer. J. Epidemiol., 131, 454–467 (1990).
HU, J. and 15 OTHERS, Risk factors for oesophageal cancer in northeast
China. Int. J. Cancer, 57, 38–46 (1994).
KOLICHEVA, N.I., Epidemiology of oesophageal cancer in the USSR. In:
D.L. Levin (ed.), Cancer epidemiology in the USA and USSR, pp. 191–197,
NIH, Washington (1980).
LAUNOY, G., FAIVRE, J., PIENKOWSKI, P., MILAN, CH., GIGNOUX, M. and
POTTIER, D., Changing pattern of oesophageal cancer incidence in France.
Int. J. Epidemiol., 23, 246–251 (1994).
LAUNOY, G., MILAN, CH., FAIVRE, J., PIENKOWSKI, P., MILAN, CL. and
GIGNOUX, M., Alcohol, tobacco and oesophageal cancer. Effects of the
duration of consumption, mean intake, current and former consumption.
Brit. J. Cancer (1997). (In press).
LOQUET, C., TOUSSAINT, G. and LE TALAER, J.Y., Studies on mutagenic
constituents of apple brandy and various alcoholic beverages collected in
western France, a high incidence area for oesophageal cancer. Mutation
Res., 32, 155–164 (1981).
MARTINEZ, I., Factors associated with cancer of the esophagus, mouth and
pharynx in Puerto Rico. J. nat. Cancer Inst., 42, 1069–1094 (1969).
METTLIN, C., GRAHAM, S., PRIORE, R., MARSHALL, J. and SWANSON, M., Diet
and cancer of the esophagus. Nutr. Cancer, 2, 143–147 (1981).
MUÑOZ, N., VICTORA, C.G., CRESPI, M., SAUL, C., BRAGA, N.M. and
CORREA, P., Hot maté drinking and pre-cancerous lesions of the oesophagus: an endoscopic survey in Southern Brazil. Int. J. Cancer, 39, 708–709
(1987).
NEWBERNE, P.M., CARLTON, W.W. and BROWN, W.R., Histopathological
evaluation of proliferative liver lesions in rats fed trans-anethole in chronic
studies. Food chem. Toxicol., 27, 21–26 (1989).
PERA, M., CARDESA, A., PERA, C. and MOHR, U., Nutritional aspects in
oesophageal carcinogenesis. Anticancer Res., 7, 301–308 (1987).
923
POTTERN, L., ZIEGLER, R.G., MORRIS, L., BLOT, W., HOOVER, R. and
FRAUMENI, J.F., Esophageal cancer among black men in Washington, D.C. I.
Alcohol, tobacco, and other risk factors. J. nat. Cancer Inst., 67, 777–783
(1981).
POTTIER, D., LAUNOY, G., CHERIE, L., CHUBERRE, C. and GIGNOUX, M., Le
cancer de l’oesophage dans le département du Calvados. Facteurs d’inégalités
géographiques et sociales. Bull. Cancer, 76, 111–119 (1989).
SEGAL, I., REINACH, S.G. and DE BEER, M., Factors associated with
oesophageal cancer in Soweto, South Africa. Brit. J. Cancer, 58, 681–686
(1988).
SEGI, M., Tea-gruel as a possible factor for cancer of the esophagus. Gann,
66, 199–202 (1975).
TRUHAUT, R., LE BOURHIS, B., ATTIA, M., GLOMOT, R., NEWMAN, J. and
CALDWELL, J., Chronic toxicity/carcinogenicity study of trans-anethole in
rats. Food chem. Toxicol., 27, 11–20 (1989).
TUYNS, A., PÉQUIGNOT, G. and ABBATUCCI, J.S., Oesophageal cancer and
alcohol consumption: importance of type of beverage. Int. J. Cancer, 23,
443–447 (1979).
VASSALLO, A., CORREA, P., DE STEFANI, E., CENDAN, M., ZAVALA, A., CHEN,
V., CARZOGLIO, J. and DENEO-PELLEGRINI, H., Esophageal cancer in
Uruguay: a case control study. J. nat. Cancer Inst., 75, 1005–1009 (1985).
VICTORA, C.G., MUÑOZ, N., DAY, N.E., BARCELOS, L.B., PECCIN, D.A. and
BRAGA, N.M., Hot beverages and oesophageal cancer in southern Brazil: a
case-control study. Int. J. Cancer, 39, 710–716 (1987).
WYNDER, E.L. and BROSS, I.J., A study of etiological factors in cancer of the
oesophagus. Cancer, 14, 389–413 (1961).
YAMADA, Y., WELLER, R.O., KLEIHUE, S. and LUDEKE, B.I., Effects of
ethanol and various alcoholic beverages on the formation of 0 6methyldeoxyguanosine from concurrently administered N-nitrosomethylbenzylamine in rats: a dose-response study. Carcinogenesis, 13, 1171–1175
(1992).
YIORIS, N., IVANKOVIC, S. and LEHNERT, T., Effect of thermal injury and oral
administration of N-methyl-N8-nitro-N-nitrosoguanidine on the development of esophageal tumors in Wistar rats. Oncol., 41, 36–38 (1984).
Документ
Категория
Без категории
Просмотров
2
Размер файла
56 Кб
Теги
121
1/--страниц
Пожаловаться на содержимое документа