FIBER INTAK E AND THE RISK OF ORAL, PHARYNGEALAND ESOPHAGEAL CANCERMaria SOLER

1, Cristina BOSETTI1* , Silvia FRANCESCHI

2, Eva NEGRI1, Paola ZAMBON

3, Renato TALAMINI4, Ettore CONTI

5 andCarlo LA VECCHIA

1,6

1Istituto di Ricerche Farmacologiche “Mari o Negri,” Milan, Italy2Field and Intervention Studies Unit, International Agency for Research on Cancer, Lyon Cedex, France3Servizio di Epidemiologia dei Tumori, Registro Tumori del Veneto, Padua, Italy4Servizio di Epidemiologia. Centro di Riferimento Oncologico, Aviano (PN), Italy5Servizio di Epidemiologia e Oncogenesi, Istituto “Regina Elena” per lo Studio e la Cura dei Tumori, Rome, Italy6Istituto di Statistica Medica e Biometria, Universita degli Studi di Milano, Milan, Italy

The relation between various types of fiber and oral, pharyn-geal and esophageal cancer was investigated using data from acase-control study conducted between 1992 and 1997 in Italy.Cases were 271 hospital patients with incident, histologicallyconfirmed oral cancer, 327 with pharyngeal cancer and 304 withesophageal cancer. Controls were 1,950 subjects admitted tothe same network of hospitals as the cases for acute, nonneo-plastic diseases. Cases and controls were interviewed duringtheir hospital stay using a validated food frequency question-naire. Odds ratios (OR) were computed after allowance for age,sex, and other potential confounding factors, including alcohol,tobacco consumption, and energy intake. The ORs for the high-est vs. the lowest quintile of intake of oral, pharyngeal andesophageal cancer combined were 0.40 for total (Englyst) fiber,0.37 for soluble fiber, 0.52 for cellulose, 0.48 for insoluble noncellulose polysaccharide, 0.33 for total insoluble fiber and 0.38for lignin. The inverse relation were similar for vegetable fiber(OR 5 0.51), fruit fiber (OR 5 0.60) and grain fiber (OR 5 0.56),and were somewhat stronger for oral and pharyngeal cancerthan for esophageal cancer. The ORs were similar for the twosexes and strata of age, education, alcohol and tobacco con-sumption, and total non-alcohol energy intake. Our study indi-cates that fiber intake may have a protective role on oral,pharyngeal and esophageal cancer.© 2001 Wiley-Liss, Inc.

Key words: fiber; esophageal cancer; oral cancer; pharyngealcancer

Consumption of whole grain foods has been inversely related tothe risk of oral and pharyngeal, as well as of esophageal cancer.1,2

The relative risk (RR) estimates of oral and pharyngeal cancer forthe highest level of whole grain bread and cereal intake were 0.7in a hospital-based case-control study including 227 cases of oraland pharyngeal cancer conducted in Southern United States,3 0.7for male and 0.5 for females in a population-based one of 871cases and 979 controls from various areas of the United States.4TheRRswere0.4 for the intermediateand 0.3 for thehighest levelof whole grain bread and pasta intake in an Italian case-controlstudy including 181 cancersof theoral cavity and pharynx and 316of the esophagus.5 In an American study of 275 cases of esopha-geal cancer and 275 controls,6 the RR was 0.7 for the highest levelof whole grain bread intake. In aSwiss study7 whole grain cerealswereapparently protectiveon the risk of cancersof theoral cavity,pharynx, esophagus and larynx, with RRs between 0.3 and 0.7,whereas refined cerealswereassociated with elevated risk. In moststudies, the inverse association between whole grain foods andcancer risk persisted after allowance for tobacco, alcohol and othermajor determinants of upper digestive tract neoplasms.1,2

Only scanty data are available on apotential role of fiber per seon the process of oral and esophageal carcinogenesis.

The inverse relation of whole grain food with oral and esoph-ageal cancer could be related to their fiber content. Whole grainfoods, however, are rich not only in fiber, but also in severalnutrients, such as antioxidants, indoles, phenolic compounds, in-cluding flavonoids, isothiocyanates and phytoestrogens, that areconsidered beneficial aspects of the human diet.8–10

It is, consequently, important to understand whether the apparentprotection observed of whole grain foods on upper digestive tractneoplasms is due to fiber, or to other components of these foods.

We have therefore considered the role of fiber, including spe-cific types of fiber, on the risk of upper digestive tract neoplasmsusing data from a case-control study conducted in Italy, whereinformation was available on alcohol, tobacco, total energy intakeand other major factors of potential interest.

MATERIAL AND METHODS

A hospital-based case-control study on cancers of the upperaerodigestive tract has been conducted between January 1992 andDecember 1997 in Italy.

For oral cavity and pharynx, cases were patients below the ageof 79 years with incident, histologically confirmed cancer, diag-nosed within one year before the interview in the major teachingand general hospitals of two Italian areas (the province of Porde-none, in Northeastern Italy and those of Rome and Latina, incentral Italy).11 Cases of lip, salivary glands and nasopharyngealcancers were not included in the present study. For esophagealcancer, cases were patients below age 79 years with incident,histologically confirmed diagnosis of squamous-cell carcinoma ofthe esophagus enrolled in the major hospitals of three Italian areas(the provinces of Padua, Pordenone, and the greater Milan area, innorthern Italy). A total of 271 patients with cancer of the oralcavity (219 men and 52 women), 327 with cancer of the pharynx(293 men and 34 women), and 304 with cancer of the esophagus(275 men and 29 women) were interviewed.

Controlswere1,950 subjects (1,374 men and 576 women), agedunder 79 years, with no history of cancer, admitted to majorhospitals in the same catchment areas as cases for acute, non-neoplastic conditions unrelated to smoking, alcohol drinking andlong-term modification of diet. Controls were frequency-matchedwith cases on age (5-year group), gender, year of interview andarea of residence. To compensate for the rarity of these neoplasmsin women, a greater control-to-case ratio was chosen for females.Thirty-two percent of controls were admitted for non-traumaticorthopaedic disorders, 26% for traumatic conditions, 19% foracute surgical conditions, and 23% for other miscellaneous ill-nesses (such as ear, nose and throat, skin or dental disorders). On

Grant sponsor: Italian Association for Cancer Research, Milan.

*Correspondence to: Istituto di Ricerche Farmacologiche “Mario Ne-gri,” Via Eritrea 62, 20157 Milan, Italy. Fax: 139 (02) 33200231.E-mail: bosetti@irfmn.mnegri.it

Received 24 July 2000; Revised 14 September 2000; Accepted 25September 2000

Int. J. Cancer: 91, 283–287 (2001)© 2001 Wiley-Liss, Inc.

Publication of the International Union Against Cancer

average, 5% of cases and 6% of controls approached during theirhospital stay refused to be interviewed.

The interviewer-administered questionnaire included informa-tion on sociodemographic characteristics, lifetime smoking andalcohol-drinking habits, physical activity, anthropometric mea-sures at various ages, a problem oriented medical history, andfamily history of cancer. A validated food frequency questionnairewas used to assess the usual diet during the 2 years preceding thediagnosis (for cases) or hospital admission (for controls), includingquestions on 78 specific foods or dishes divided into 6 sections: 1)bread, cereals, and first courses; 2) second courses (meat and othermain dishes); 3) side dishes (i.e., vegetables); 4) fruits; 5) sweets,desserts, and soft drinks; and 6) milk, hot beverages, and sweet-eners. For a few seasonal vegetables and fruits, consumption inseason and the corresponding duration were elicited. For 40 fooditems, the portion was defined in “natural” units (e.g., 1 teaspoonof sugar, one egg), and for the remaining items, it was defined assmall, average, or large with the help of pictures. The reproduc-ibility and validity of the food questionnaire were satisfactory.12,13

To compute total energy, nutrient, and fiber intake, Italian foodcomposition databases were used for about 80% of the food items,and integrated with other sources when needed.14

Dietary fiber intake was derived from published values based onthe Englyst procedure,15,16 that measures fiber as nonstarch poly-saccharides. A value was obtained for total fiber, soluble andinsoluble fiber. A modification of the method allows cellulose to bemeasured separately from insoluble fiber (non cellulose polysac-charide, NCP). Values for lignin, a minor component of the human

diet, were also provided separately. We did not include resistantstarch in the computation of total fiber, because its amount dependson the processing to which each food has been subjected, andrelated information and food composition tables were unavailable.Fiber intake was also divided according to the source,i.e., vege-table, fruit or cereal; vegetable fiber derived from 24 items of thequestionnaire, fruit fiber from 10 items and grain fiber from 25items.

Data analysisOdds ratio (OR) of oral and pharyngeal cancer, and of esopha-

geal cancer, for various measures of fiber intake were derivedusing unconditional multiple logistic regression models.17 Themodels included also terms for center, sex, age, education, alcoholand tobacco consumption, and non-alcohol energy intake, plus aglobal measure of vegetable and fruit consumption, when indicat-ed.17 The various types of fiber were entered in the models both asquintiles of the distribution of controls and as continuous vari-ables; in the latter case, the unit was the difference between the80th and 20th centile of the control distribution (i.e., between thelower cutpoints of the 5th and the 2nd quintiles).

RESULTS

Table I shows the distribution of cases and controls according togender, age, education, tobacco and alcohol consumption, vegeta-ble, fruit and non-alcohol energy intake. Cases were more oftenmale, and those with esophageal cancer were older than controls;

TABLE I – DISTRIBUTION OF 598 CASES OF ORAL AND PHARYNGEAL CANCER, 304 CASES OF ESOPHAGEAL CANCER AND 1950 CONTROLSACCORDING TO SELECTED COVARIATES, ITALY 1992–1997

Oral and pharyngeal cancer N (%) Esophageal cancer N (%) Controls N (%)

GenderMale 512 (85.6) 275 (90.5) 1,374 (70.5)Female 86 (14.4) 29 (9.5) 576 (29.5)

Age (years),50 133 (22.2) 29 (9.6) 454 (23.3)50–59 205 (34.3) 112 (36.8) 627 (32.2)60–69 211 (35.3) 119 (39.1) 656 (33.6)$70 49 (8.2) 44 (14.5) 213 (10.9)

Education (years),7 410 (68.6) 216 (71.0) 1,152 (59.1)7–11 130 (21.7) 65 (21.4) 525 (26.9)$12 58 (9.7) 23 (7.6) 273 (14.0)

Smoking status1

Never smokers 49 (8.2) 33 (10.9) 735 (37.7)Ex-smokers 153 (25.6) 109 (35.9) 667 (34.2)Current smokers

,15 cig/day 98 (16.4) 37 (12.2) 226 (11.6)$15 cig/day 298 (49.8) 124 (41.0) 320 (16.5)

Alcohol consumption (drinks/week)1

0–20 97 (16.3) 34 (11.2) 1,045 (53.7)21–34 76 (12.8) 50 (16.5) 436 (22.4)35–55 94 (15.7) 62 (20.5) 244 (12.6)56–83 165 (27.7) 85 (28.1) 148 (7.6)$84 164 (27.5) 72 (23.7) 72 (3.7)

Vegetable intake (portions/week),9 223 (37.3) 142 (46.7) 470 (24.1)9–11 105 (17.5) 50 (16.5) 421 (21.6)12–15 141 (23.6) 76 (25.0) 564 (28.9)$16 129 (21.6) 36 (11.8) 495 (25.4)

Fruit intake (portions/week),11 239 (40.0) 139 (45.7) 493 (25.3)12–15 122 (20.4) 59 (19.4) 449 (23.0)16–22 133 (22.2) 50 (16.5) 531 (27.2)$23 104 (17.4) 56 (18.4) 477 (24.5)

Non-alcohol energy intake (quartiles)I (lowest) 128 (21.4) 82 (27.0) 485 (24.8)II 152 (25.4) 68 (22.4) 490 (25.2)III 157 (26.3) 73 (24.0) 488 (25.0)IV (highest) 161 (26.9) 81 (26.6) 487 (25.0)

1The sum does not add up to the total because of some missing values.

284 SOLERET AL.

moreover they were somewhat less educated than controls andreported significantly higher tobacco (OR for the highest vs. thelowest level was 8.0 for oral and 4.2 for esophageal cancer) andalcohol consumption (OR 15.2 and 18.8). Vegetables intake wasless frequent in cases (OR 0.6 for oral and 0.5 for esophagealcancer), as well as fruit intake (OR 0.7 for oral and 0.8 foresophageal cancer), whereas non-alcohol energy intake was simi-lar for cases and controls (ORs 0.9 for oral and 0.9 for esophagealcancer). Allowance for these variables was therefore made in theanalyses.

Table II gives the ORs of oral, pharyngeal, and esophagealcancer and their combination according to intake of various typesof fiber, both in quintiles and as continuous variables. There was asignificant inverse association between total fiber intake, and itscomponents, and the risk of both oral and pharyngeal, and esoph-ageal cancer. The ORs of oral, pharyngeal and esophageal cancercombined, for a difference between the 80th and 20th centiles ofthe control distribution were 0.58 (95% CI 0.47–0.71) for totalfiber, 0.57 (95% CI 0.47–0.70) for soluble NCPs, 0.59 (95% CI0.48–0.71) for total insoluble fiber, 0.62 (95% CI 0.52–0.75) forcellulose, 0.63 (95% CI 0.52–0.78) for insoluble NCPs, and 0.60(95% CI 0.50–0.72) for lignin. The ORs were in the same direc-tion for the two cancers, but the inverse relation tended to beapparently stronger for oral and pharyngeal cancer (OR5 0.51 fortotal fiber) than for esophageal cancer (OR5 0.70). With regard tothe food source of the fiber, vegetable, fruit and cereal fiber wereprotective for the two cancers and the combined OR was 0.72 forvegetable, 0.78 for fruit, and 0.76 for grain fiber.

Table III presents the ORs of oral, pharyngeal and esophagealcancer for the various types of fiber separately in men and women.The OR for total fiber was 0.60 in men and 0.46 in women. Forseveral types of fiber considered the ORs were somewhat lower inwomen, although the pattern was less clear with reference to thesources of fiber.

TABLE II – ODDS RATIOS1 (OR) AND 95% CONFIDENCE INTERVALS (CI) OF ORAL, PHARYNGEAL AND ESOPHAGEAL CANCER ACCORDING TO THEINTAKE OF VARIOUS TYPES OF FIBER, ITALY 1992–1997

Type of fiberQuintile of intake2

xtrend2

Continuous OR (CI 95%)

Q2 Q3 Q4 Q5 P80–P20 unit

Total (Englyst) fiber (12.70) (15.48) (17.97) (21.74)Oral cavity and pharynx 0.55 0.41 0.48 0.36 18.3 0.51 (0.40–0.66)Esophagus 0.47 0.49 0.42 0.44 6.6 0.70 (0.51–0.96)Both 0.52 0.44 0.47 0.40 22.2 0.58 (0.47–0.71)Soluble NCP3 (6.25) (7.60) (8.80) (10.47)

Oral cavity and pharynx 0.53 0.40 0.43 0.27 26.7 0.48 (0.37–0.62)Esophagus 0.59 0.48 0.51 0.45 6.1 0.74 (0.54–1.01)Both 0.54 0.43 0.46 0.37 27.9 0.57 (0.47–0.70)

Total insoluble fiber (9.57) (11.68) (13.57) (16.22)Oral cavity and pharynx 0.57 0.41 0.46 0.32 23.4 0.51 (0.40–0.65)Esophagus 0.46 0.48 0.46 0.44 6.9 0.71 (0.52–0.96)Both 0.49 0.42 0.47 0.33 26.6 0.59 (0.48–0.71)Cellulose (3.19) (4.05) (4.78) (5.79)

Oral cavity and pharynx 0.68 0.46 0.44 0.51 15.9 0.58 (0.46–0.72)Esophagus 0.66 0.29 0.47 0.55 7.9 0.68 (0.51–0.91)Both 0.66 0.41 0.46 0.52 21.9 0.62 (0.52–0.75)

Insoluble NCP3 (3.12) (3.80) (4.46) (5.38)Oral cavity and pharynx 0.58 0.44 0.36 0.46 12.0 0.59 (0.46–0.76)Esophagus 0.55 0.44 0.41 0.47 6.6 0.70 (0.51–0.97)Both 0.56 0.44 0.43 0.48 16.8 0.63 (0.52–0.78)

Lignin (1.21) (1.47) (1.76) (2.15)Oral cavity and pharynx 0.46 0.39 0.47 0.25 24.0 0.46 (0.36–0.59)Esophagus 0.80 0.51 0.52 0.57 4.3 0.82 (0.62–1.10)Both 0.57 0.47 0.53 0.38 21.4 0.60 (0.50–0.72)

Vegetable fiber (4.22) (5.27) (6.25) (7.77)Oral cavity and pharynx 0.74 0.52 0.57 0.42 17.2 0.66 (0.53–0.81)Esophagus 0.74 0.46 0.81 0.73 1.3 0.81 (0.65–1.07)Both 0.75 0.56 0.61 0.51 16.5 0.72 (0.61–0.85)

Fruit fiber (3.08) (4.84) (6.31) (8.53)Oral cavity and pharynx 0.63 0.70 0.66 0.57 7.1 0.75 (0.62–0.90)Esophagus 0.53 0.57 0.56 0.63 3.1 0.82 (0.64–1.06)Both 0.63 0.66 0.65 0.60 9.1 0.78 (0.67–0.92)

Grain fiber (3.39) (4.32) (5.17) (6.63)Oral cavity and pharynx 0.95 0.70 0.85 0.48 6.3 0.70 (0.55–0.89)Esophagus 0.73 0.90 0.78 0.57 1.9 0.87 (0.65–1.17)Both 0.94 0.82 0.91 0.56 5.3 0.76 (0.63–0.93)

1Estimates from multiple logistic regression models including terms for center, sex, age, education, alcohol and tobacco consumption andnon-alcohol energy intake.–2Quintiles are computed on the distribution of controls. The reference category is the first (lowest) quintile. The lowercutpoint of the quintile (in grams) is given in parenthesis.–3Non cellulose polysaccharides.

TABLE III – ODDS RATIOS1 (OR) AND 95% CONFIDENCE INTERVALS (CI)OF ORAL, PHARYNGEAL AND ESOPHAGEAL CANCER ACCORDING TO

INTAKE OF VARIUS TYPES OF FIBER IN MEN AND WOMEN, ITALY1992–1997

Type of fiberOR1 (95% CI)

Men Women

Total (Englyst) fiber 0.60 (0.48–0.75) 0.46 (0.28–0.77)Soluble NCP2 0.61 (0.48–0.76) 0.41 (0.24–0.70)Total insoluble fiber 0.61 (0.49–0.76) 0.44 (0.27–0.73)

Cellulose 0.64 (0.52–0.79) 0.51 (0.32–0.80)Insoluble NCP2 0.63 (0.51–0.78) 0.64 (0.40–1.02)

Lignin 0.64 (0.52–0.79) 0.42 (0.26–0.70)Vegetable fiber 0.71 (0.59–0.85) 0.75 (0.50–1.16)Fruit fiber 0.83 (0.70–0.99) 0.58 (0.39–0.86)Grain fiber 0.72 (0.58–0.89) 0.99 (0.63–1.53)1Estimates from multiple logistic regression models including terms

of center, age, education, alcohol and smoking consumption, andnon-alcohol energy intake. The unit of intake is the difference betweento 80th and 20th percentiles of the distribution of controls (i.e.,between the lower cutpoints of the 5th and 2nd quintiles).–2Noncontrols cellulose polysaccharides.

285FIBER AND UPPER DIGESTIVE TRACT CANCER

In Table IV, the association between total fiber intake and therisk of oral, pharyngeal and esophageal cancer is considered instrata of selected covariates. No apparent heterogeneity was foundacross strata of age, but the inverse relation tended to be strongerin more educated subjects, in alcohol drinkers and in those whohad lower intake of non-alcohol energy.

DISCUSSION

Our study indicates that there is a strong inverse relation be-tween fiber consumption and oral, pharyngeal and esophagealcancer. The apparent protection was of similar magnitude forvarious types of fiber, including soluble, insoluble fiber or lignin,as well as for vegetable, fruit or grain fiber. The present resultsconfirm the findings of other studies of upper digestive tractneoplasms conducted in North America and Europe, dealingmainly with whole grain cereals,1–6 and add further evidence on atopic with relevant implications for prevention.

The interpretation of a potential protection of fiber on oral,pharyngeal and esophageal carcinogenesis is however not clear.Fiber are known to slow digestion and absorption of carbohydratesand the consequent glycemic response, thus reducing hyperinsu-

linemia and insulin-like growth factors (IGF).18,19 Fiber may alsobind carcinogens and thereby limit their contact with upper diges-tive tract epithelia.10 High-fiber foods, moreover, generally tend tohave a higher content of antioxidants.10

Fruit, vegetables, and whole grain cereals were three major foodgroups inversely related to upper digestive tract neoplasms in thispopulation, and the strength of the association was stronger thanfor most other digestive and non-digestive tract neoplasms, includ-ing cancers of colon, rectum and breast.5,11,20,21It is thus difficultto disentangle the potential effect of fiber from that of othercomponents of these foods. When we also included a term for fruitand vegetable consumption in the multivariate analysis, the ORsfor the highest level of total fiber intake were 0.41 for oral andpharyngeal cancer and 0.55 for esophageal cancer, suggesting thatthe protection is, at least in part, attributable to fiber itself, ratherthan to other components of foods rich in fiber. Because of thecollinearity between fruit, vegetables and fiber, it is howeverdifficult to understand whether, and how much, such a modelrepresents an over adjustment.

High fiber intake may simply be an indicator of a diet rich infruit, vegetables, and whole grains, and consequently poor inrefined cereals, meat and saturated fats, that have been associatedto elevated oral and esophageal cancer risk in the Italian popula-tion.11,21Fiber may also be an indicator of a better general lifestylehabit. However, if allowance for tobacco, alcohol, as well asselected dietary factors related to the risk of upper digestive tractneoplasms (such as fruit, vegetable, and meat consumption) wasunable to explain the association observed.

The present study, one of the largest case-control investigationsof the nutritional etiology of upper digestive tract cancers, hassome limitations as well as several strengths of the hospital-basedcase-control studies.17 Dietary habits of hospital controls maydiffer from those of the general population, but we excluded fromthe control group all diagnoses related to tobacco and alcoholconsumption, or which might have involved any long term mod-ification of diet. To reduce the possibility of recall bias due tochanges in diet subsequent to cancer onset, we asked about foodintake in the 2 years before the diagnosis. The questionnaire wasadministered to cases and controls by the same interviewers undersimilar conditions, thus minimizing information bias. Among otherstrengths of our study are the high response rate of the studyparticipants, the comparable catchment areas of cases and controls,and the possibility of controlling for major risk factors for oral,pharyngeal and esophageal cancer. The use of a validated fre-quency food questionnaire allowed us to adjust risk estimates fortotal energy intake, too.12,13

In conclusion, the present data lend support to the hypothesisthat fiber may exert a protection against oral, pharyngeal andesophageal cancer in western countries. Such favorable effectseems to be independent from smoking, drinking, and not largelyexplained throughout other dietary factors related to the risk ofupper digestive tract neoplasms.

ACKNOWLEDGEMENTS

This work was conducted with the contribution of the ItalianAssociation for Cancer Research, Milan, Italy. The authors thankMrs. M. Paola Bonifacino for editorial assistance.

REFERENCES

1. Jacobs DR, Slavin J, Marquart L. Whole grain intake and cancer: areview of the literature. Nutr Cancer 1995;24:221–29.

2. Jacobs DR Jr, Marquart L, Slavin J, Kushi LH. Whole-grain intakeand cancer: An expanded review and meta-analysis. Nutr Cancer1998;30:85–96.

3. Winn DM, Ziegler RG, Pickle LW, Gridley G, Blot WJ, Hoover RN.Diet in the etiology of oral and pharyngeal cancer among women fromthe Southern United States. Cancer Res 1984;44:1216–22.

4. McLaughlin JK, Gridley G, Block G, Winn DM, Preston-Martin S,

Schoenberg JB, et al. Dietary factors in oral and pharyngeal cancer.J Natl Cancer Inst 1988;80:1237–43.

5. Chatenoud L, Tavani A, La Vecchia C, Jacobs DR Jr, Negri E, LeviF. et al. Whole grain food intake and cancer risk. Int J Cancer1998;77:24–8.

6. Yu MC, Garabrant DH, Peters JM, Mack TM. Tobacco, alcohol, diet,occupation, and carcinoma of the esophagus. Cancer Res 1988;48:3843–8.

7. Levi F, Pasche C, Lucchini F, Chatenoud L, Jacobs DR Jr, La Vecchia

TABLE IV – ODDS RATIOS1 (OR) AND 95% CONFIDENCE INTERVALS (CI)OF ORAL, PHARYNGEAL AND ESOPHAGEAL CANCER ACCORDING TOTOTAL FIBER INTAKE, IN STRATA OF SELECTED COVARIATES, ITALY,

1992–1997

Stratum OR1 (95% CI)

Age (years),50 0.62 (0.40–0.96)50–59 0.50 (0.35–0.70)60–69 0.60 (0.42–0.86)$70 0.70 (0.34–1.48)

Education (years),7 0.60 (0.46–0.77)7–11 0.63 (0.42–0.94)$12 0.35 (0.17–0.73)

Smoking statusNever smokers 0.62 (0.38–1.02)Ex-smokers 0.53 (0.36–0.77)Current smokers

,15 cigarettes/day 0.62 (0.36–1.07)$15 cigarettes/day 0.61 (0.43–0.86)

Alcohol intake (drinks/week)0–20 0.80 (0.54–1.19)21–34 0.62 (0.38–1.04)35–55 0.57 (0.35–0.91)56–83 0.41 (0.26–0.69)$84 0.48 (0.29–0.79)

Non-alcohol energy intake (quartiles)I (lowest) 0.44 (0.27–0.72)II 0.35 (0.21–0.59)III 0.68 (0.43–1.07)IV (highest) 0.66 (0.48–0.89)

1Estimates from multiple logistic regression models including termsfor center, sex, age, education, alcohol and smoking consumption, andnon-alcohol energy intake. The unit of intake is the difference betweento 80th and 20th percentiles of the distribution of controls (i.e.,between the lower cutpoints of the 5th and 2nd quintiles).

286 SOLERET AL.

C. Refined and whole grain cereals and the risk of oral, esophagealand laryngeal cancer. Eur J Clin Nutr 2000;54:487–9.

8. Potter JD, Steinmetz K. Vegetables, fruit and phytoestrogens as pre-ventive agents. In: Stewart BW, McGregor D, Kleihues O. IARCPubl. 139, Principles of chemoprevention. Lyon: IARC, 1996:61–90.

9. Adlercreutz H, Mazur W. Phytoestrogens and western diseases. AnnMed 1997;29:95–120.

10. Slavin J, Jacobs D, Marquart L. Whole-grain consumption andchronic disease: protective mechanisms. Nutr Cancer 1997;27:14–21.

11. Franceschi S, Favero A, Conti E, Talamini R, Volpe R, Negri E, et al.Food groups, oils and butter, and cancer of the oral cavity andpharynx. Br J Cancer 1999;80:614–20.

12. Franceschi S, Barbone F, Negri E, Decarli A, Ferraroni M, Filiberti R,et al. Reproducibility of an Italian food frequency questionnaire forcancer studies. Results for specific nutrients. Ann Epidemiol 1995;5:69–75.

13. Decarli A, Franceschi S, Ferraroni M, Gnagnarella P, Parpinel MT, LaVecchia C, et al. Validation of food-frequency questionnaire to assessdietary intakes in cancer studies in Italy. Results for specific nutrients.Ann Epidemiol 1996;6:110–8.

14. Salvini S, Gnagnarella P, Parpinel MT, Boyle P, Decarli A, Ferraroni

M, et al. The food composition database for an Italian food frequencyquestionnaire. J Food Composition Anal 1996;9:57–71.

15. Englyst NH, Bingham SA, Runswick SA, Collinson E, Cummings JH.Dietary fibre (non-starch polysaccharides) in fruit, vegetables andnuts. J Hum Nutr Diet 1988;1:247–86.

16. Englyst HN, Bingham SA, Runswick SA, Collinson E, Cummings JH.Dietary fibre (non-starch polysaccharides) in cereal products. J HumNutr Diet 1989;2:253–71.

17. Breslow NE, Day NE. Statistical methods in cancer research. Vol. 1. Theanalysis of case-control studies, IARC Sci. Publ. 32. Lyon: IARC, 1980.

18. Giovannucci E. Insulin and colon cancer. Cancer Causes Control1995;6:164–79.

19. Chatenoud L, La Vecchia C, Franceschi S, Tavani A, Jacobs DR,Parpinel MT, et al. Refined cereal intake and risk of selected cancersin Italy. Am J Clin Nutr 1999;70:1107–10.

20. La Vecchia C, Chatenoud L, Franceschi S, Soler M, Parazzini F,Negri E. Vegetables and fruit and human cancer: updated of an Italianstudy. Int J Cancer 1999;82:151–2.

21. Bosetti C, La Vecchia C, Talamini R, Simonato P, Zambon P, NegriE, et al. Food groups and risk of squamous cell esophageal cancer inNorthern Italy. Int J Cancer 2000;87:289–94.

287FIBER AND UPPER DIGESTIVE TRACT CANCER