Nutrient dietary patterns and the risk of laryngeal cancer : an Italian case-control study

Introduction. Several studies have indicated a role of diet in the etiology of laryngeal cancer. Since foods and nutrients tend to be closely related and act synergistically, the pecific effect of each dietary component of interest may be difficult to identify and can be partly confounded by other dietary components. Dietary patterns have thus been proposed as a practical tool to describe the association between diet and cancer, given their ability to capture the variations in overall food intake (Newby, Tucker 2004). A few studies have investigated the role of diet on laryngeal cancer through factor analysis. Aims We applied exploratory principal component factor analysis (PCFA) to identify a posteriori dietary patterns for a multicentric case-control study conducted in Italy on cancer of the larynx. This a posteriori technique allows to integrate several dietary exposures (i.e. foods, food groups, nutrients) into a smaller number of dietary patterns, that are independent from one another and can be evaluated as risk factors in subsequent analysis for the assessment of cancer risk. Methods A case-control study of cancer of the larynx was conducted from 1992 and 2000 in the provinces of Milan and Pordenone, in the Northern Italy. Cases were 460 subjects (415 men, 45 women) admitted to major teaching and general hospitals in the study areas with incident, histologically confirmed squamous cell cancer of the larynx, diagnosed no longer than 1 year before the interview. Controls were 1088 subjects (863 men, 225 women) admitted to the same hospitals for a wide spectrum of acute, non-neoplastic conditions, unrelated to smoking or alcohol drinking, or long term modifications of diet. The subjects’ diet was assessed using a valid and reproducible food frequency questionnaire (FFQ) including 78 foods and beverages, as well as a range of the most common Italian recipes. Subjects were asked to indicate the average weekly consumption for each dietary item; intakes lower than once a month were coded as 0.5 per week. To estimate the intake of various nutrients, an Italian food composition database was used. We performed an exploratory PCFA on a selected set of 28 major macro- and micro-nutrients. We preliminarily evaluated the correlation matrix to determine if it was factourable, trough visual inspection and statistical procedure (Bartlett’s test of sphericity). Moreover, we evaluated the Kaiser-Meyer-Olkin measure and individual measures of sampling adequacy (Kaiser, 1974). We chose the number of factors to retain based on the following criteria: factor eigenvalue greater than 1, scree plot examination and factor interpretability. We applied a varimax rotation to the factor loadings matrix to achieve a simpler and more interpretable solution. We used nutrients with rotated factor loading greater or equal to 0.63 on a given factor to name the pattern. Factor scores were defined for each subject and for each factor following the weighted least square method. They indicate the degree to which each subject’s diet conforms to one of the identified patterns. To examine the robustness of the identified dietary patterns, we performed a principal axis factor analysis on the standardized nutrients and a maximum likelihood factor analysis after logarithmic transformation of the original nutrients. We calculated factor scores referring to the multiple regression method and standardizing the results. The correlations between scores referring to the same factor calculated with different methods were equal to 1 for all the comparisons. We also performed factor analysis separately within male and female subsamples and within different centers. All these checks yielded dietary patterns consistent with PCFA the ones obtained on the overall sample. To assess the reliability and refine the identified factors, we evaluated the internal consistency of those nutrients with a loading greater than 0.40 using standardized Cronbach’s coefficient alpha. We calculated coefficient alphas for each factor and coefficient alphas when item deleted (Cronbach, 1951). To confirm the internal reproducibility of the identified patterns, individuals were randomly placed into one of two equally sized groups, and PCFA was performed separately in both subsamples. For each factor, we grouped participants into three categories according to quintiles of factor scores among the control population, and estimated the odds ratio and corresponding 95% confidence intervals using unconditional multiple logistic regression models, including all the factors simultaneously. The model was adjusted for sex, age, study center, education, body mass index, physical activity, tobacco smoking, and alcohol drinking. Results Five factors were retained according to the defined criteria. These factors explained 79% of the total variance of the original nutrients. The first pattern, named Animal products, had the greatest loadings on calcium, phosphorus, riboflavin, animal protein, saturated fatty acids, zinc, and cholesterol. The second pattern, named Starch-rich, had the greatest loadings on starch, vegetable protein, and sodium. The third pattern, named Vitamins and fiber, had the greatest loadings on vitamin C, total fiber, beta-carotene equivalents, and total folate. The fourth pattern, named Seed oils, had the greatest loadings on linoleic acid, vitamin E, and linolenic acid. The fifth pattern, named Fish-rich, had the greatest loadings on other polyunsaturated fatty acids, and vitamin D. A direct association was observed between the Animal products pattern and laryngeal cancer (OR=1.94, 95% CI: 1.39-2.70). A borderline direct association was observed between the Starch-rich pattern and laryngeal cancer (OR=1.30, 95% CI: 0.93-1.81). An inverse relationship was observed between the Vitamins and fiber pattern and laryngeal cancer (OR=0.56, 95% CI: 0.41-0.76). No relationship was evident between the Seed oils pattern and laryngeal cancer (OR=0.98, 95% CI: 0.70-1.37). A direct association was found between the Fish-rich pattern and the laryngeal cancer (OR=2.09, 95% CI: 1.51-2.90). Conclusions The role of dietary habits on the risk of laryngeal cancer was evaluated through exploratory PCFA on 28 major nutrients of interest. We identified 5 major dietary patterns, explaining about 80% of the total variance of the original nutrients. Our results indicated that the Animal products and Fish-rich patterns are potentially unfavourable indicators of risk for laryngeal cancer, while the Vitamins and fiber pattern is inversely related to laryngeal cancer.