Beta-Cryptoxanthin and Lung Cancer in Shanghai, China—An Examination of Potential Confounding with Cigarette Smoking Using Urinary Cotinine as a Biomarker for True Tobacco Exposure

Abstract:

Diet may be a modifier of smoking-related cancer risk, with protective effects of intake of fruits and vegetables and associated antioxidants found in many observational studies. We previously reported serum β -cryptoxanthin levels being inversely associated with smoking-related lung cancer incidence in a cohort of Chinese men. We noted, however, that serum β -cryptoxanthin is negatively correlated with smoking. Since self-reports of smoking intensity undoubtedly contain errors, this negative correlation indicates a potential bias in assessing the effects of β -cryptoxanthin, due to confounding with the unmeasured (residual) portion of cigarette exposure. We evaluated cotinine levels in pre-diagnostic spot urine samples to attempt to improve smoking assessment. We noted that urinary cotinine levels correlated significantly with cigarette consumption overall and that cotinine was strongly predictive of lung cancer risk. Urinary cotinine, however, was not as strong a predictor of lung cancer risk in current smokers as were self-reports of cigarette consumption and cotinine remained only a marginally significant predictor of lung cancer risk after adjustment for self-reports. An apparent benefit of β -cryptoxanthin remained evident when including both urinary cotinine and self-reported cigarette consumption and cotinine in the statistical model. However, we conclude that cotinine measured from a single spot urine seems to have only limited value in augmenting self-reports of cigarette consumption so that, at present, the apparent protective effects of β -cryptoxanthin, as seen in our own study, should continue to be regarded as unproven. We believe that future epidemiological evaluation of the association between β -cryptoxanthin (and other antioxidants) and reduced lung cancer risk must utilize improved biomarkers to augment smokers' own self-reports of smoking amount.

∗Deceased

Acknowledgments and Notes

The study was supported by grants R01 CA43092, R01 CA 54281, 5P30 ES07048, P01 CA 17054, and U01 GM074496 from the United States National Institutes of Health.

Notes

∗Deceased

∗2-sided Ps based on t-test or chi-square test.

^†Arithmetic means and 95% confidence intervals in parentheses.

^‡Percentages in total cases or control subjects.

^§Geometric means and 95% confidence intervals in parentheses.

^a Among ever smokers only.

∗2-sided Ps derived from analysis of variance.

∗2-sided P < 0.01.

∗Self-reported smoking variables were adjusted for urinary level of cotinine in quartiles. Urinary cotinine was adjusted for smoking status (never, ex-, and current smokers), number of cigarettes per day, and age at starting to smoke.

∗Cut-off values defining the quartiles were 0–1.80, 1.81–3.00, 3.01–4.53, and greater than 4.53 mcg/dl for the 1st through 4th quartiles, respectively.

^†Adjusted for smoking status (never, ex-, and current smokers), number of cigarettes smoked per day, and age at starting to smoke; OR, odds ratio; CI, confidence interval.

^‡Adjusted for smoking status (never, ex-, and current smokers), number of cigarettes smoked per day, age at starting to smoke, and urinary levels of cotinine in quartiles.