|Home | About | Journals | Submit | Contact Us | Français|
The purpose of this study was to examine associations of pretreatment serum carotenoids, tocopherols, and quercetin with prognosis in 154 newly diagnosed head and neck cancer (HNC) patients.
Pretreatment blood and health surveys were collected. Serum micronutrients were measured by HPLC. Data on recurrence and death were collected annually. Cox proportional hazards models measured associations of serum nutrient concentrations with recurrence and overall survival.
During a median follow-up time of 37 months, there were 32 recurrences and 27 deaths. After controlling for covariates, subjects with high versus low serum xanthophyll and total carotenoid concentrations had significantly longer recurrence-free time (P = 0.002 and P = 0.02, respectively). Overall survival time was significantly longer in subjects with high versus low serum xanthophyll concentrations (P = 0.02).
Future research should evaluate the possible benefits of interventions to increase intakes of rich food sources of xanthophylls in this patient population.
Higher fruit and vegetable intake is associated with decreased risk of head and neck cancer (HNC) development and mortality.1–6 We recently reported that a whole foods pattern, characterized by high intakes of vegetables, fruits, whole grains, poultry, and fish was associated with lower recurrence and mortality rates in a prospective cohort of newly diagnosed HNC cases.7 This same dietary pattern was associated with lower levels of pro-inflammatory cytokines that have been hypothesized to facilitate tumor growth and dissemination in patients with HNC.8
While our previous work provides evidence that consuming a healthy diet prior to HNC treatment is beneficial to clinical and biological HNC outcomes, a limitation of these studies was the use of self-reported diet via food frequency questionnaire (FFQ), which is susceptible to measurement error. Examining biomarkers of dietary intake in relation to HNC disease outcomes is important, as consistency across results will reduce the likelihood that bias or confounding was responsible for previously observed associations with reported diet.9
Blood carotenoids have been established as valid biomarkers of fruit and vegetable intake.10 Tocopherols are found mainly in vegetable oils and high fat foods like nuts. The roles of carotenoids and tocopherols in cancer chemoprevention have been of great interest over the past two decades due to their antioxidant properties.11 Quercetin is a flavonol that accounts for about 75% of typical U.S. intake of flavonoids.12 It is widely distributed in many fruits and vegetables; tea, onions, allium vegetables and apples with skin are major food sources of quercetin in the U.S., and these foods are low in the highly pigmented carotenoids.12–15 Dietary intakes of β-carotene and quercetin have been reported to be moderately correlated (r = 0.27 and r = 0.32 for women and men, respectively),12 suggesting measurement of quercetin may capture fruit and vegetable intakes that are not well represented by carotenoids.16 Like carotenoids and tocopherols, quercetin has a potential role in cancer prevention17, 18 due to its anti-inflammatory effects and promotion of apoptosis and autophagy in cancer cells.19–23 Few studies have examined associations of serum carotenoids and tocopherols with HNC outcomes3, 6, 24 and quercetin concentrations have never been reported in relation to HNC mortality.
This study evaluated the associations of pretreatment serum carotenoids, tocopherols, and quercetin with recurrence-free and overall survival time in a cohort of newly diagnosed HNC patients. The hypothesis was that a protective relationship would be observed between the examined serum micronutrients and outcomes.
This prospective cohort study utilized data collected from 154 newly diagnosed HNC patients enrolled in the University of Michigan Head and Neck Specialized Program of Research Excellence (UM HN-SPORE). All study activities were approved by the internal Institutional Review Board at University of Michigan and study participants gave signed, informed consent to participate. Recruitment and data collection methods have been described elsewhere.8 Briefly, subjects were recruited between November 2008 and August 2012. Exclusion criteria included being < 18 years of age, pregnant, non-English speaking, diagnosed as mentally unstable (e.g. presence of schizophrenia, severe, unmanaged depression or any other state or condition that study personnel feel would make active study participation difficult or impossible), diagnosed with another non-upper aerodigestive tract cancer, or diagnosed with any other head and neck primary within the past five years.
Enrolled participants completed a self-administered epidemiologic health questionnaire before treatment that collected data on demographics, tobacco use, alcohol consumption, weight status, and comorbidities, as well as an FFQ.9 Clinical data were recorded from medical records. Pretreatment peripheral blood samples (30 mL) were collected using routine venipuncture. Sera were collected after centrifugation of blood and stored in the HN-SPORE Tissue Core in 0.5 mL aliquots at −80 °C until testing. All serum samples were barcoded and analyzed blinded to subject identifiers or outcomes.
During the study period a total of N=520 evaluable subjects were recruited, yielding a 92% participation rate. For the current analysis, we analyzed serum nutrient values for those subjects for whom pretreatment serum samples, epidemiologic health questionnaires and FFQs were available. N=219 subjects (42.1%) completed baseline health questionnaires and FFQs with valid data. FFQ data was considered valid according to standard criteria, i.e. no complete pages missing, no more than 70 missing items and total energy intake ≤5000 kcals/day and ≥200 kcals per day.9 N=65 subjects did not have baseline blood collected or adequate samples available for measurement, leaving a final sample size of N=154.
Quercetin, morin, citric acid, ammonium acetate, hydrochloric acid, and enzymes were obtained from Sigma Aldrich (St. Louis, MO). HPLC grade hexane, acetonitrile, water, methyl-tert-butyl-ether and methanol were Burdick and Jackson brand, HPLC-grade (Honeywell, Morristown, NJ). Tocol was a gift of Hoffman La Roche (Basel, Switzerland). High performance liquid chromatography (HPLC) was performed using a Shimadzu system consisting of two LC-20AT pumps, a CTO-20A autosampler, a CBM-20A controller and two detectors in tandem: SPD-20AV UV/visible detector and a 5600A ESA CoulArray Multi-Electrode Detector (ESA Biosciences, Chelmsford, MA). Chromatograms were analyzed using ESA CoulArray Software (220.127.116.11, 2002).
Analysis of serum carotenoids and tocopherols has been described elsewhere.8 For quercetin, serum aliquots, 200 μL, were buffered with 21 μL 0.55 M sodium acetate buffer (pH 5.0) followed by addition of 800 units β-glucuronidase from Helix pomatia, that contains sulfatase activity, and incubated at 37°C for 17 hours to release conjuagtes, based on the methods of Erlund et al.25 The internal standard morin, 10 μL of 10 μg/mL in methanol, was then added along with 0.1M hydrochloric acid (25 μL) followed by vortexing. Extraction with ethyl acetate (500 μL) was performed twice. The organic extracts were dried in the SpeedVac and reconstituted in 100 μL of HPLC mobile phases A and B mixed in a 1:1 ratio before HPLC analysis. Detection was with the ESA detector set at 50mV, 100mV, and 150mV.
The separations were performed with a reversed phased YMC-Pack Pro C18 column (150 X 4.6 mm, catalog # AS12SO3-1546WT, YMC, Allentown, PA). The mobile phases used were A: 10% acetonitrile, 80% water, 10% buffer (250mM ammonium acetate, 750 mM citric acid, pH 2.7), and B: 50% acetonitrile, 40% water, 10% buffer. The gradient began at 40% B and then increased to 100% B over 9 minutes. There was a 3 minute final hold, and then a 5 minute re-equilibration period. The flow rate was 1.0mL/min. The serum samples were analyzed in six batches, and aliquots of the same standards were used for calibration with each batch. The inter-day coefficients of variation for the individual carotenoids were between 3% and 11%. Our laboratory participated in the National Institutes of Standards Technology Round Robin for carotenoids, and the values obtained by our laboratory were within two standard deviations for all the samples analyzed.
Sociodemographic variables included age and sex. Tobacco and alcohol use data were categorized as current, former, or never, where “current” reflected use in the 12 months prior to cancer diagnosis. Body mass index (BMI) (kg/m2) was calculated at diagnosis based on self-reported weight and height. Comorbidities were recorded using the Adult Comorbidity Evaluation-27 and categorized into none, mild, moderate, or severe comorbidities.26 As previously described, tumor human papillomavirus (HPV) status was determined by an ultrasensitive method using real-time competitive polymerase chain reaction and matrix-assisted laser desorption/ionization time of flight mass spectroscopy with separation of products on a matrix loaded silicon chip array.27 As a result of having equivocal HPV-status data for n=34 (22.1%) participants, tumor site was recorded from medical records and categorized into four groups, which allowed for consideration of known HPV-status: 1) oral cavity, 2) oropharynx HPV+, 3) oropharynx HPV- or unknown and 4) larynx. To increase statistical power, cancer stage was categorized a priori into two groups, with Stages 1 and 2 collapsed, and Stages 3 and 4 collapsed to increase statistical power. The method used to determine HPV-status has been previously described.8, 27
Patients were followed according to National Comprehensive Cancer Network guideline intervals. New tumor events and status (recurrence, residual disease, persistent disease, and second primary) were updated at each visit and annually via medical record review. For subjects who did not return to our institution for surveillance after their treatment, information on tumor events and status was obtained through self-report and contact with local physicians. Deaths were confirmed through the Social Security Death Index, yearly survey updates, notification from family, and medical record reviews. Survival time was censored to February 1, 2014. Recurrence events were censored to the last date of each subject’s annual medical record review. Persistent disease was assigned a follow-up time of one day for calculation of time to recurrence. Subjects were considered to be lost to follow-up if they had not returned for care, and there was a failure to contact the subject or to collect updates when contacting the referring physician. Subjects lost to follow-up were censored until their last known status.
Descriptive statistics (means and frequencies) were generated for demographic, epidemiologic and clinical variables. To assess relationships between serum nutrient concentrations (categorized into tertiles) and demographic, clinical, and behavioral variables, chi-squared tests were used for categorical variables and Wilcoxon rank-sum tests were used for continuous variables. To examine whether degradation of serum nutrient concentrations may have occurred over time, Spearman correlations were computed between sample storage time (number of days between date of assay and date blood samples were collected) and serum nutrient concentrations.
Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for the associations of recurrence-free and overall survival time with serum nutrient concentrations. To arrive at a parsimonious final multivariable model, variables were considered for inclusion using a backwards selection strategy based on a priori knowledge and input from treating physicians of this study population. Final multivariable models were adjusted for age, BMI, tumor site, cancer stage, and tobacco use. Sex, alcohol use, comorbidities and sample storage time were considered as covariates but excluded from final models as they did not substantially alter parameter estimates. Treatment modality was not considered as a covariate in this analysis due to the high collinearity of treatment and disease site (P<0001 for chi-square test with p-value approximated by Monte-Carlo simulation). Other behavioral variables (e.g. physical activity, red meat intake) were not considered as covariates based on previous findings which showed a lack of association of these variables with mortality in this study population.1, 7 All serum nutrient concentrations were categorized into tertiles (low, medium, and high concentrations) for use in models and display in Kaplan-Meier plots. HRs and 95% CIs were estimated for each tertile of serum nutrient concentration and compared with the lowest, tertile 1. The SAS® system Version 9.4 (SAS Institute Inc., Cary, NC) was used for all analyses. P-values of 0.05 or less were considered statistically significant and no multiplicity adjustments were performed.
During a median longitudinal follow-up time of 37 months, there were 32 (20.8%) recurrence events and 27 deaths (17.5%). Epidemiological characteristics of the study population are given in Table 1. The mean age of study participants was 59 years. The majority of participants were men (79%). The mean BMI at diagnosis was 27.2 kg/m2, with 67% of the population considered overweight or obese. Only 19% of the participants reported never smoking, while 37% reported smoking in the last 12 months. Tumors were most commonly of the oral cavity (37%) or HPV+ oropharynx (30%) and diagnosed at an advanced stage (66% stage IV).
Select patient characteristics across tertiles of serum total carotenoids, xanthophylls, quercetin, and α- and γ-tocopherol are shown in Table 2. Age was positively associated with tocopherols. Mean BMI increased significantly across increasing tertiles of quercetin. A significantly higher proportion of patients with stage III or IV cancers had low serum xanthophyll concentrations. Participants with the highest serum xanthophyll and total carotenoid concentrations were significantly less likely to be current smokers. Patients self-reporting higher intake of green leafy vegetables, dark yellow and orange vegetables and fruit on the FFQ had significantly higher serum xanthophyll concentrations. Serum total carotenoid concentrations were significantly positively associated with intake of dark yellow and orange vegetables and fruit on the FFQ. We previously reported serum total carotenoids, α-carotene, β-carotene, β-cryptoxanthin, lutein, zeaxanthin, and α-tocopherol to be significantly correlated with reported dietary intake of each micronutrient in this study population.8 Significant, negative spearman correlations were found between sample days and β-carotene (ρ=−0.29, p = 0.001), α-carotene (ρ=−0.28, p=0.001) and total carotenoids (ρ=−0.22, p=0.01).
Kaplan-Meier curves and associated p-values for the associations between each serum nutrient (categorized into tertiles) with time-to-recurrence and time-to-death (any cause)are shown in Figure 1. HRs and 95% CIs for recurrence and mortality by serum micronutrient levels are shown in Table 3. After controlling for covariates, participants with high serum xanthophylls and total carotenoids were significantly less likely to experience a recurrence than those with low concentrations. Participants with high serum xanthphylls also had significantly lower mortality than those with low concentrations. Although not statistically significant, there appeared to be trends toward lower recurrence with higher serum β-carotene, lycopene, and α-tocopherol levels, and lower mortality with high serum γ-tocopherol levels. There were no apparent associations between recurrence and α-carotene, quercetin, and γ-tocopherol, or between mortality and α-carotene, β-carotene, lycopene, total carotenoids, quercetin, and α-tocopherol.
This study of pretreatment serum micronutrient concentrations in newly diagnosed HNC patients showed that high pretreatment serum xanthophyll concentrations were associated with a lower risk of recurrence and mortality, independent of other factors known to influence these outcomes. Results also showed that high pretreatment total carotenoid concentrations were associated with decreased risk of recurrence. The results of this study are of particular importance because they are consistent with the previous findings of our group that a whole foods dietary pattern, derived from Principal Components Analysis (PCA), was associated with lower risk of HNC mortality and recurrence.7 Foods that were among the most significant contributors to this dietary pattern (i.e. had the highest factor loadings) included green leafy vegetables, dark-yellow and orange vegetables, and fruit—all rich food sources of xanthophylls and total carotenoids. Therefore it is less likely that bias or confounding were responsible for these previously observed associations based on reported diet.
Similar to our results, Sakhi et al. found that higher pretreatment serum carotenoids measured in 60 patients, predicted better progression-free survival.6 However, other studies suggest the relationship between HNC mortality and carotenoids and tocopherols may be modified by smoking status. As has been previously reported, it is possible that carotenoids and tocopherols have a detrimental effect, especially in smokers, when consumed in high doses from supplements that are unlikely to be attained through dietary intake alone.28, 29 An RCT carried out in a sample of 540 newly diagnosed HNC patients showed that smokers randomized to the α-tocopherol and β-carotene supplementation group had increased recurrence, mortality, and second primary cancers compared to non-smokers and placebo.30 Interestingly, however, higher reported β-carotene dietary intake (i.e. not from supplements) before randomization and subsequent radiotherapy was associated with reduced risk of local recurrence, regardless of smoking status.24 In another study Mayne et al. reported protective effects of total serum carotenoids on mortality, but increased mortality with higher α-tocopherol in post-treatment HNC patients who were smokers and disease-free at the time of blood sampling.3 We did not have the statistical power to assess effect modification by smoking status in the current study but this issue warrants further investigation in the future.
We found that xanthophylls were strongly associated with better HNC outcomes. Consistent with our findings, Hughes et al. found that xanthophylls, either individually or grouped, were inversely associated with a urinary marker of oxidative stress 31. The xanthophylls include the oxygenated carotenoids lutein, β–cryptoxanthin and zeaxanthin. It is unclear if their protective effects in cells are the result of chemical or biochemical properties of these carotenoids or of some other aspect of foods that are high in xanthophylls. While α- and β-carotenes are distributed in a wider variety of fruits and vegetables than the other carotenoids measured, lutein and zeaxanthin are high mainly in green leafy vegetables that are typically low in the U.S. diet.32 Men tend to eat less fruit than women,33 and β–cryptoxanthin is especially high in dark orange fruits. More heterogeneity in dietary intakes and blood concentrations may be present for xanthophylls than the other carotenoids due to their distribution in a smaller variety of fruits and vegetables. Few supplements containing xanthophylls are currently available and so it is unknown whether supplemental xanthophylls would behave the same way as β-carotene and tocopherols and interact with smoking and/or increase cancer risk and mortality.
Unlike carotenoids, higher quercetin concentrations were not protective. Concentrations of quercetin, kaempferol, naringenin, and hesperitin in fasting plasma have been shown to be reasonable biomarkers of recent dietary intakes with a short half-life.34 Carotenoids are highly non-polar with much longer half-lives 35–37. Quercetin is more polar and has both poor and highly variable bioavailability 38. Therefore, the findings that quercetin was not associated with survival may not negate the potential protective effects of foods that contain quercetin.
The results of this study should be interpreted in light of several strengths and limitations. Strengths include the prospective study design, ability to adjust for some of the most significant potential confounding factors and the ability to account for HPV-status of the tumor. Numerous studies have reported a survival advantage in HNC patients with HPV+ tumors compared to those with HPV− tumors.39 Our group previously reported that HPV-positivity was associated with significantly higher intakes of chemoprotective micronutrients.40 In the current study, we were able to account for HPV-status when adjusting for tumor site, decreasing the potential for confounding by this characteristic. Due to the small sample size and incomplete data on post-treatment smoking status on 30.6% of the study population we did not have the statistical power to consider tobacco use as a time-dependent variable in statistical models, which is a limitation. Another limitation is that we did not have power to evaluate patients by type of cancer treatment received. In stage IV patients, which were the majority of the study group, the efficacy of chemotherapy and radiation therapy is variable. We also found significant, inverse correlations between sample storage time and α-carotene, β-carotene and total carotenoids, suggesting there may have been some degradation of these nutrients in the older serum samples that were used in this analysis. However, we hypothesize that any degradation of nutrients that may have occurred in our samples over time likely resulted in an attenuation of their associations with recurrence-free and overall survival time. Finally, the prospective study design only allowed for the determination of associations, and we cannot rule out the possibility of reverse causality and potential unmeasured and residual confounding.
In summary, we found strong protective associations of serum xanthophyll and total carotenoid concentrations on overall survival time and protective associations of serum xanthophyll concentrations with recurrence-free time in HNC patients This indicates that in this population, dietary intakes of other micronutrients with putative preventive properties was either already sufficiently high for maximal benefit and/or that intakes of foods high in xanthophylls and carotenoids was deficient. Future research could evaluate the possible benefits of an intervention to increase intakes of foods that are rich sources of xanthophylls in HNC patients or in persons at increased risk for HNC.
This project was supported by Federal Funds from the National Cancer Institute, National Institutes of Health under the University of Michigan Specialized Programs of Research Excellence (UM SPORE): P50CA097248; the UM Cancer Center Support Grant P30CA046592; Anna Arthur was supported by NIH/NCI Cancer Prevention and Control Training Grant: R25 CA047888, a Colgate-Palmolive Fellowship in Nutrition and Oral Health, and a UM Rackham Predoctoral Fellowship; Christie Mueller was supported by the UM Cancer Biology Training Grant 5T32CA009676; and Ethan Harris was supported by the UM Rackham Summer Research Opportunity Program. We thank the patients, clinicians, and principal investigators of the individual projects at the UM Head and Neck SPORE program, who were responsible for the recruitment, treatment, and follow-up of patients and provided access to the longitudinal clinical database included in this report. These investigators included Carol Bradford, Avraham Eisbruch, Theodore Lawrence, Mark Prince, Jeffrey Terrell, Shaomeng Wang, Frank Worden, Joseph Helman, Brent Ward, and Andrea Haddad.
Meeting at which this manuscript was presented: American Society of Preventive Oncology, March 2015, Birmingham, AL
The authors declare no conflicts of interest with the work.