Diet and proinflammatory cytokine levels in head and neck squamous cell carcinoma

Abstract

Background

Proinflammatory cytokine levels may be associated with cancer stage, recurrence, and survival. A study was undertaken to determine if cytokine levels were associated with dietary patterns and fat-soluble micronutrients in previously untreated head and neck squamous cell carcinoma (HNSCC) patients.

Methods

This was a cross-sectional study of 160 newly diagnosed HNSCC patients who completed pretreatment food frequency questionnaires (FFQ) and health surveys. Dietary patterns were derived from FFQs using principal component analysis. Pretreatment serum levels of the proinflammatory cytokines IL-6, TNF-α, and IFN-γ were measured by ELISA and serum carotenoid and tocopherol levels by HPLC. Multivariable ordinal logistic regression models examined associations between cytokines and quartiles of reported and serum dietary variables.

Results

Three dietary patterns emerged: whole foods, Western, and convenience foods. In multivariable analyses, higher whole foods pattern scores were significantly associated with lower levels of IL-6, TNF-α, and IFN-γ (P = <0.001, P = 0.008, and P = 0.03, respectively). Significant inverse associations were reported between IL-6, TNF-α, and IFN-γ levels and quartiles of total reported carotenoid intake (P = 0.006, P = 0.04, and P = 0.04, respectively). There was an inverse association between IFN-γ levels and serum α-tocopherol levels (P = 0.03).

Conclusions

Consuming a pretreatment diet rich in vegetables, fruit, fish, poultry and whole grains may be associated with lower proinflammatory cytokine levels in patients with HNSCC.

Introduction

Higher fruit and vegetable intake and levels of serum carotenoids are associated with more favorable head and neck squamous cell carcinoma (HNSCC) prognoses.^1–4 We reported that a whole foods pattern, characterized by high intakes of vegetables, fruits, whole grains, poultry, and fish, and a BMI ≥ 25 kg/m² at the time of diagnosis, were associated with lower recurrence and mortality rates in a large prospective cohort of newly diagnosed HNSCC cases.⁵ These findings suggest diet as a potential area of intervention to improve HNSCC prognosis. However, research exploring potential mechanisms underlying these associations is warranted to deepen understanding of how nutritional interventions might influence patient survival.

It is widely accepted that chronic inflammation can be a driver of cancer development and progression.⁶ The proinflammatory cytokines interleukin (IL)-6 and tumor necrosis factor (TNF)-α are mediators of the immune response and are thought to be involved in malignant transformation, progression, and prognosis.⁶ Endothelial cells in the tumor microenvironment, and HNSCC cells themselves, have been shown to secrete high levels of IL-6, inducing tumor invasion and metastasis.^{7, 8} In observational studies, higher IL-6 has been associated with later cancer stage^9–11 and IL-6^12–15 and TNF-α¹⁶ have been associated with increased HNSCC recurrence and mortality, including within our own HNSCC study population.^{17, 18}

Dietary intake is known to be involved in the physiological response to inflammation and oxidative stress.¹⁹ Therefore, we hypothesize that nutrients with antioxidant and anti-inflammatory properties such as carotenoids and vitamin E may mitigate the effects of proinflammatory cytokines in the body, reducing the likelihood of metastasis and prolonging survival. To our knowledge, the relationship between diet and cytokines has not been studied in HNSCC. The objective of this study was to examine associations of dietary patterns, carotenoids and tocopherols with pretreatment serum levels of proinflammatory cytokines in a cohort of newly diagnosed HNSCC cases.

Methods

Design, setting and subjects

This was a cross-sectional study of patients newly diagnosed with HNSCC and enrolled in the University of Michigan Head and Neck Specialized Program of Research Excellence (HN-SPORE). Institutional Review Board approval was granted from the University of Michigan Health System (Ann Arbor, MI). Patients were recruited between November 2008 and November 2012. Exclusion criteria included: 1) < 18 years of age; 2) pregnant; 3) non-English speaking; 4) diagnosed as mentally unstable; 5) a diagnosis of another non-upper aerodigestive tract cancer; or 6) any head and neck primary diagnosed within the past five years.

Participants completed a self-administered health questionnaire prior to treatment that collected data on demographics, tobacco use, alcohol use, weight status, and comorbidities. Pretreatment dietary intake was assessed using the 2007 self-administered, semi-quantitative Harvard FFQ.²⁰ Medical records were reviewed to collect data on tumor site and stage.

Peripheral blood samples (30 mL) were collected prior to treatment using routine venipuncture technique. Sera were collected after centrifugation of blood and stored in 0.5 mL aliquots at −80 °C until testing. To ensure patient confidentiality and blind laboratory personnel, all serum samples were immediately barcoded and assigned a numerical identifier before being stored in the HN-SPORE Tissue Core.

Measures

Reported diet

The validated 2007 Harvard FFQ was designed to assess respondents’ usual dietary intake from food and supplements over the past year.^21–23 The FFQ includes standard portion sizes for each item and allows participants to choose their average frequency of consumption over the past year. Total energy and nutrient intakes were estimated by summing intakes from each food based on the portion size, frequency of consumption, and nutrient content of each food. Daily food servings were estimated by summing the frequency weights of each food item based on reported daily frequencies of consumption.²³

Serum micronutrients

Serum carotenoids and tocopherols were extracted and analyzed by high pressure liquid chromatography as previously described.²⁴ Briefly, serum was mixed with an equal volume of ethanol containing butylated hydroxytoluene and extracted with hexane. Tocol as was the internal standard. A YMC C30 reverse phase column (2×150 mm with a 2.0×20mm guard) was used to separate carotenoids and tocopherols using gradient elution at 0.2 ml/min total flow on a Shimadzu LC-20AT HPLC system. Detection was at 450–472 nm for carotenoids. Electrochemical detection was used for Tocol and tocopherols with a Coularray electrochemical detector set at 310, 390 and 470 mV. Samples were analyzed in six batches of approximately 30 samples per batch.

Covariates

Sociodemographic variables were age, sex and level of education. Highest educational level attained was dichotomized as “≤ high school diploma or equivalent” and “some college or more”. Smoking data were categorized as current, former, or never smoker, where status “current” reflects use in the 12 months prior to cancer diagnosis. Alcohol abuse was measured using the previously validated Alcohol Use Disorders Identification Test (AUDIT).²⁵ An AUDIT score ≥8 was considered problem drinking. Tumor site was recorded from medical records and categorized into three groups: 1) oral cavity, 2) oropharynx and 3) larynx. In order to increase statistical power of stage-wise comparisons, cancer stage was categorized a priori into three groups, with Stages 1 and 2 collapsed, and Stage 3 and Stage 4 considered separately. Tumor 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, as previously described.²⁶ Comorbidities were recorded using the Adult Comorbidity Evaluation-27 and categorized into none or mild comorbidities compared with moderate to severe comorbidities.²⁷

Serum cytokines

Cytokine levels were determined using standard methodology established in our Cancer Center Immune Monitoring Core using commercially available paired-antibody ELISA kits. All blood samples were obtained in the outpatient clinic, stored at 4 degrees until transfer to the Tissue core (within 2 hours) where they were immediately centrifuged and sera separated into 2ul aliquots and frozen immediately for storage at −80 °C. Internal controls were used for all ELISA plates, and all patient samples were assayed in the same batch because the microplate arrays for analysis accommodated 174 duplicate samples at a time. Briefly, serum sample aliquots frozen at −80 °C were thawed then incubated overnight @ 4°C in duplicate (25ul/well) on microtiter plates pre-coated with monoclonal antibody specific for IL-6, TNF-α, or IFN-γ, accordingly. Any unbound substances were washed away and biotin-linked polyclonal antibody specific for the cytokine was introduced. After incubation for two hours at room temperature, the plates were washed and incubated with streptavidin-HRP for an additional hour. After a final wash, substrate solution was then added and color development stopped after 25 minutes at room temperature. A microplate reader was then used to determine colorimetric densities for each sample as calculated from a standard curve. The test sensitivity was determined per manufacturer guidelines. For quality control, multiple plasma aliquots from a single donor were prepared and one aliquot analyzed with the batch of samples. Frozen aliquots were stored for repetitive testing if necessary.

Statistical analysis

Descriptive statistics (means and frequencies) were generated for all demographic, epidemiologic and clinical variables. Dietary intake data was assessed for missing values and energy outliers using the Rosner method.^{23, 28} Food consumption data from FFQs were classified a priori into 40 foods and food groups using methods similar to those described in previous studies of dietary patterns and disease^{29, 30} and pretreatment dietary patterns were derived by using principal component analysis (PCA), as previously reported.⁵ PCA is a data reduction method that derives dietary patterns by aggregating food variables together whose intakes are correlated.³¹ Pattern factor scores were calculated for each study and categorized into quartiles for analysis with serum cytokine levels.

Odds ratios (ORs) and associated 95% confidence intervals (CIs) were calculated from multivariable logistic regression models to assess the relationships between dietary variables and cytokine levels. Cytokine values were categorized into two or three levels according to their observed empirical count distribution. IFN-γ and IL-6 were categorized into three levels: zero level, non-zero low level and non-zero high level, where the medians of the non-zero values were used to separate the latter two levels. TNF-α was dichotomized into zero and non-zero levels, due to the high proportion of zero-values (64%). For the outcomes with three levels an ordinal logistic model with the proportional odds assumption was used.³² When possible, the proportional odds assumption was imposed (i.e., we assumed the odds ratios between any two neighboring categories were the same). This allowed us to achieve more efficient estimation especially given the small sample size. Dietary pattern scores and carotenoid and tocopherol levels were categorized into quartiles for use in statistical models. Cochran-Armitage tests for linear trend across quartiles were conducted between cytokines and nutrient quartiles.³³ All final models were adjusted for age, sex, tumor site, cancer stage, smoking, and alcohol problem. Education, BMI and ACE-comorbidities were considered as covariates but excluded from final models. Stratified analyses were conducted in which associations were examined separately for subjects with stage 1, 2, or 3 cancers and subjects with stage 4 cancers to assess potential effect modification by cancer stage. To examine HPV-status as a potential confounder, analyses were repeated in the subset of 122 patients on which HPV-status was available. The SAS^® system Version 9.3 (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.

Results

Of 737 eligible patients approached, 670 consented, yielding a response rate of 91%. Additional patients were excluded who did not complete a baseline food frequency questionnaire (FFQ) (n = 304), withdrew from the study (n = 24), had reported daily energy intake <200 or >5000 kilocalories (kcals) per day (n = 6), or did not have baseline serum available for use in this study (n = 175). The final sample size included 160 newly diagnosed HNSCC patients.

Characteristics of the study population are displayed in Table 1. The mean age of study participants was approximately 60 years. The majority of participants were white (94.4%) and male (80.1%). About 60% of the participants had at least some college education. The most frequent tumor location was the oropharynx (47.8%) and most participants presented to UMHS clinics with late-stage (III or IV) cancers (84.9%). About 74% of participants were ever smokers, with about <1/5^th reporting an alcohol problem (16.2% with AUDIT score ≥ 8). The mean BMI at the time of diagnosis was 27.2 kg/m², which is considered overweight, but lower than the average BMI of 28.7 kg/m² in the United States population from 1999 – 2010 ³⁴. Of the 122 subjects for which HPV-status of the tumor was available, 61.5% were positive and 38.5% were negative. Correlations between serum and reported antioxidants are displayed in Supplementary Table 1. Mean cytokine levels by patient characteristics are reported in Supplementary Table 2.

Table 1.

Pretreatment characteristics of newly diagnosed head and neck cancer patients

Characteristic	No. of Patients (n = 160)	%
Age, years
Mean	59.7
SD	10.7
Range	25 – 93
Sex
Male	128	80.0
Female	32	20.0
Race
White/Non-Hispanic	151	94.4
Non-white/Hispanic	9	5.6
Education
High School or Less	63	39.4
Some College or More	97	60.6
Site
Oral Cavity	48	30.0
Oropharynx	77	48.1
Larynx	35	21.9
Stage
1,2	25	15.6
3	24	15.0
4	111	69.4
HPV-status†
Positive	47	38.5
Negative	75	61.5
ACE-27 comorbidity score
None or mild	114	71.2
Moderate or severe	46	28.8
Smoking
Current	25	15.6
Former	93	58.1
Never	42	26.3
Alcohol Problem (AUDIT ≥8)
Yes	26	16.2
No	134	83.8
BMI (kg/m2)
Mean	27.2
SD	5.5
Range	13.1 – 54.4

^†HPV-status was available for 122 study participants

Three major dietary patterns emerged from PCA. The first pattern, termed the “whole foods pattern”, was characterized by high intakes of vegetables, fruits, legumes, fish, poultry, fruit juice, water, and wine, and low intakes of sugar sweetened beverages and beer. The second pattern, termed the “Western pattern”, was characterized by high intakes of red and processed meats, refined grains, condiments, eggs, coffee, butter, and high-fat dairy, and low intakes of fruits, legumes and cereals. The third pattern, termed the “convenience foods pattern,” was characterized by high intakes of cereals, pizza, desserts, diet beverages, and energy bars, and low intakes of other vegetables and beer.

IL-6, TNF-α, and IFN-γ levels significantly decreased across increasing quartiles of whole foods pattern score (Table 2). IL-6, TNF-α, and IFN-γ levels were also significantly inversely associated with total reported carotenoid intake. Significant inverse associations also were observed across increasing quartiles of lycopene intake for IL-6 and total vitamin E intake for IFN-γ. No significant associations were reported for relationships between IL-6 and TNF-α and serum carotenoids (Table 3). There was a significant trend towards decreased IFN-γ levels across increasing quartiles of serum α-tocopherol and β-cryptoxanthin levels. Results did not differ significantly when stratified by cancer stage, nor when analyses were performed that included HPV-status as a covariate on the subset of 122 patients on which HPV-status was available.

Table 2.

Multivariable^† odds ratios and 95% confidence intervals for proinflammatory cytokine levels by quartile (Q) of FFQ reported dietary patterns and nutrient intake^‡ (n=160)

	IL-6§
	Q2	Q3	Q4	Ptrend
Whole foods pattern	0.41 (0.16, 1.04)	0.32 (0.12, 0.86)	0.16 (0.06, 0.45)	<0.001*
Western pattern	1.19 (0.49, 2.87)	0.48 (0.19, 1.22)	0.70 (0.27, 1.77)	0.20
Convenience foods pattern	0.90 (0.36, 2.23)	1.19 (0.49, 2.90)	0.80 (0.32, 1.98)	0.80
Total vitamin E (μg/day)	0.81 (0.32, 2.01)	0.90 (0.36, 2.26)	0.54 (0.22, 1.34)	0.23
α-carotene (μg/day)	0.48 (0.20, 1.18)	0.49 (0.19, 1.24)	0.61 (0.24, 1.60)	0.33
β-carotene (μg/day)	0.57 (0.23, 1.42)	0.55 (0.21, 1.43)	0.59 (0.22, 1.59)	0.31
Lycopene (μg/day)	1.01 (0.39, 2.58)	1.09 (0.40, 2.97)	0.35 (0.13, 0.94)	0.03*
β-cryptoxanthin (μg/day)	0.30 (0.11, 0.81)	0.64 (0.25, 1.63)	0.40 (0.15, 1.04)	0.27
Lutein + zeaxanthin (μg/day)	0.67 (0.28, 1.64)	0.57 (0.23, 1.43)	0.45 (0.17, 1.20)	0.11
Total carotenoids (μg/day)	0.42 (0.17, 1.06)	0.59 (0.23, 1.53)	0.19 (0.23, 1.53)	0.006*

	TNF-α ‡‡
	Q2	Q3	Q4	Ptrend
Whole foods pattern	0.41 (0.14, 1.18)	0.29 (0.09, 0.92)	0.19 (0.06, 0.64)	0.008*
Western pattern	0.91 (0.34, 2.45)	0.62 (0.21, 1.81)	0.63 (0.21, 1.83)	0.31
Convenience foods pattern	1.21 (0.43, 3.43)	1.18 (0.42, 3.25)	0.44 (0.14, 1.33)	0.16
Total vitamin E (μg/day)	0.92 (0.32, 2.63)	1.09 (0.38, 3.14)	0.76 (0.27, 2.16)	0.69
α-carotene (μg/day)	0.42 (0.15, 1.22)	0.60 (0.21, 1.72)	0.72 (0.25, 2.10)	0.67
β-carotene (μg/day)	0.36 (0.12, 1.05)	0.30 (0.10, 0.95)	0.60 (0.20, 1.83)	0.44
Lycopene (μg/day)	0.45 (0.15, 1.37)	0.99 (0.32, 3.11)	0.24 (0.07, 0.84)	0.08
β-cryptoxanthin (μg/day)	0.52 (0.17, 1.59)	0.61 (0.21, 1.77)	0.69 (0.24, 2.02)	0.67
Lutein + zeaxanthin (μg/day)	0.48 (0.17, 1.36)	0.68 (0.25, 1.90)	0.31 (0.10, 0.99)	0.09
Total carotenoids (μg/day)	0.38 (0.13, 1.09)	0.64 (0.22, 1.88)	0.22 (0.07, 0.72)	0.04*

	IFN-γ §
	Q2	Q3	Q4	Ptrend
Whole foods pattern	0.58 (0.23, 1.48)	0.40 (0.15, 1.10)	0.32 (0.12, 0.90)	0.03*
Western pattern	0.90 (0.37, 2.17)	0.75 (0.29, 1.92)	1.33 (0.52, 3.42)	0.64
Convenience foods pattern	1.34 (0.53, 3.36)	1.59 (0.64, 3.94)	0.84 (0.34, 2.12)	0.79
Total vitamin E (μg/day)	0.53 (0.21, 1.36)	0.32 (0.12, 0.84)	0.27 (0.10, 0.70)	0.004*
α-carotene (μg/day)	0.76 (0.31, 1.88)	0.79 (0.31, 2.01)	0.79 (0.31, 2.07)	0.67
β-carotene (μg/day)	0.87 (0.35, 2.20)	0.46 (0.17, 1.23)	0.61 (0.22, 1.65)	0.21
Lycopene (μg/day)	0.85 (0.33, 2.21)	1.17 (0.43, 3.21)	0.60 (0.22, 1.62)	0.38
β-cryptoxanthin (μg/day)	0.16 (0.06, 0.47)	0.51 (0.20, 1.35)	0.27 (0.10, 1.35)	0.14
Lutein + zeaxanthin (μg/day)	1.08 (0.44, 2.68)	0.58 (0.23, 1.47)	0.56 (0.21, 1.49)	0.14
Total carotenoids (μg/day)	0.45 (0.18, 1.16)	0.69 (0.26, 1.81)	0.29 (0.10, 0.78)	0.04*

^†Adjusted for age, sex, tumor site, cancer stage, smoking, and alcohol problem

^‡Odds ratios are comparing each of the upper quartiles to the lowest quartile 1

^*Indicates significance at P < 0.05

^§Odds ratios for higher cytokine value versus lower cytokine value for cytokines categorized into three levels

^‡‡Odds ratios comparing non-zero cytokine level versus zero for cytokines categorized into two levels

Table 3.

Multivariable^† odds ratios and 95% confidence intervals for proinflammatory cytokine levels by quartile (Q) of nutrient biomarker levels^‡ (n=151)

	IL-6§
	Q2	Q3	Q4	Ptrend
Biomarker (μg/ml)
α-tocopherol	0.80 (0.33, 1.93)	1.07 (0.44, 2.64)	1.21 (0.49, 2.98)	0.56
γ-tocopherol	1.80 (0.74, 4.37)	0.92 (0.38, 2.21)	1.23 (0.51, 2.96)	0.97
α-carotene	0.84 (0.34, 2.05)	0.97 (0.38, 2.48)	0.97 (0.36, 2.59)	0.98
β-carotene	0.87 (0.35, 2.15)	1.48 (0.58, 3.75)	1.25 (0.50, 3.15)	0.43
Lycopene	1.02 (0.42, 2.48)	0.64 (0.26, 1.59)	0.78 (0.30, 1.99)	0.43
β-cryptoxanthin	0.30 (0.11, 0.81)	0.64 (0.25, 1.63)	0.40 (0.15, 1.04)	0.27
Lutein	0.83 (0.34, 2.02)	0.52 (0.20, 1.35)	0.97 (0.38, 2.52)	0.73
Zeaxanthin	0.81 (0.33, 1.97)	0.77 (0.31, 1.91)	0.60 (0.23, 1.52)	0.29
Total carotenoids	0.63 (0.26, 1.55)	1.15 (0.45, 2.93)	0.71 (0.27, 1.81)	0.70

	TNF-α ‡‡
	Q2	Q3	Q4	Ptrend
Biomarker (μg/ml)
α-tocopherol	1.39 (0.47, 4.06)	2.70 (0.90, 8.10)	1.92 (0.65, 5.69)	0.28
γ-tocopherol	1.98 (0.61, 6.41)	2.10 (0.68, 6.50)	1.85 (0.57, 6.03)	0.33
α-carotene	2.54 (0.73, 8.89)	2.44 (0.71, 8.33)	1.51 (0.44, 5.19)	0.57
β-carotene	2.45 (0.75, 8.02)	3.31 (0.91, 12.06)	1.02 (0.26, 3.99)	0.96
Lycopene	1.28 (0.42, 3.88)	0.88 (0.27, 2.86)	1.35 (0.41, 4.45)	0.72
β-cryptoxanthin	2.25 (0.70, 7.28)	1.39 (0.44, 4.37)	1.18 (0.34, 4.06)	0.92
Lutein	1.73 (0.54, 5.53)	2.33 (0.68, 7.97)	2.35 (0.71, 7.80)	0.16
Zeaxanthin	1.14 (0.35, 3.71)	2.31 (0.76, 6.97)	0.80 (0.23, 2.77)	0.88
Total carotenoids	0.79 (0.26, 2.41)	1.21 (0.38, 3.87)	0.87 (0.26, 2.91)	0.93

	IFN-γ §
	Q2	Q3	Q4	Ptrend
Biomarker (μg/ml)
α-tocopherol	0.99 (0.41, 2.42)	0.90 (0.35, 2.26)	0.62 (0.25, 1.57)	0.03*
γ-tocopherol	2.68 (0.99, 7.27)	1.61 (0.62, 4.22)	1.39 (0.51, 3.78)	0.90
α-carotene	1.51 (0.54, 4.27)	1.51 (0.54, 4.21)	1.20 (0.42, 3.40)	0.76
β-carotene	0.88 (0.34, 2.32)	1.62 (0.55, 4.76)	0.91 (0.30, 2.78)	0.91
Lycopene	0.98 (0.38, 2.53)	0.60 (0.22, 1.68)	1.25 (0.44, 3.52)	0.82
β-cryptoxanthin	0.94 (0.35, 2.48)	0.60 (0.23, 1.57)	0.35 (0.12, 1.01)	0.03*
Lutein	1.07 (0.42, 2.73)	0.94 (0.34, 2.64)	0.73 (0.27, 1.99)	0.51
Zeaxanthin	0.68 (0.26, 1.79)	0.72 (0.28, 1.85)	0.42 (0.15, 1.20)	0.14
Total carotenoids	0.76 (0.30, 1.95)	0.61 (0.22, 1.66)	0.53 (0.18, 1.54)	0.23

^†Adjusted for age, sex, tumor site, cancer stage, smoking, and alcohol problem

^‡Odds ratios are comparing each of the upper quartiles to the lowest quartile 1

^*Indicates significance at P < 0.05

^§Odds ratios for higher cytokine value versus lower cytokine value for cytokines categorized into three levels

^‡‡Odds ratios comparing non-zero cytokine level versus zero for cytokines categorized into two levels

Discussion

High whole foods dietary pattern scores and reported intake of total carotenoids before treatment were associated with significantly lower serum levels of proinflammatory cytokines IL-6, TNF-α, and IFN-γ, independent of other factors known from previous research to modulate cytokine levels.³⁵ Additionally, higher reported intake of lycopene was significantly associated with lower IL-6 levels, and higher reported vitamin E intake and serum α-tocopherol levels were associated with lower IFN-γ levels.

To our knowledge, this is the first study to examine the relationship of dietary patterns and carotenoids with serum proinflammatory cytokine levels in HNSCC patients. Our results support findings from prior research examining this relationship in other study populations. In a prospective analysis conducted among a population of older adults as part of the Health ABC Study, a ‘healthy’ dietary pattern (consistent with our whole foods dietary pattern) was associated with lower IL-6 levels than a dietary pattern high in sweets and high-fat dairy products.³⁶ An intervention trial conducted among males with metabolic syndrome reported that those randomized to a Mediterranean diet, rich in foods characteristic of our whole foods dietary pattern, had significantly lower IL-6 and TNF-α after five weeks compared to the control group.³⁷

Our findings show a statistically significant inverse association between IL-6 levels and reported total carotenoids. Similarly, higher serum concentrations of total carotenoids were found to be correlated with lower IL-6 and TNF-α levels in subjects with recent hip fractures.³⁸ In contrast, a case-control study of breast cancer risk reported no association between IL-6 levels and reported intake of β-carotene.³⁹ Low α-tocopherol plasma concentrations were associated with increased expression of the gene encoding IFN-γ in another study conducted among breast cancer cases.⁴⁰ These results are consistent with our findings that high reported intake of total vitamin E and serum levels of α-tocopherol were associated with lower levels of IFN-γ in HNSCC subjects.

One mechanism by which the whole foods dietary pattern, carotenoids, and tocopherols may help to reduce circulating cytokine levels in HNSCC patients is through modulation of the NF-κB pathway. A number of epidemiologic and clinical studies have implicated NF-κB as a driver of cancer progression, making this pathway a potential target for therapeutic intervention.⁴¹ IL-6, TNF-α and IFN-γ can all stimulate the NF-κB pathway.^{41, 42} Oxidative stress has also been reported to activate the NF-κB family of transcription factors, further upregulating expression of proinflammatory cytokines.^{39, 43} Nutrients that can reduce oxidative stress and inflammation in the body such as carotenoids and tocopherols—both found in abundance in foods that characterize the whole foods dietary pattern—have been shown to suppress NF-κB in laboratory studies.⁴⁴

We found several significant associations between reported dietary variables and cytokine levels, but did not find any significant associations between cytokine levels and serum carotenoids and tocopherols, with the exception of α-tocopherol and β-cryptoxanthin with IFN-γ. This suggests perhaps other dietary variables that are highly correlated with dietary carotenoid intakes may be driving the associations of reported carotenoid intake with cytokine levels. These could include bioactive compounds found in fruits and vegetables such as polyphenolic compounds and flavonoids. Flavonoids have been shown to inhibit NF-κB signaling and downregulate the expression of proinflammatory markers; and, therefore, should be examined in relation to proinflammatory cytokines in HNSCC in the future.⁴⁵

This study used comprehensive dietary and serum data collected from subjects and had the ability to adjust for multiple potential confounding factors, which strengthened our conclusions. No causal inferences could be made due to the cross-sectional design, which is a limitation of the study. Random measurement error in dietary reporting may have led to misclassification bias. However, such bias is likely to attenuate associations towards the null.²³ Data on additional variables that may have influenced results such as steroid use or autoimmune history of patients were not readily available, which is a limitation of the study.

In summary, pretreatment diet, particularly a whole foods dietary pattern and total carotenoid intake may be associated with lower systemic inflammation in newly diagnosed HNSCC cases, as measured by serum levels of IL-6, TNF-α, and IFN-γ. These associations may provide mechanistic evidence for our previous described findings of the association between the whole foods dietary pattern and prolonged survival.⁵ The results of this study can serve as the basis for translational intervention research in HNSCC populations, where the aim is to increase the consumption of foods characteristic of the whole foods dietary pattern, and abundant in dietary carotenoids.