This study utilized data from the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial, a large-scale randomized controlled trial that enrolled 154,887 participants aged 55–74 years from 10 screening centers across the United States between 1993 and 2001 (15). The PLCO trial was designed to evaluate cancer screening effectiveness in reducing mortality from prostate, lung, colorectal, and ovarian cancers, with participants randomly assigned to either screening intervention or usual care control groups. Follow-up continued through 2009 for cancer incidence and through 2018 for mortality outcomes. As part of the trial protocol, participant information was systematically collected through a Baseline Questionnaire (BQ) and a Dietary History Questionnaire (DHQ).

To investigate the association between FQI and HNC incidence, we excluded participants with incomplete BQ, missing or invalid DHQ, a history of any cancer before study entry, an HNC diagnosis occurring before completion of the DHQ, or missing key covariates. Participants meeting these criteria were included in the final analytical cohort; the selection process is shown in Figure 1. The original PLCO trial received approval from the National Cancer Institute and Institutional Review Boards at each screening center, with informed consent obtained from all participants. The current analysis was approved by the NCI under Project ID PLCO-1886.

The participants’ baseline information, including demographic, lifestyle, and clinical factors, was collected through a self-administered BQ. In this study, the following variables were included as covariates: age (continuous), sex (male/female), race (Non-Hispanic White, Non-Hispanic Black, Hispanic, Other), education (some college or less, college graduate, postgraduate), body mass index (BMI; kg/m², calculated as weight in kilograms divided by height in meters squared), pipe smoking status (never, former, current), cigar smoking status (never, former, current), daily cigarette consumption (0 per day, 1–20 per day or >20 per day), alcohol drinking status (no or yes), alcohol consumption (continues, g/days), and family history of HNC (no, yes or possibly).

The DHQ is a validated 137-item self-administered food frequency questionnaire (FFQ) designed to assess dietary intake over the previous 12 months, administered approximately 3 years after enrollment (16). Daily intakes are estimated by combining reported consumption frequency with portion size, while daily energy and nutrient intakes are derived using the nutrient database established from the 1994–1996 US Department of Agriculture Continuing Survey of Individual Intakes (17). The DHQ enables comprehensive evaluation of energy, macronutrient and fatty acid contributions, and fruit and vegetable consumption. Its validity was confirmed in the “Eating at America’s Table” study using 24-h recalls from a nationally representative sample of 1,640 US participants. DHQ performed favorably relative to the other two commonly used FFQs (Block and Willett questionnaires) for absolute intake estimation (16).

The FQI was calculated as the ratio of beneficial fatty acids (monounsaturated fatty acids [MUFA] and polyunsaturated fatty acids [PUFA]) to potentially harmful fatty acids (saturated fatty acids [SFA] and trans fatty acids [TFA]): FQI = (MUFA + PUFA)/(SFA + TFA) (13). Participants were categorized into quartiles based on their FQI scores for analysis.

HNC cases were identified through annual study update questionnaires sent to all surviving participants, requesting information on cancer diagnoses including type, date, anatomical site, and healthcare provider details. All self-reported cancers were verified through medical records using standardized data extraction forms. Cancer cases and anatomical locations were confirmed by study physicians blinded to participants’ exposure and risk factor status. HNC was defined according to the International Classification of Diseases for Oncology, Second Edition (ICD-O-2), encompassing oral cavity and pharyngeal cancers (C00.1–C06.9, C09.0–C14.0), laryngeal cancer (C32.0–C32.9), and nasal cavity and middle ear cancers (C30.0, C30.1) (18, 19). The primary outcome was overall HNC incidence. Secondary analyses examined oral cavity/pharyngeal cancer and laryngeal cancer separately.

The Cox proportional hazards regression model was used to estimate the hazard ratio (HR) and 95% confidence interval (CI) for the association between FQI and the incidence of HNC, with follow-up duration as the time measure. The follow-up period was defined as the time from completion of the first DHQ to the date of exit, defined as the earliest occurrence of HNC diagnosis, death, loss to follow-up, or end of follow-up (December 31, 2009). To examine the linear trend across FQI quartiles, each participant was assigned the median value of their respective quartile. This value was modeled as a continuous variable in the regression analysis, with the lowest quartile as the reference group. The p-value for trend was used to assess the statistical significance of the linear association. Additionally, FQI was modeled as a continuous variable to estimate the risk associated with each unit increase. Based on prior literature and clinical expertise (18, 20, 21), potential confounding factors were identified and included in the Cox regression models to control for confounding. Model 1 adjusted for age, sex, race, and education level. Model 2 further adjusted for family history of HNC, baseline BMI, pipe and cigar smoking status, daily cigarette consumption, alcohol consumption, alcohol use, total protein and carbohydrate intake, energy-adjusted fruit and vegetable intake, and total energy intake.

For secondary outcomes, we applied the same Cox regression framework to assess the associations between FQI and the incidence of site-specific HNC, including oral cavity/pharyngeal cancer and laryngeal cancer. To further explore whether specific types of fatty acids contribute differently to HNC risk, we conducted parallel analyses treating each component of the FQI—monounsaturated, polyunsaturated, saturated, and trans fatty acids—as separate variables in the model. To characterize the dose–response relationship between FQI and HNC incidence across its full range, we fitted restricted cubic spline models with knots placed at the 10th, 50th, and 90th percentiles of FQI distribution (22). The median value of the first quartile (1.27) was used as the reference point. Nonlinearity was assessed by testing the null hypothesis that the coefficient of the second spline term equals zero.

To assess the combined impact of dietary fat quality and quantity on HNC risk, participants were categorized into tertiles based on FQI scores and dichotomized by fat-derived energy percentage (median: 32%), creating six joint exposure groups. Multivariable Cox proportional hazards models estimated hazard ratios (HRs) and 95% confidence intervals (CIs) for HNC incidence, using the lowest FQI tertile with low fat intake (<32%) as the reference. Models were adjusted for covariates described in Model 2. Interaction between FQI and fat intake was tested using a multiplicative term, with significance evaluated by likelihood ratio test.

To explore potential effect modification, we conducted pre-specified subgroup analyses stratified by age (<65 vs. ≥65 years), sex, family history of HNC (absent vs. present/possible), smoking status (pipe: never vs. ever; cigarettes: never vs. ever), daily alcohol consumption (≤1.41 vs. >1.41 g/day), and baseline BMI (<25, 25–30, ≥30 kg/m²). Interaction was formally tested by including multiplicative terms in Cox models and comparing nested models using likelihood ratio tests. Within each subgroup, trend tests across FQI quartiles were performed by treating quartile medians as continuous variables in the regression models.

To assess the robustness of our findings, we conducted four sensitivity analyses: (1) excluding participants with implausible energy intake (<500 or >4,000 kcal/day) to minimize dietary reporting errors; (2) excluding those with extreme BMI values (top and bottom 1%) to reduce potential confounding by severe underweight or obesity; (3) excluding HNC cases diagnosed within the first 2 years of follow-up to minimize potential reverse causation; and (4) excluding participants with family history of HNC to focus on sporadic cases. Statistical analyses were performed using R software (version 4.4.2) with survival and splines packages. All tests were two-sided with statistical significance set at p < 0.05.

After applying the exclusion criteria, a final cohort of 98,560 participants was included in the analysis (Figure 1). The mean age of the participants was 65.5 (Standard Deviation, 5.7) years, and they were followed for a mean duration of 8.8 (SD, 1.9) years. The baseline characteristics of the study population, stratified by quartiles of the FQI, are detailed in Table 1. Significant differences were observed across FQI quartiles. Specifically, individuals in the highest FQI quartile (Q4), compared to those in the lowest (Q1), were more likely to be female, younger, and have a college degree or higher. They also reported less family history of HNC and a lower prevalence of heavy smoking. As would be expected by its definition, a higher FQI score was directly correlated with higher intakes of MUFAs and PUFAs, and inversely with SFAs and TFAs. Beyond these constitutive associations, individuals with a higher FQI also consumed more vegetables and fruits, and had a lower total energy intake.

Over a mean follow-up of 8.8 ± 1.9 years, corresponding to 872,167 person-years, a total of 267 incident cases of HNC were documented, including 166 cases of oral cavity and pharyngeal cancer, 91 cases of laryngeal cancer, and 10 cases involving other anatomical subsites. This yields an overall incidence rate of approximately 0.306 per 1,000 person-years. In the primary analysis treating FQI as a categorical variable, a significantly lower HNC risk was observed in the highest FQI quartile compared to the lowest (fully adjusted HR for Q4 vs. Q1: 0.62; 95% CI, 0.42–0.91), with a significant dose–response trend across quartiles (P-trend = 0.011). This inverse association was consistent when FQI was modeled as a continuous variable, where each standard deviation increase in the FQI score was associated with a 22% lower risk of HNC (fully adjusted HR: 0.78; 95% CI, 0.68–0.90). To further characterize this relationship, a restricted cubic spline analysis revealed a linear inverse dose–response relationship between FQI and HNC risk, with no evidence of nonlinearity (P-nonlinearity = 0.699; Figure 2A) per SD (Table 2).

When examining HNC by anatomical subsite, the inverse association was most pronounced for laryngeal cancer. In fully adjusted models, both categorical (HR for Q4 vs. Q1: 0.43; 95% CI: 0.21–0.88; P-trend = 0.05) and continuous analyses (HR per SD: 0.69; 95% CI: 0.53–0.89) revealed a significant protective effect of higher FQI. For oral cavity and pharyngeal cancer, a similar inverse trend was observed, although the association was not statistically significant when FQI was treated as a categorical variable (HR for Q4 vs. Q1: 0.75; 95% CI, 0.47–1.20). However, the continuous analysis did yield a significant association (HR per SD: 0.83; 95% CI, 0.70–0.99). Despite some differences in statistical significance, restricted cubic spline models indicated linear inverse dose–response relationships for both subsites (P-nonlinearity > 0.05 for both; Figures 2B,C).

In further subgroup analyses, the inverse association between FQI and HNC risk remained consistent across various participant characteristics, including age, sex, family history of HNC, smoking status, alcohol consumption, and baseline BMI. We found no evidence of statistically significant effect modification by any of these factors (all p-values for interaction > 0.05; Figure 3), suggesting the robustness of our primary finding.

The associations between different subtypes of dietary fatty acids and the risk of HNC and its anatomical subsites are presented in Table 3. A higher intake of MUFA was significantly associated with a reduced risk of HNC, with individuals in the highest quartile of MUFA consumption exhibiting a 45% lower risk compared to those in the lowest quartile (HR = 0.55; 95% CI: 0.33–0.91; p for trend = 0.012). A similar inverse association was also evident when MUFA intake was treated as a continuous variable (HR per sd increment = 0.75; 95% CI: 0.61–0.92). These protective effects extended to cancers of the oral cavity and pharynx (HR for Quartile 4 vs. Quartile 1 = 0.45; 95% CI: 0.24–0.84; p for trend = 0.016; HR per SD increment = 0.72; 95% CI: 0.56–0.93), while no statistically significant association was found for laryngeal cancer. A parallel pattern was observed for PUFA, with the highest PUFA intake quartile associated with a substantially decreased risk of HNC (HR for Quartile 4 vs. Quartile 1 = 0.63; 95% CI: 0.41–0.95; p for trend = 0.013; HR per SD increment = 0.77; 95% CI: 0.66–0.91). This protective association extended to oral cavity and pharyngeal cancer when PUFA was treated as a continuous variable (HR per SD increment = 0.76; 95% CI: 0.62–0.92), while the comparison of extreme quartiles showed a similar but non-significant effect (HR = 0.61; 95% CI: 0.36–1.05; p for trend = 0.027). In contrast, no significant associations were detected between SFA and HNC or its subsites. Similarly, TFA showed no meaningful correlation with HNC incidence, although a non-significant elevated risk was noted among participants in the highest TFA quartile (HR for Quartile 4 vs. Quartile 1 = 1.30; 95% CI: 0.96–1.76; p for trend = 0.359).

Figure 4 illustrates the HR for HNC according to the joint stratification by the tertiles of the FQI and the proportion of energy intake from fat (<32% vs. ≥32%). Across all FQI tertiles, individuals with a higher proportion of energy from fat (≥32%) demonstrated consistently lower HR of HNC compared to their counterparts with lower fat intake. Notably, in the highest FQI tertile (Tertile 3), the HR for HNC was substantially lower among participants with fat intake ≥32% (HR = 0.36; 95% CI: 0.21–0.59), relative to the reference group (Tertile 1 with fat <32%). However, no significant interaction was observed between dietary fat intake and FQI on HNC risk (p for interaction = 0.895), suggesting that the protective association of higher FQI is not strongly modified by the level of fat consumption.

A series of sensitivity analyses were conducted to evaluate the robustness of the primary findings. Overall, the inverse association between FQI and the risk of HNC remained largely unchanged across various analytical scenarios.