Longgang Zhao

Date of Award

Spring 2023

Document Type

Open Access Dissertation


Epidemiology and Biostatistics

First Advisor

Susan E. Steck


Background: Studies have shown positive associations between ultra-processed food (UPF) intake and obesity and type 2 diabetes, two risk factors for liver cancer. However, the relationship between UPF intake and liver cancer has scarcely been studied. We aimed to evaluate the association between UPF intake and the risk of liver cancer in three prospective cohorts and further explore the role of genetic predisposition to fatty liver diseases and biomarkers of liver health on modifying the association.

Methods: We used data from the Nurses’ Health Study (1986-2012), the Health Professionals Follow-up Study (1986-2012), and UK Biobank (2006-2020). Diet was assessed with food frequency questionnaires or 24-hour dietary recalls. UPF was defined according to the NOVA classification system. We employed Cox proportional hazards regression to estimate hazard ratios (HR) and 95% confidence intervals (CIs) for evaluating the association between UPF intake and liver cancer risk. We adjusted the model for potential confounders such as age, sex, race, smoking status, alcohol use, physical activity, aspirin use, body mass index, and history of diabetes. We tested multiplicative interaction terms using Wald tests by including a product of UPF intake and modifiers (genetic risk score [GRS], C-reactive protein [CRP], and fatty liver index [FLI]).

Results: Among 153,508 participants from the NHS and HPFS cohorts with an average follow-up period of 24.2 years, 173 new cases of liver cancer were documented. The analysis found no significant association between UPF intake as a percentage of energy intake and risk of liver cancer, after adjusting for potential confounders (HR Tertile 3 vs. tertile 1=1.12, 95% CI: 0.77-1.64; HR Per SD=1.01, 95% CI: 0.0.83-1.23). Among different UPF components, intake of meat and meat-substitute-based products was positively associated with liver cancer risk (HRTertile 3 vs. tertile 1=1.57, 95% CI: 1.06-2.32; Ptrend=0.02). In the UK Biobank study of 175,530 participants with a median follow-up of 8.9 years, 154 cases of liver cancer were recorded. The results of the age-adjusted model revealed a positive association between higher intake of UPF and liver cancer (HR Per 500 kcal/day from UPF=1.20, 95% CI=1.01-1.43). However, no association was found after controlling for potential confounders (HR Per 500 kcal/day from UPF=1.02, 95% CI=0.81-1.29). We confirmed positive associations between GRS, CRP, and FLI and risk of liver cancer. We also found that higher UPF intake was associated with adverse levels of serum CRP, HbA1c, and alkaline phosphatase. However, we did not find evidence for effect modification by genetic predisposition, serum CRP or FLI on the UPF-liver cancer association in the UK Biobank (Pinteraction>0.05).

Conclusion: The overall analysis did not provide substantial evidence linking higher UPF intake to an increased risk of liver cancer. Moreover, the relationship between UPF intake and liver cancer was not modified by genetic predisposition or clinical risk factors for fatty liver disease. However, further research with a larger sample size of liver cancer cases is warranted to confirm these findings.


© 2023, Longgang Zhao

Available for download on Thursday, May 15, 2025

Included in

Epidemiology Commons