Introduction:

In the era of precision medicine, there is a continuous pursuit for less invasive and more powerful methods to screen for diseases, predict treatment responses, and detect recurrences. While tools like circulating tumor DNA (ctDNA) and other forms of liquid biopsy have proven useful in managing urothelial cancer, similar diagnostic and prognostic tools for renal cell carcinoma (RCC) remain to be elucidated. We therefore aimed to investigate serum protein expression in a cohort of patients before their RCC diagnosis compared to non-cancer controls. Our goal was to identify protein-cancer associations most likely to have a causal role in cancer development and progression.

Methods:

Utilizing the UK Biobank, a prospective database from the United Kingdom including data for over 500,000 patients, we analyzed serum proteomic data encompassing over 2900 proteins using Olink in a cohort of patients previously diagnosed with RCC (n=2245). Additionally, we examined data for patients without evidence of RCC at the time of serum collection who were later diagnosed with RCC (n=1225). Both groups were compared to a third cohort of non-cancer controls (n=44386). All groups were adjusted for age, gender, BMI, and smoking status. All analyses were conducted using Stata release 18.1 and R version 4.1.2. We estimated hazard ratios (HRs) and 95% confidence intervals (CI) using Cox proportional hazards regression models. We investigated protein and RCC cancer-risk associations to examine the effects of reverse causality. Additionally, we conducted sensitivity and specificity analyses and generated area under the curve (AUC) values to evaluate diagnostic performance of the identified biomarkers.

Results:

We identified associations between elevated levels of tafa5, klk11, klk8, csdn, and clmp proteins and undiagnosed RCC (Figure 1A). All five most expressed proteins in each cohort showed increased hazard ratios (HR) for cancer association (95% CI), with tafa5 exhibiting the highest HR in the treated vs healthy group HR 9.7 (95% CI 6.3-12.01,p<0.001)(Figure 1B). Receiver operating characteristic (ROC) curve analysis demonstrated significant diagnostic performance, with AUC values indicating strong predictive capabilities: igfbp4 (AUC 0.9127, SN 0.8056, SP 0.8606), tafa5 (AUC 0.9134, SN 0.7500, SP 0.8996), tgfbr2 (AUC 0.9057, SN 0.6667, SP 0.9083), scarb2 (AUC 0.8968, SN 0.7500, SP 0.9197), and nectin4 (AUC 0.9269, SN 0.7222, SP 0.9299) (Figures 1C and 1D). These findings indicate that specific proteomic biomarkers may have a significant association with RCC diagnosis, with high AUC values supporting their potential utility in RCC diagnosis and prognosis. Prospective validation in clinical settings is warranted.

Conclusion:

In the era of personalized oncologic treatment, the search for reliable biomarkers for RCC remains ongoing. Our study highlights distinct protein elevation profiles between patients presenting with RCC and those initially disease-free but later diagnosed with RCC. Although further studies are warranted, the favorable HR, AUC, SN, and SP values for several proteins identified in this study underscore their potential as attractive biomarkers for RCC diagnosis and prognosis.

Funding: N/A