Renal cell carcinoma (RCC), prostate cancer (PCa), and bladder cancer (BC) are the most common tumors of the urinary system, of which the incidence has been rising rapidly in recent years. Renal cell carcinoma is a prevalent cancer in Western countries, accounting for about 3% of all cancers (1). Prostate and bladder cancer are the leading causes of male cancer death worldwide ([2], [3], [4]). Early screening contributes to lowering mortality. The main challenge is utilizing new technologies to improve early cancer detection.

The non-invasive approach is an ideal method for cancer screening. As a routine screening method, imaging has some disadvantages, such as relatively high cost, a long waiting time, and not being sensitive enough for very early stages. Computed Tomography may even have a radiative injury to the human body. Body fluids such as plasma and urine are easily collected samples for tumor screening. However, most urinary system-derived tumors lack specific biomarkers up to now.

Even if prostate-specific antigen has been widely used in prostate cancer, its specificity is still unsatisfactory (5). Many other diseases can also lead to prostate-specific antigen elevation, such as benign prostatic hyperplasia, prostatitis, urinary retention, etc. (6). Especially for the elderly with prostatic hyperplasia when the prostate-specific antigen is in the range of 4-10 ng/mL, it usually leads to an unnecessary prostate biopsy, which increases the patient's injury and infection risk and costs. The early stages of RCC and BC are generally asymptomatic. The screening mainly relies on ultrasound. Numerous patients are not diagnosed until gross hematuria appears (7). Some of them have already progressed to an advanced stage. Exploring simple and efficient cancer screening methods is an important topic.

Raman spectroscopy is an inelastic light scattering. It can provide information about molecular structural features and material composition, often called molecular fingerprints (8,9). However, the inherent flaw of Raman spectroscopy is that the average Raman scattering cross-section of the molecule is relatively small, leading to a very weak Raman signal. This defect seriously hinders the further development of Raman spectroscopy in clinical diagnosis ([10], [11], [12]). By adsorbing analytes onto the surface of gold or silver nanoparticles, surface-enhanced Raman spectroscopy (SERS) technology effectively improves the target Raman scattering signals, overcoming the limitations of Raman spectroscopy and greatly expanding its application in the biomedical fields ([13], [14], [15], [16]), such as rapid screening (17) and diagnosis for cancer disease (18).

In recent years, there has been an initial exploration of applying SERS to screening for urinary system tumors. Shaoxin Li et al. used SERS to differentiate bladder cancer patients from normal controls by testing their serum. They built a diagnostic model using six already known Raman spectral bands with an accuracy of 94.5% (19). Shou Chen et al. used SERS to detect urine samples and reported as high as 97.8% diagnostic accuracy in distinguishing between bladder cancer patients and normal controls (20). Dayu Hu et al. analyzed the urinary supernatant and sediment of bladder cancer patients using SERE combined with the Principal component analysis (PCA) and linear discriminant analysis (LDA) methods, and the result showed 100% diagnostic sensitivity and 98.85% specificity (21). In 2022, based on plasma SERS and three different machine algorithms, Sevda Mert et al. achieved an average accuracy of 0.77 in diagnosing kidney cancer compared with normal subjects (22). Shaoxin Li et al. measured SERS of plasma samples from prostate cancer patients and normal individuals and classified the two groups with an accuracy of 98.1% (23). Andrei Stefancu et al. also reported that combining serum SERS with serum PSA can increase prostate cancer diagnosis rate with an accuracy of 94% (5). Most previous studies used dichotomous methods that compared a single urologic tumor with a normal sample. We need to verify whether a new test or model can distinguish between different urinary tumors and the multi-classification approach is more in line with clinical practice. However, the use of SERS for a comprehensive analysis of multiple cancers of the genitourinary system is still lacking.

This study performed SERS detection followed by classification and regression multivariable statistic algorithms to screen urinary system tumors (BC, PCa, and RCC). To our knowledge, this is the first attempt to screen multiple urinary system tumors at the same time through SERS detection of plasma. Our findings demonstrate that the plasma-SERS in conjunction with the PCA-LDA statistic algorithm is an effective means to identify biochemical changes in blood and offers a potential choice for the clinical use of plasma samples in common urinary system cancer screening.