Using single-cell and bulk RNA data to predict outcomes and immunotherapy response in prostate cancer

Prostate cancer (PCa) is primarily an epithelial malignancy that occurs in the prostate gland and is one of the most common cancers among men worldwide. Although the cause of PCa is not known, studies have shown that factors such as age, family history, ethnicity, and gene regulation are all involved in its development¹. PCa exhibits geographical variability in its incidence, being more prevalent in developed nations, especially in North America and Europe, when compared to their developing counterparts. Age is a significant risk factor, with the incidence increasing substantially after the age of 50². Several racial and ethnic disparities have been observed in PCa epidemiology. It is more common in African-American men compared to Caucasian men, and African-American men also tend to have a higher mortality rate associated with PCa³. As a consequence of the indiscernible clinical manifestations of early-stage prostate cancer, an approximate majority of patients are diagnosed belatedly, at an intermediate or advanced stage, or when distant metastases have already occurred, resulting in elevated rates of patient mortality⁴. Therefore, early screening and diagnosis of PCa are essential to reducing its mortality rate. With the rapid development of genetic analysis techniques, we have achieved a deeper understanding of the molecular pathogenesis of PCa, which in turn can be used for early diagnosis and treatment in clinics^5,6.

The tumour microenvironment (TME) is a dynamic environment made up of multiple cells and is considered one of the most important factors influencing tumourigenesis, invasion, and metastasis⁷. Cancer-associated fibroblasts (CAFs), a major component of the TME, contribute to cancer progression by interacting with other cells to influence tumour angiogenesis and immunity evasion⁸. CAFs could induce abnormal proliferation and invasion of tumour cells by secreting large amounts of growth factors, cytokines, and chemokines and degrading extracellular matrix proteins⁹. It has been shown that CAFs are highly heterogeneous in esophageal squamous cell carcinoma and can be used as a valid predictive target for esophageal squamous cell carcinoma prognosis¹⁰. Studies have reported that aberrant molecular crosstalk between cancer cells and CAF is extremely important for tumour progression and metastasis, leading to poor prognosis in cancers^11,12. In addition, a foreign study found that CAFs promoted the aggressiveness of PCa cells through mitochondrial transfer¹³. Collectively, CAFs are closely related to the development of tumours.

Despite the voluminous body of research on CAFs carried out by a plethora of scholars, a comparatively insufficient comprehension of the properties of CAFs persists, including their correlation with prostate cancer prognosis and immunotherapy. With advances in sequencing technology, increasing numbers of scholars are turning their attention to single-cell RNA-sequencing (scRNA-seq). The scRNA-seq technology could focus on detecting the expression of single cell subpopulations in tissues, bridge the shortage of traditional sequencing technologies, and help us further explore and construct the gene expression profiles of various cells to better understand the pathological processes of diseases¹⁴. The scRNA-seq can analyze the composition and gene expression of thousands of individual cells at high resolution, making it better suited for exploring tumour heterogeneity¹⁵. More and more researches are combining RNA-seq and scRNA-seq data to analyze tumor heterogeneity and construct models and nomograms to predict cancer patient prognostic outcomes, thus providing new approaches to tumor clinical treatment^16,17. A study demonstrated that significant heterogeneity within CAFs was found in human PCa tissue and revealed that CAF subpopulations stimulate the growth and development of PCa by aberrant recruitment of immune cells¹⁸. However, at present, there are relatively few scRNA-seq analyses focused on exploring the correlation between CAFs and PCa. Therefore, we have explored the relationship between PCa and CAFs by combining bulk RNA-seq and scRNA-seq techniques to provide new ideas for the clinical diagnosis and treatment of PCa.

In the study, we acquired the scRNA-seq and RNA-seq data for PCa from the GEO and TCGA datasets and conducted batch correction to merge the two datasets. Then we analyzed the scRNA-seq data, constructed a prognostic model associated with PCa, and validated the model. We further identified differentially expressed genes (DEGs) between risk groups and performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses. We also explored differences in somatic mutations between different risk groups. Finally, we filtered 31 chemotherapeutic agents and targeted therapeutics that are closely related to PCa. Overall, our study may provide novel thinking on the pathogenesis of PCa and new perspectives for the personalized treatment of PCa patients.

PCa is one of the leading diseases affecting men's health around the world, and its incidence has been increasing in recent years³⁷. The current treatment of PCa is mainly based on hormonal therapy, radiotherapy, and chemotherapy, but the majority of patients have a worse prognosis and mortality remains high³⁸. As a consequence, there is a pressing demand in clinical practice to seek novel therapeutic strategies to develop individualized treatments for PCa patients in order to improve their prognosis. Numerous studies have confirmed that dynamic communication between tumour cells and CAFs accelerates the growth and progression of tumours^39,40. With the development of biotechnology, single-cell sequencing analysis can help us gain comprehension the mechanism of the role of CAFs in the pathological progression of PCa and more accurately search for potential new biomarkers, thus effectively improving therapeutic outcomes.

As tumour heterogeneity is quite pronounced, the response of the same stage patients to the same treatment methods has a strongly variable individual outcome and affects the prognosis of the patients⁴¹. Research on PCa has found significant differences in outcomes for patients with similar clinical stages who receive the same therapeutic approach⁴². Therefore, a more extensive investigation into the association between the diversity of tumour characteristics in PCa and its corresponding implications on treatment response and patient prognoses is warranted. In our study, based on scRNA-seq data, we divided the PCa cells into 4 clusters, screened 463 CAFRGs, and performed functional enrichment analysis of CAFRGs. On the basis of the screening above, we constructed a prognostic model for PCa incorporating 10 CAFRGs by using multivariate Cox analysis. Next, we validated the reliability and stability of the prognostic model using multiple datasets for internal and external validation, respectively. We investigated differences in somatic mutations between high-risk and low-risk groups. We also explored differences in the immune microenvironment landscape and ICG expression levels in the different groups. Lastly, we predicted the therapeutic response of PCa and searched for anti-tumour drugs by calculating the IC50 in different groups.

For our constructed prognostic model, we found that most of the 10 CAFRGs were involved in the pathogenesis of PCa. THBS1 could accelerate malignant cancer cell invasion in PCa and promote melanoma metastasis^43,44. LDHA is highly expressed in PCa, and inhibition of LDHA can delay the progression of PCa⁴⁵. A study found that MIR-138-5P could inhibit the development of PCa by targeting FOXC1⁴⁶. The expression level of NR4A1 in PCa tissue is significantly elevated in comparison to that in the normal surrounding tissue⁴⁷. MYC promoted PCa oncogenic signaling via the PI3K/AKT/mTOR pathway, thereby stimulating massive cancer cell growth⁴⁸. The most recent investigation has established that UBE2S, an E2 ubiquitin-conjugating enzyme, plays a critical role in stabilizing β-linked proteins via K16-linked ubiquitination, which is known to trigger enhanced migration and invasion of tumor cells in the context of prostate cancer bone metastases and promote oncogenic activities. Furthermore, it has been suggested that UBE2S-mediated ubiquitination of β-linked proteins may contribute to tumorigenesis and the progression of PCa bone metastases by facilitating alterations in key signaling pathways that promote tumor cell survival and proliferation⁴⁹. Although there is no direct evidence linking HOPX, LY6E, SAMD4A, and CRISPLD2 to the development of PCa, numerous studies have shown their close relationship with the growth and invasion of other tumours^50–53. The aforementioned findings have the potential to serve as novel biomarkers that can aid in the development of new therapeutic approaches for the treatment of PCa.

TP53 is a tumour suppressor gene that normally acts as a guardian to protect the organism⁵⁴. When TP53 is mutated, the body loses its protection, which may lead to uncontrolled tumour growth. Growing evidence has demonstrated the prevalence of mutated TP53 in cancer and its association with cancer susceptibility and drug resistance⁵⁵. The risk of aggressive PCa may be higher in men who have mutations in the TP53 gene⁵⁶. Our analysis results are generally consistent with previous studies. TTN is a commonly mutated gene in cancer and has a prominent role in predicting immune checkpoints and prognosis in solid tumours⁵⁷. TTN mutations have been shown to be an independent risk factor for thyroid cancer and to predict a poorer prognosis for patients⁵⁸. The sequencing data of lung adenocarcinoma patients showed that patients with TTN gene mutations had higher immunogenicity, immune scores, and inflammatory TME, which stimulates tumour growth⁵⁹. Nevertheless, the current number of studies on TTN in PCa is limited. Our study elucidates some references for the role of TTN in PCa.

The TME is a complex structure that provides the environment for tumor cell proliferation, invasion, and metastasis based on the presence of a large number of immune cells and immunomodulatory molecules in the tumor microenvironment⁶⁰. Our results revealed that the majority of immune cell infiltration levels were higher in the low-risk group. Significant increases in Treg cells have been found in many malignant cancers⁶¹. NK cells are core cells of the natural immune system that have been found to be abnormally expressed in prostate tumours, resulting in more rapid tumour development⁶². Breakthroughs have been made in NK cell-based cancer treatment, but there are still relatively few treatment strategies for PCa⁶³. Our findings may provide new inspiration for the utilization of NK cells in the treatment of PCa. And we found that ICGs such as CD274, LAG3, and PDCD1 expression differed between high- and low-risk groups. A study revealed that the CD274 (PD-L1) gene could stimulate the immune escape of tumor cells⁶⁴. High expression of LAG3 in peripheral blood T cells and lymphocytes is strongly correlated with a worse prognosis in PCa⁶⁵. The above studies are broadly consistent with our findings, demonstrating the reliability of our outcomes.

Despite the notable findings reported in this study, it is important to acknowledge certain limitations that may impact the interpretation of the results. Firstly, our analysis data is drawn from public datasets, and the results may be influenced by the datasets themselves, resulting in some possible bias in the results. Thus, we adopted several datasets for multiple validations from internal and external validation datasets, respectively, to support the reliability of the prognostic model we built. Secondly, we included a relatively smaller sample size of PCa, which may lead to analyses with some bias. Thirdly, due to the difficulty of collecting PCa tissue samples, we have not conducted experimental validation for the moment. In the future, we will endeavor to experimentally validate the results of our analyses, which will be the main direction of our future work.

Herein, we analyzed CAFs from PCa single-cell sequencing data, grouped them into four clusters, and identified CAFRGs. We also constructed a nomogram of PCa prognosis based on CAFRGs to predict patient prognosis and response to immunotherapy. In summary, our study will provide a novel clinical perspective on the targeted treatment of PCa and better predict the prognostic outcome of patients.