Using Latent Dirichlet Allocation Topic Modeling to Uncover Latent Research Topics and Trends in Renal Cell Carcinoma: Bibliometric Review

TL;DR

This study uses topic modeling to analyze 50 years of research on kidney cancer, identifying trends and future directions in treatment and understanding of the disease.

Contribution

The paper introduces a novel application of latent Dirichlet allocation to uncover hidden research themes and trends in renal cell carcinoma.

Findings

RCC research has shifted from diagnosis to understanding disease progression and genetic factors.

Emerging topics include diagnostic imaging, microRNA signatures, and drug resistance in RCC.

Neglected areas include ferroptosis, long-term treatment evaluation, and AI applications in RCC.

Abstract

Renal cell carcinoma (RCC) is a common, often lethal kidney cancer that originates in the renal cortex. Its incidence is rising, and major factors include smoking, obesity, and hypertension, though its etiology is uncertain. While surgery is effective for localized RCC, treatments for metastatic RCC have advanced significantly due to better diagnostic, prognostic, and predictive tools. Despite this progress, challenges remain, including long-term drug resistance and the complexity of RCC as a diverse group of diseases rather than a single entity.

The aim of this bibliometric review was a comprehensive analysis of the topics and trends in RCC research, offering a foundation for future investigations.

We used R “Bibliometrix” to conduct a bibliographic search in Scopus and PubMed covering publications from 1975 to 2023 to statistically assess the distribution of publications associated with RCC by year, journal, and country. Topic modeling of RCC research was conducted using latent Dirichlet allocation, a Bayesian network-based probabilistic algorithm that identifies unobserved thematic clusters in a collection of text documents. Trends in the retrieved themes were then characterized by using regression slopes over time, across countries, and in different journals. These trends were visualized as a heatmap, which was then used for hierarchical clustering to group similar topics based on their correlation strengths.

A total of 35,228 documents from 3070 sources were found, with a steady yearly growth of 9.86% and 118 participating countries. Thirty topics with the best coherence score were found in 8 crucial domains: treatment and therapies, biomolecular and genetic characteristics, disease characteristics and progression, diagnosis and evaluation, metastasis and dissemination, epidemiology and risk factors, related conditions, and pathological features. The pertinent clustergrams that resulted from the heatmaps mirrored the latent Dirichlet allocation’s algorithm identification of major RCC research subjects.

Over 50 years, RCC research’s focus has shifted from diagnosis and assessment to a more thorough understanding of disease characteristics and progression. Because many patients are diagnosed with abdominal imaging studies, an emerging topic in RCC is diagnostic imaging and radiological evolution. The advances in omics technologies and the function of microRNA signature in the progression, diagnosis, therapy targeting, and prognosis of RCC have garnered a lot of attention. The discovery of the genetic background has enhanced our understanding of the growth of RCC. Drug resistance, local RCC ablation, and postoperative surveillance of RCC recurrence following nephrectomy are key future research avenues. The next generation of drug-targeted therapy and immunotherapy will make it possible to successfully treat metastatic RCC following nephrectomy. Neglected topics include the association between ferroptosis and RCC, the long-term assessment of novel treatments, and the application of artificial intelligence on RCC. Our bibliographic review delivered pertinent data for clinical decision-making and the planning of future RCC research.

Figures