Predicting Oncologic Outcomes in Small Renal Tumors

Histopathology based AI model predicts anti-angiogenic therapy response in renal cancer clinical trial

Anti-angiogenic (AA) therapy is a cornerstone of metastatic clear cell renal cell carcinoma (ccRCC) treatment, but not everyone responds, and predictive biomarkers are lacking. CD31, a marker of vasculature, is insufficient, and the Angioscore, an RNA-based angiogenesis quantification method, is costly, associated with delays, difficult to standardize, and does not account for tumor heterogeneity. Here, we developed an interpretable deep learning (DL) model that predicts the Angioscore directly from ubiquitous histopathology slides yielding a visual vascular network (H&E DL Angio). H&E DL Angio achieves a strong correlation with the Angioscore across multiple cohorts (spearman correlations of 0.77 and 0.73). Using this approach, we found that angiogenesis inversely correlates with grade and stage and is associated with driver mutation status. Importantly, DL Angio expediently predicts AA response in both a real-world and IMmotion150 trial cohorts, out-performing CD31, and closely approximating the Angioscore (c-index 0.66 vs 0.67) at a fraction of the cost.

In Pursuit of KI-RADS: Toward a Single, Evidence-based Imaging Classification of Renal Masses

Despite the successful application of Imaging Reporting and Data Systems to improve the radiologic description and management of disease in many organs, one does not yet exist for the kidney. Instead, the radiologic approach to the kidney has focused on the Bosniak classification system, which is based on imaging characteristics for cystic renal masses, and detecting macroscopic fat within solid renal masses. Radiologically, cystic and solid renal masses are categorized and evaluated separately because of historical precedent, differences in appearance at imaging, and differences in biologic behavior. However, the World Health Organization classification of renal neoplasms does not support such separation. Further, the primary goal has been cancer diagnosis. Differentiating benign from malignant masses is important, but data show that many renal cancers, particularly when small, will not cause harm. Therefore, a critical goal of any unifying, single, imaging-based classification of kidney masses (ie, a Kidney Imaging Reporting and Data System) should be predicting the biologic behavior or aggressiveness of suspected kidney cancer. This system could inform the need for treatment or active surveillance and reduce prevalent overdiagnosis and overtreatment. This review describes the rationale for and challenges in creating such a system and the research needed for it to be developed.

Differences in mutations across tumour sizes in clear-cell renal cell carcinoma

OBJECTIVE

To assess the distribution of key mutations across tumour sizes in clear-cell renal cell carcinoma (ccRCC), and secondarily to examine the prognostic impact of aggressive mutations in smaller ccRCCs.

PATIENT AND METHODS

The distribution of mutations (VHL, PBRM1, SETD2, BAP1 and CDKN2A loss) across tumour sizes was assessed in 1039 ccRCCs treated with nephrectomy in cohorts obtained from the Tracking Cancer Evolution (TRACERx), The Cancer Genome Atlas (TCGA) and the Cancer Genomics of the Kidney (CAGEKID) projects. Logistic regression was used to model the presence of each mutation against size. In our secondary analysis, we assessed a subset of ccRCCs ≤7 cm for associations of key aggressive mutations (SETD2, BAP1, and CDKN2A loss) with metastasis, invasive disease and overall survival, while controlling for size. A subset of localised tumours ≤7 cm was also used to assess associations with recurrence after nephrectomy.

RESULTS

On logistic regression, each 1-cm increase in tumour size was associated with aggressive mutations, SETD2, BAP1, and CDKN2A loss, at odds ratios (ORs) of 1.09, 1.10 and 1.19 (P < 0.001), whereas no significant association was observed between tumour size and PBRM1 (OR 1.02; P = 0.23). VHL was mildly negatively associated with a 1-cm increase in size (OR 0.95; P = 0.01). Among tumours ≤7 cm, SETD2 and CDKN2A loss were associated with metastatic disease at ORs of 3.86 and 3.84 (P < 0.05) while controlling for tumour size. CDKN2A loss was associated with worse overall survival, with a hazard ratio (HR) of 2.19 (P = 0.03). Among localised tumours ≤7 cm, SETD2 was associated with worse recurrence-free survival (HR 2.00; P = 0.03).

CONCLUSION

Large and small ccRCCs are genomically different. Aggressive mutations, namely, SETD2, BAP1, and CDKN2A loss, are rarely observed in small ccRCCs and are observed more frequently in larger tumours. However, when present in tumours ≤7 cm, SETD2 mutations and CDKN2A loss were still independently associated with invasive disease, metastasis, worse survival, and recurrence after resection, after controlling for size.

Multiparameter computed tomography (CT) radiomics signature fusion-based model for the preoperative prediction of clear cell renal cell carcinoma nuclear grade: a multicenter development and external validation study

Background

The preoperative prediction of the pathological nuclear grade of clear cell renal cell carcinoma (CCRCC) is crucial for clinical decision making. However, radiomics features from one or two computed tomography (CT) phases are required to predict the CCRCC grade, which reduces the predictive performance and generalizability of this method. We aimed to develop and externally validate a multiparameter CT radiomics-based model for predicting the World Health Organization/International Society of Urological Pathology (WHO/ISUP) grade of CCRCC.

Methods

A total of 500 CCRCC patients at The First, Second, and Yongchuan Hospitals of Chongqing Medical University between January 2016 and May 2022 were retrospectively enrolled in this study. The patients were divided into the training set (n=268), internal testing set (n=115), and two external testing sets (testing set 1, n=62; testing set 2, n=55). Radiomics features were extracted from multi-phase CT images, and radiomics signatures (RSs) were created by least absolute shrinkage and selection operator (LASSO) regression. In addition, a clinical model was developed. A combined model was also established that integrated the RSs with the clinical factors, and was visualized via a nomogram. The performance of the established model was assessed using area under the curve (AUC) values, a calibration curve analysis, and a decision curve analysis (DCA).

Results

Among the four RSs and the clinical model, the RS-Triphasic had the best predictive performance with AUCs of 0.88 [95% confidence interval (CI): 0.85-0.91] and 0.84 (95% CI: 0.74-0.95) in the training and testing sets, respectively, and 0.82 (95% CI: 0.72-0.93) and 0.82 (95% CI: 0.71-0.93) in external testing sets 1 and 2. Integrating the RS-Triphasic, RS-corticomedullary phase (CMP), RS-nephrographic phase (NP), RS-non-contrast phase (NCP) with the clinical risk factors, a combined model was established with AUCs of 0.92 (95% CI: 0.89-0.94), 0.86 (95% CI: 0.76-0.95), 0.84 (95% CI: 0.73-0.95), and 0.82 (95% CI: 0.70-0.94) for the training, internal testing, and external testing sets 1 and 2, respectively. The DCA indicated that the nomogram had a greater overall net benefit than the clinical and radiomics models.

Conclusions

The multiparameter CT RS fusion-based model had high accuracy in differentiating between high- and low-grade CCRCC preoperatively. Thus, it has great potential as a useful tool for personalized treatment planning and clinical decision making for CCRCC patients.

The evolving management of small renal masses

Small renal masses (SRMs) are a heterogeneous group of tumours with varying metastatic potential. The increasing use and improving quality of abdominal imaging have led to increasingly early diagnosis of incidental SRMs that are asymptomatic and organ confined. Despite improvements in imaging and the growing use of renal mass biopsy, diagnosis of malignancy before treatment remains challenging. Management of SRMs has shifted away from radical nephrectomy, with active surveillance and nephron-sparing surgery taking over as the primary modalities of treatment. The optimal treatment strategy for SRMs continues to evolve as factors affecting short-term and long-term outcomes in this patient cohort are elucidated through studies from prospective data registries. Evidence from rapidly evolving research in biomarkers, imaging modalities, and machine learning shows promise in improving understanding of the biology and management of this patient cohort.

Histopathology Based AI Model Predicts Anti-Angiogenic Therapy Response in Renal Cancer Clinical Trial

Background

Predictive biomarkers of treatment response are lacking for metastatic clearcell renal cell carcinoma (ccRCC), a tumor type that is treated with angiogenesis inhibitors, immune checkpoint inhibitors, mTOR inhibitors and a HIF2 inhibitor. The Angioscore, an RNA-based quantification of angiogenesis, is arguably the best candidate to predict anti-angiogenic (AA) response. However, the clinical adoption of transcriptomic assays faces several challenges including standardization, time delay, and high cost. Further, ccRCC tumors are highly heterogenous, and sampling multiple areas for sequencing is impractical.

Approach

Here we present a novel deep learning (DL) approach to predict the Angioscore from ubiquitous histopathology slides. In order to overcome the lack of interpretability, one of the biggest limitations of typical DL models, our model produces a visual vascular network which is the basis of the model's prediction. To test its reliability, we applied this model to multiple cohorts including a clinical trial dataset.

Results

Our model accurately predicts the RNA-based Angioscore on multiple independent cohorts (spearman correlations of 0.77 and 0.73). Further, the predictions help unravel meaningful biology such as association of angiogenesis with grade, stage, and driver mutation status. Finally, we find our model is able to predict response to AA therapy, in both a real-world cohort and the IMmotion150 clinical trial. The predictive power of our model vastly exceeds that of CD31, a marker of vasculature, and nearly rivals the performance (c-index 0.66 vs 0.67) of the ground truth RNA-based Angioscore at a fraction of the cost.

Conclusion

By providing a robust yet interpretable prediction of the Angioscore from histopathology slides alone, our approach offers insights into angiogenesis biology and AA treatment response.

Molecular underpinnings of dedifferentiation and aggressiveness in chromophobe renal cell carcinoma

Sarcomatoid dedifferentiation is common to multiple renal cell carcinoma (RCC) subtypes, including chromophobe RCC (ChRCC), and is associated with increased aggressiveness, resistance to targeted therapies, and heightened sensitivity to immunotherapy. To study ChRCC dedifferentiation, we performed multiregion integrated paired pathological and genomic analyses. Interestingly, ChRCC dedifferentiates not only into sarcomatoid but also into anaplastic and glandular subtypes, which are similarly associated with increased aggressiveness and metastases. Dedifferentiated ChRCC shows loss of epithelial markers, convergent gene expression, and whole genome duplication from a hypodiploid state characteristic of classic ChRCC. We identified an intermediate state with atypia and increased mitosis but preserved epithelial markers. Our data suggest that dedifferentiation is initiated by hemizygous mutation of TP53, which can be observed in differentiated areas, as well as mutation of PTEN. Notably, these mutations become homozygous with duplication of preexisting monosomes (i.e., chromosomes 17 and 10), which characterizes the transition to dedifferentiated ChRCC. Serving as potential biomarkers, dedifferentiated areas become accentuated by mTORC1 activation (phospho-S6) and p53 stabilization. Notably, dedifferentiated ChRCC share gene enrichment and pathway activation features with other sarcomatoid RCC, suggesting convergent evolutionary trajectories. This study expands our understanding of aggressive ChRCC, provides insight into molecular mechanisms of tumor progression, and informs pathologic classification and diagnostics.

Renal Mass Imaging with MRI Clear Cell Likelihood Score: A User's Guide

Small solid renal masses (SRMs) are frequently detected at imaging. Nearly 20% are benign, making careful evaluation with MRI an important consideration before deciding on management. Clear cell renal cell carcinoma (ccRCC) is the most common renal cell carcinoma subtype with potentially aggressive behavior. Thus, confident identification of ccRCC imaging features is a critical task for the radiologist. Imaging features distinguishing ccRCC from other benign and malignant renal masses are based on major features (T2 signal intensity, corticomedullary phase enhancement, and the presence of microscopic fat) and ancillary features (segmental enhancement inversion, arterial-to-delayed enhancement ratio, and diffusion restriction). The clear cell likelihood score (ccLS) system was recently devised to provide a standardized framework for categorizing SRMs, offering a Likert score of the likelihood of ccRCC ranging from 1 (very unlikely) to 5 (very likely). Alternative diagnoses based on imaging appearance are also suggested by the algorithm. Furthermore, the ccLS system aims to stratify which patients may or may not benefit from biopsy. The authors use case examples to guide the reader through the evaluation of major and ancillary MRI features of the ccLS algorithm for assigning a likelihood score to an SRM. The authors also discuss patient selection, imaging parameters, pitfalls, and areas for future development. The goal is for radiologists to be better equipped to guide management and improve shared decision making between the patient and treating physician. © RSNA, 2023 Quiz questions for this article are available in the supplemental material. See the invited commentary by Pedrosa in this issue.

Small Renal Masses: The Evolving Histologic, Imaging, and Genomic Landscapes

According to the American Cancer Society, it is currently estimated that approximately 81,800 new cases of kidney cancer will be diagnosed in the United States in 2023 [...].