Phenotyping Hepatic Immune-Related Adverse Events in the Setting of Immune Checkpoint Inhibitor Therapy.

PURPOSE: We present and validate a rule-based algorithm for the detection of moderate to severe liver-related immune-related adverse events (irAEs) in a real-world patient cohort. The algorithm can be applied to studies of irAEs in large data sets. METHODS: We developed a set of criteria to define hepatic irAEs. The criteria include: the temporality of elevated laboratory measurements in the first 2-14 weeks of immune checkpoint inhibitor (ICI) treatment, steroid intervention within 2 weeks of the onset of elevated laboratory measurements, and intervention with a duration of at least 2 weeks. These criteria are based on the kinetics of patients who experienced moderate to severe hepatotoxicity (Common Terminology Criteria for Adverse Events grades 2-4). We applied these criteria to a retrospective cohort of 682 patients diagnosed with hepatocellular carcinoma and treated with ICI. All patients were required to have baseline laboratory measurements before and after the initiation of ICI. RESULTS: A set of 63 equally sampled patients were reviewed by two blinded, clinical adjudicators. Disagreements were reviewed and consensus was taken to be the ground truth. Of these, 25 patients with irAEs were identified, 16 were determined to be hepatic irAEs, 36 patients were nonadverse events, and two patients were of indeterminant status. Reviewers agreed in 44 of 63 patients, including 19 patients with irAEs (0.70 concordance, Fleiss' kappa: 0.43). By comparison, the algorithm achieved a sensitivity and specificity of identifying hepatic irAEs of 0.63 and 0.81, respectively, with a test efficiency (percent correctly classified) of 0.78 and outcome-weighted F1 score of 0.74. CONCLUSION: The algorithm achieves greater concordance with the ground truth than either individual clinical adjudicator for the detection of irAEs.

Nondestructive, quantitative viability analysis of 3D tissue cultures using machine learning image segmentation.

Ascertaining the collective viability of cells in different cell culture conditions has typically relied on averaging colorimetric indicators and is often reported out in simple binary readouts. Recent research has combined viability assessment techniques with image-based deep-learning models to automate the characterization of cellular properties. However, further development of viability measurements to assess the continuity of possible cellular states and responses to perturbation across cell culture conditions is needed. In this work, we demonstrate an image processing algorithm for quantifying features associated with cellular viability in 3D cultures without the need for assay-based indicators. We show that our algorithm performs similarly to a pair of human experts in whole-well images over a range of days and culture matrix compositions. To demonstrate potential utility, we perform a longitudinal study investigating the impact of a known therapeutic on pancreatic cancer spheroids. Using images taken with a high content imaging system, the algorithm successfully tracks viability at the individual spheroid and whole-well level. The method we propose reduces analysis time by 97% in comparison with the experts. Because the method is independent of the microscope or imaging system used, this approach lays the foundation for accelerating progress in and for improving the robustness and reproducibility of 3D culture analysis across biological and clinical research.

Proteogenomic analysis of enriched HGSOC tumor epithelium identifies prognostic signatures and therapeutic vulnerabilities.

We performed a deep proteogenomic analysis of bulk tumor and laser microdissection enriched tumor cell populations from high-grade serous ovarian cancer (HGSOC) tissue specimens spanning a broad spectrum of purity. We identified patients with longer progression-free survival had increased immune-related signatures and validated proteins correlating with tumor-infiltrating lymphocytes in 65 tumors from an independent cohort of HGSOC patients, as well as with overall survival in an additional 126 HGSOC patient cohort. We identified that homologous recombination deficient (HRD) tumors are enriched in pathways associated with metabolism and oxidative phosphorylation that we validated in independent patient cohorts. We further identified that polycomb complex protein BMI-1 is elevated in HR proficient (HRP) tumors, that elevated BMI-1 correlates with poor overall survival in HRP but not HRD HGSOC patients, and that HRP HGSOC cells are uniquely sensitive to BMI-1 inhibition.

Detecting Disruption of HER2 Membrane Protein Organization in Cell Membranes with Nanoscale Precision.

The spatiotemporal organization of proteins within the cell membrane can affect numerous biological functions, including cell signaling, communication, and transportation. Deviations from normal spatial arrangements have been observed in various diseases, and a better understanding of this process is a key stepping stone to advancing development of clinical interventions. However, given the nanometer length scales involved, detecting these subtle changes has primarily relied on complex super-resolution and single-molecule imaging methods. In this work, we demonstrate an alternative fluorescent imaging strategy for detecting protein organization based on a material that exhibits a unique photophysical behavior known as aggregation-induced emission (AIE). Organic AIE molecules have an increase in emission signal when they are in close proximity, and the molecular motion is restricted. This property simultaneously addresses the high background noise and low detection signal that limit conventional widefield fluorescent imaging. To demonstrate the potential of this approach, the fluorescent molecule sensor is conjugated to a human epidermal growth factor receptor 2 (HER2)-specific antibody and used to investigate the spatiotemporal behavior of HER2 clustering in the membrane of HER2-overexpressing breast cancer cells. Notably, the disruption of HER2 clusters in response to an FDA-approved monoclonal antibody therapeutic (Trastuzumab) is successfully detected using a simple widefield fluorescent microscope. While the sensor demonstrated here is optimized for sensing HER2 clustering, it is an easily adaptable platform. Moreover, given the compatibility with widefield imaging, the system has the potential to be used with high-throughput imaging techniques, accelerating investigations into membrane protein spatiotemporal organization.

KRAS Sequence Variation as Prognostic Marker in Patients With Young- vs Late-Onset Colorectal Cancer.

Importance: The understanding of the association between KRAS sequence variation status and clinical outcomes in colorectal cancer (CRC) has evolved over time. Objective: To characterize the association of age at onset, tumor sidedness, and KRAS sequence variation with survival among patients diagnosed with CRC. Design, Setting, and Participants: This cross-sectional study used data extracted from the Surveillance, Epidemiology, and End Results database. Patients diagnosed with adenocarcinoma of the colon or rectum from 2010 through 2015 were included and were classified as having young-onset (YO) cancer if diagnosed between ages 20 to 49 years and late-onset (LO) cancer if diagnosed at age 50 years or older. Data were analyzed from April 2021 through August 2023. Main Outcomes and Measures: CRC cause-specific survival (CSS) was summarized using Fine and Gray cumulative incidence and Kaplan-Meier curves. Estimation of subdistribution hazard ratios (sHRs) for the association of KRAS status, age at onset, and tumor location with CRC CSS was conducted using the Fine and Gray competing risk model. Cox proportional hazards regression was used to estimate and compare HRs. Results: Among 21 661 patients with KRAS sequence variation status (mean [SD] age at diagnosis, 62.50 [13.78] years; 9784 females [45.2%]), 3842 patients had YO CRC, including 1546 patients with KRAS variants, and 17 819 patients had LO CRC, including 7311 patients with KRAS variants. There was a significant difference in median CSS time between patients with variant vs wild-type KRAS (YO: 3.0 years [95% CI, 2.8-3.3 years] vs 3.5 years [95% CI, 3.3-3.9 years]; P = .02; LO: 2.5 years [95% CI, 2.4-2.7 years] vs 3.4 years [95% CI, 3.3-3.6 years]; P < .001). Tumors with variant compared with wild-type KRAS were associated with higher risk of CRC-related death (YO: sHR, 1.09 [95% CI, 1.01-1.18]; P = .03; LO: sHR, 1.06 [95% CI, 1.02-1.09]; P = .002). Among patients with YO cancer, mortality hazards increased by location, from right (sHR, 1.02 [95% CI, 0.88-1.17) to left (sHR, 1.15 [95% CI, 1.02-1.29) and rectum (sHR, 1.16 [95% CI, 0.99-1.36), but no trend by tumor location was seen for LO cancer. Conclusions and Relevance: In this study of patients diagnosed with CRC, KRAS sequence variation was associated with increased mortality among patients with YO and LO tumors. In YO cancer, variant KRAS-associated mortality risk was higher in distal tumors than proximal tumors.

Trends in Epidemiology of Esophageal Cancer in the US, 1975-2018.

Importance: Esophageal cancer (EC) is the 7th most common cancer worldwide and 14th in the US. More data are needed to study the changing incidence patterns of its 2 primary histologic subtypes, squamous cell carcinoma of the esophagus (SCE) and adenocarcinoma of the esophagus (ACE). Objective: To examine temporal trends in incidence rates of EC, ACE, and SCE from 1975 through 2018. Design, Setting, and Participants: In this population-based cross-sectional study, data were derived from 9 Surveillance, Epidemiology, and End Results (SEER) registries from January 1975 through December 2018 and from all 21 registries for January 2000 through December 2018 for patients with a diagnosis of EC from 1975 through 2018 (International Classification of Disease-Oncology, Third Edition codes). Age-adjusted incidence rates (AAIRs) of EC, ACE, and SCE were calculated. The timing and magnitude of the annual percentage change (APC) in incidence were examined using Joinpoint regression analyses. Data analysis was started in 2021 and updated and completed in 2023. Main Outcome and Measures: The APC for age-adjusted EC incidence rates as stratified by histology, anatomical location, stage, sex, age, race and ethnicity, and geographic region. Results: A total of 47 648 patients with a diagnosis of EC were retained for analysis. These included 22 419 (47.1%) with a diagnosis of SCE, 22 217 (46.6%) with ACE, and 3012 (6.3%) with other subtypes. The AAIR for EC changed from 4.14 per 100 000 population in 1975 to 4.18 in 2018, AAIRs of SCE declined from 3.06 in 1975 to 1.15 in 2018 as well as for ACE, and AAIRs increased from 0.42 in 1975 to 2.78 in 2018. From 1975 through 2004, EC incidence significantly increased (APC, 0.53; 95% CI, 0.4 to 0.7) but significantly decreased (APC, -1.03; 95% CI, -1.3 to -0.7) from then until 2018. The APC of SCE significantly continued to decline (-2.80, 95% CI, -3.0 to -2.6), and ACE increased from 2000 to 2006 (APC, 2.51; 95% CI, 1.0 to 4.0) but has since stabilized from 2006 to 2018. Conclusions and Relevance: The results of this cross-sectional study suggest that the incidence of EC modestly declined since 2004 and that the incidence of SCE continued to decline while the incidence rate of ACE plateaued for more than a decade. Understanding factors associated with plateaued rates of ACE may help inform public health interventions.

Racial/Ethnic Disparities Impact the Real-World Effectiveness of a Multicomponent Maternal Smoking Cessation Program: Findings from the CTTP Cohort.

INTRODUCTION: Smoking during pregnancy adversely affects perinatal outcomes for both women and infants. We conducted a retrospective cohort study of the state-funded Comprehensive Tobacco Treatment Program (CTTP) - the largest maternal tobacco cessation program in San Bernardino County, California - to determine the real-world program effectiveness and to identify variables that can potentially improve effectiveness. METHODS: During 2012-2019, women who smoked during pregnancy were enrolled in CTTP's multicomponent behavioral smoking cessation program that implemented components of known efficacy (i.e., incentives, biomarker testing, feedback, and motivational interviewing). RESULTS: We found that 40.1% achieved prolonged abstinence by achieving weekly, cotinine-verified, 7-day abstinence during 6 to 8 weeks of enrollment. Using intention-to-treat analyses, we computed that the self-reported point prevalence abstinence rate (PPA) at the six-month telephone follow-up was 36.7%. Cohort members achieving prolonged abstinence during the CTTP were five times more likely to achieve PPA six months after CTTP. Several non-Hispanic ethnicities (Black, Native American, White, or More than one ethnicity) in the cohort were two-fold less likely (relative to Hispanics) to achieve prolonged abstinence during CTTP or PPA at six months after CTTP. This disparity was further investigated in mediation analysis. Variables such as quitting during the first trimester and smoking fewer cigarettes at enrollment were also associated with achieving PPA at six months. DISCUSSION: Racial/ethnic health disparities that have long been linked to a higher rate of maternal smoking persist even when the pregnant smoker enrolls in a smoking cessation program.

Chemotherapy Treatment Costs and Clinical Outcomes of Colon Cancer in the U.S. Military Health System's Direct and Private Sector Care Settings.

INTRODUCTION: Identifying low-value cancer care may be an important step in containing costs associated with treatment. Low-value care occurs when the medical services, tests, or treatments rendered do not result in clinical benefit. These may be impacted by care setting and patients' access to care and health insurance. We aimed to study chemotherapy treatment and the cost paid by the Department of Defense (DoD) for treatment in relation to clinical outcomes among patients with colon cancer treated within the U.S. Military Health System's direct and private sector care settings to better understand the value of cancer care. MATERIALS AND METHODS: A cohort of patients aged 18 to 64 years with primary colon cancer diagnosed between January 1, 1999, and December 31, 2014, were identified in the Military Cancer Epidemiology database. Multivariable time-dependent Cox proportional hazards regression models were used to assess the relationship between chemotherapy treatment and the cost paid by the DoD (in quartiles, Q) and the outcomes of cancer progression, cancer recurrence, and all-cause death modeled as adjusted hazard ratios (aHRs) and 95% confidence intervals (95% CIs). The Military Cancer Epidemiology data were approved for research by the Uniformed Services University of the Health Sciences' Institutional Review Board. RESULTS: The study included 673 patients using direct care and 431 patients using private sector care. The median per patient chemotherapy costs in direct care ($111,202) were lower than in private sector care ($350,283). In direct care, higher chemotherapy costs were associated with an increased risk of any outcome but not with all-cause death. In private sector care, higher chemotherapy costs were associated with a higher risk of any outcome and with all-cause death (aHR, 2.67; 95% CI, 1.20-5.92 for Q4 vs. Q1). CONCLUSIONS: The findings in the private sector may indicate low-value care in terms of the cost paid by the DoD for chemotherapy treatment and achieving desirable survival outcomes for patients with colon cancer in civilian health care. Comprehensive evaluations of value-based care among patients treated for other tumor types may be warranted.

An Inflection Point in Cancer Protein Biomarkers: What was and What's Next.

Biomarkers remain the highest value proposition in cancer medicine today-especially protein biomarkers. Despite decades of evolving regulatory frameworks to facilitate the review of emerging technologies, biomarkers have been mostly about promise with very little to show for improvements in human health. Cancer is an emergent property of a complex system, and deconvoluting the integrative and dynamic nature of the overall system through biomarkers is a daunting proposition. The last 2 decades have seen an explosion of multiomics profiling and a range of advanced technologies for precision medicine, including the emergence of liquid biopsy, exciting advances in single-cell analysis, artificial intelligence (machine and deep learning) for data analysis, and many other advanced technologies that promise to transform biomarker discovery. Combining multiple omics modalities to acquire a more comprehensive landscape of the disease state, we are increasingly developing biomarkers to support therapy selection and patient monitoring. Furthering precision medicine, especially in oncology, necessitates moving away from the lens of reductionist thinking toward viewing and understanding that complex diseases are, in fact, complex adaptive systems. As such, we believe it is necessary to redefine biomarkers as representations of biological system states at different hierarchical levels of biological order. This definition could include traditional molecular, histologic, radiographic, or physiological characteristics, as well as emerging classes of digital markers and complex algorithms. To succeed in the future, we must move past purely observational individual studies and instead start building a mechanistic framework to enable integrative analysis of new studies within the context of prior studies. Identifying information in complex systems and applying theoretical constructs, such as information theory, to study cancer as a disease of dysregulated communication could prove to be "game changing" for the clinical outcome of cancer patients.

Diagnosis and Management of Adult Malignancy-Associated Hemophagocytic Lymphohistiocytosis.

Hemophagocytic lymphohistiocytosis (HLH) is a syndrome of severe, dysregulated inflammation driven by the inability of T cells to clear an antigenic target. When associated with malignancy (mHLH), the HLH syndrome is typically associated with extremely poor survival. Here, we review the diagnosis of secondary HLH (sHLH) syndromes in adults, with emphasis on the appropriate workup and treatment of mHLH. At present, the management of HLH in adults, including most forms of mHLH, is based on the use of corticosteroids and etoposide following the HLH-94 regimen. In some cases, this therapeutic approach may be cohesively incorporated into malignancy-directed therapy, while in other cases, the decision about whether to treat HLH prior to initiating other therapies may be more complicated. Recent studies exploring the efficacy of other agents in HLH, in particular ruxolitinib, offer hope for better outcomes in the management of mHLH. Considerations for the management of lymphoma-associated mHLH, as well as other forms of mHLH and immunotherapy treatment-related HLH, are discussed.

Contrived Materials and a Data Set for the Evaluation of Liquid Biopsy Tests: A Blood Profiling Atlas in Cancer (BLOODPAC) Community Study.

The Blood Profiling Atlas in Cancer (BLOODPAC) Consortium is a collaborative effort involving stakeholders from the public, industry, academia, and regulatory agencies focused on developing shared best practices on liquid biopsy. This report describes the results from the JFDI (Just Freaking Do It) study, a BLOODPAC initiative to develop standards on the use of contrived materials mimicking cell-free circulating tumor DNA, to comparatively evaluate clinical laboratory testing procedures. Nine independent laboratories tested the concordance, sensitivity, and specificity of commercially available contrived materials with known variant-allele frequencies (VAFs) ranging from 0.1% to 5.0%. Each participating laboratory utilized its own proprietary evaluation procedures. The results demonstrated high levels of concordance and sensitivity at VAFs of >0.1%, but reduced concordance and sensitivity at a VAF of 0.1%; these findings were similar to those from previous studies, suggesting that commercially available contrived materials can support the evaluation of testing procedures across multiple technologies. Such materials may enable more objective comparisons of results on materials formulated in-house at each center in multicenter trials. A unique goal of the collaborative effort was to develop a data resource, the BLOODPAC Data Commons, now available to the liquid-biopsy community for further study. This resource can be used to support independent evaluations of results, data extension through data integration and new studies, and retrospective evaluation of data collection.

Analysis of BRCA2 Copy Number Loss and Genomic Instability in Circulating Tumor Cells from Patients with Metastatic Castration-resistant Prostate Cancer.

BACKGROUND: BRCA2 alterations predict for a response to poly-ADP-ribose polymerase inhibition in metastatic castration-resistant prostate cancer (mCRPC). However, detection is hindered by insufficient tumor tissue and low sensitivity of cell-free DNA for detecting copy number loss. OBJECTIVE: To evaluate the BRCA2 loss detection using single-cell, shallow whole-genome sequencing (sWGS) of circulating tumor cells (CTCs) in patients with mCRPC. DESIGN, SETTING, AND PARTICIPANTS: We analyzed CTC samples collected concurrently with tumor biopsies intended for clinical sequencing in patients with progressing mCRPC. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Differences in proportions were evaluated using the chi-square test. Correlations between assays were analyzed in linear regression models. Associations between alterations and genomic instability were assessed on the single-cell level using mixed-effect negative binomial models. RESULTS AND LIMITATIONS: We identified 138 patients with concurrent CTC and biopsy samples. CTC sWGS generated copy number profiles in a similar proportion of patients to biopsy samples (83% vs 78%, p = 0.23), but was more effective than bone biopsies (79% vs 50%; p = 0.009). CTC sWGS detected BRCA2 loss in more patients than tissue at the ≥1 (42% vs 16%; p < 0.001) and ≥2 (27% vs 16%; p = 0.028) CTC thresholds. The overall prevalence of BRCA2 loss was not increased in CTCs using sample-level composite z scores (p = 0.4), but was significantly increased compared with a lower-than-expected prevalence in bone samples (21% vs 3%, p = 0.014). Positive/negative predictive values for CTC BRCA2 loss were 89%/96% using the ≥1 CTC threshold and 67%/92% using the composite z score. CTC BRCA2 loss was associated with higher genomic instability in univariate (1.4-fold large-scale transition difference, 95% confidence interval [CI]: 1.2-1.6; p < 0.001) and multivariable analysis (1.4-fold difference, 95% CI: 1.2-1.6; p < 0.001). CONCLUSIONS: Copy number profiles can reliably be generated using CTC sWGS, which detected a majority of tissue-confirmed BRCA2 loss and "CTC-only" losses. BRCA2 losses were supported by increases in genomic instability. PATIENT SUMMARY: Current testing strategies have limitations in their ability to detect BRCA2 loss, a relatively common alteration in prostate cancer that is used to identify patients who may benefit from targeted therapy. In this paper, we evaluated whether we could detect BRCA2 loss in individual tumor cells isolated from patient blood samples and found this method to be suitable for further analysis.

Proteogenomic analysis of lung adenocarcinoma reveals tumor heterogeneity, survival determinants, and therapeutically relevant pathways.

We present a deep proteogenomic profiling study of 87 lung adenocarcinoma (LUAD) tumors from the United States, integrating whole-genome sequencing, transcriptome sequencing, proteomics and phosphoproteomics by mass spectrometry, and reverse-phase protein arrays. We identify three subtypes from somatic genome signature analysis, including a transition-high subtype enriched with never smokers, a transversion-high subtype enriched with current smokers, and a structurally altered subtype enriched with former smokers, TP53 alterations, and genome-wide structural alterations. We show that within-tumor correlations of RNA and protein expression associate with tumor purity and immune cell profiles. We detect and independently validate expression signatures of RNA and protein that predict patient survival. Additionally, among co-measured genes, we found that protein expression is more often associated with patient survival than RNA. Finally, integrative analysis characterizes three expression subtypes with divergent mutations, proteomic regulatory networks, and therapeutic vulnerabilities. This proteogenomic characterization provides a foundation for molecularly informed medicine in LUAD.

Agonizing over the Stimulatory Immune Checkpoint ICOS.

SUMMARY: Vopratelimab, an anti-ICOS (inducible costimulator of T cells) agonist, alone and in combination with nivolumab, possesses limited toxicity and modest clinical activity in a large phase I/II trial. This treatment induced ICOS expression of CD4+ T cells, which may enable biomarkers for patient selection. Nevertheless, T-cell agonists as cancer immunotherapies continue to be challenging. See related article by Yap et al., p. 3695.

Adaptation of Imaging Mass Cytometry to Explore the Single Cell Alloimmune Landscape of Liver Transplant Rejection.

Rejection continues to be an important cause of graft loss in solid organ transplantation, but deep exploration of intragraft alloimmunity has been limited by the scarcity of clinical biopsy specimens. Emerging single cell immunoprofiling technologies have shown promise in discerning mechanisms of autoimmunity and cancer immunobiology. Within these applications, Imaging Mass Cytometry (IMC) has been shown to enable highly multiplexed, single cell analysis of immune phenotypes within fixed tissue specimens. In this study, an IMC panel of 10 validated markers was developed to explore the feasibility of IMC in characterizing the immune landscape of chronic rejection (CR) in clinical tissue samples obtained from liver transplant recipients. IMC staining was highly specific and comparable to traditional immunohistochemistry. A single cell segmentation analysis pipeline was developed that enabled detailed visualization and quantification of 109,245 discrete cells, including 30,646 immune cells. Dimensionality reduction identified 11 unique immune subpopulations in CR specimens. Most immune subpopulations were increased and spatially related in CR, including two populations of CD45+/CD3+/CD8+ cytotoxic T-cells and a discrete CD68+ macrophage population, which were not observed in liver with no rejection (NR). Modeling via principal component analysis and logistic regression revealed that single cell data can be utilized to construct statistical models with high consistency (Wilcoxon Rank Sum test, p=0.000036). This study highlights the power of IMC to investigate the alloimmune microenvironment at a single cell resolution during clinical rejection episodes. Further validation of IMC has the potential to detect new biomarkers, identify therapeutic targets, and generate patient-specific predictive models of clinical outcomes in solid organ transplantation.

Translating "Big Data" in Oncology for Clinical Benefit: Progress or Paralysis.

The molecular characterization of cancer through genomics, data from multiomics technologies, molecular-driven clinical trials, and internet-enabled devices capturing patient context and real-world data are creating an unprecedented big data revolution across the cancer research-care continuum. While big data has translated to benefit for some patients, it has also created new problems. Our intent in this brief communication is to explore some examples of progress and key challenges that remain. The problems are not intractable, but success will require rethinking and rebuilding an information and evidence-based learning system that moves beyond paralysis to shape a better future for patients with cancer.

Circulating tumor cell heterogeneity in neuroendocrine prostate cancer by single cell copy number analysis.

Neuroendocrine prostate cancer is an aggressive variant of prostate cancer that may arise de novo or develop from pre-existing prostate adenocarcinoma as a mechanism of treatment resistance. The combined loss of tumor suppressors RB1, TP53, and PTEN are frequent in NEPC but also present in a subset of prostate adenocarcinomas. Most clinical and preclinical studies support a trans-differentiation process, whereby NEPC arises clonally from a prostate adenocarcinoma precursor during the course of treatment resistance. Here we highlight a case of NEPC with significant intra-patient heterogeneity observed across metastases. We further demonstrate how single-cell genomic analysis of circulating tumor cells combined with a phenotypic evaluation of cellular diversity can be considered as a window into tumor heterogeneity in patients with advanced prostate cancer.