Tumor response assessment in hepatocellular carcinoma treated with immunotherapy: imaging biomarkers for clinical decision-making.

OBJECTIVE: To compare the performance of 1D and 3D tumor response assessment for predicting median overall survival (mOS) in patients who underwent immunotherapy for hepatocellular carcinoma (HCC). METHODS: Patients with HCC who underwent immunotherapy between 2017 and 2023 and received multi-phasic contrast-enhanced MRIs pre- and post-treatment were included in this retrospective study. Tumor response was measured using 1D, RECIST 1.1, and mRECIST, and 3D, volumetric, and percentage quantitative EASL (vqEASL and %qEASL). Patients were grouped into disease control vs progression and responders vs non-responders. Kaplan-Meier curves analyzed with log-rank tests assessed the predictive value for mOS. Cox regression modeling evaluated the association of clinical baseline parameters with mOS. RESULTS: This study included 37 patients (mean age, 69.1 years [SD, 8.0]; 33 men). The mOS was 16.9 months. 3D vqEASL and %qEASL successfully stratified patients into disease control and progression (vqEASL: HR 0.21, CI: 0.55-0.08, p < 0.001; %qEASL: HR 0.18, CI: 0.83-0.04, p = 0.013), as well as responder and nonresponder (vqEASL: HR 0.25, CI: 0.08-0.74, p = 0.007; %qEASL: HR 0.17, CI: 0.04-0.72, p = 0.007) for predicting mOS. The 1D criteria, mRECIST stratified into disease control and progression only (HR 0.24, CI: 0.65-0.09, p = 0.002), and RECIST 1.1 showed no predictive value in either stratification. Multivariate Cox regression identified alpha-fetoprotein > 500 ng/mL as a predictor for poor mOS (p = 0.04). CONCLUSION: The 3D quantitative enhancement-based response assessment tool qEASL can predict overall survival in patients undergoing immunotherapy for HCC and could identify non-responders. CLINICAL RELEVANCE STATEMENT: Using 3D quantitative enhancement-based tumor response criteria (qEASL), radiologists' predictions of tumor response in patients undergoing immunotherapy for HCC can be further improved. KEY POINTS: MRI-based tumor response criteria predict immunotherapy survival benefits in HCC patients. 3D tumor response assessment methods surpass current evaluation criteria in predicting overall survival during HCC immunotherapy. Enhancement-based 3D tumor response criteria are robust prognosticators of survival for HCC patients on immunotherapy.

Safety and Efficacy of Atezolizumab/Bevacizumab in Patients with Hepatocellular Carcinoma and Impaired Liver Function: A Systematic Review and Meta-Analysis.

Background: Safety and outcome of atezolizumab/bevacizumab in Child-Pugh B patients with hepatocellular carcinoma (HCC) have not been completely characterized. Objectives: In this study, we aimed at addressing safety and efficacy of atezolizumab/bevacizumab in Child-Pugh B patients by reviewing the available data and analyzing them by meta-analysis. Methods: We compared the safety and efficacy of atezolizumab/becavizumab treatment in patients with unresectable HCC and various degrees of liver dysfunction. A total of 8 retrospective, non-randomized, cohort studies were included in this meta-analysis, for a total of 1,071 Child-Pugh A and 225 Child-Pugh B patients. The albumin-bilirubin (ALBI) grade was also used to assess liver function, when available. Results: Grade ≥3 adverse events were observed in 11.8% of Child-Pugh class A and 26.8% class B patients (p = 0.0001), with an odds ratio (OR) of 0.43 (confidence interval [CI] 0.21-0.90; p = 0.02). Progression-free survival (PFS) at both 6 months (4.90 ± 2.08 vs. 4.75 ± 2.08 months; p = 0.0004) and 12 months (8.83 ± 2.32 vs. 7.26 ± 2.33 months; p = 0.002) was lower in Child-Pugh class B patients. A trend toward a higher objective response rate (ORR) was observed in Child-Pugh class A patients (219/856, 25.6%) as compared to Child-Pugh class B patients (25/138, 18.1%; p = 0.070), while the probability of obtaining an ORR was significantly greater in Child-Pugh A patients (OR 1.79, CI 1.12-2.86; p = 0.02). Median overall survival (OS) was 16.8 ± 2.0 and 6.8 ± 3.2 months in Child-Pugh A and B patients, respectively (mean difference 9.06 months, CI 7.01-11.1, p < 0.0001). Lastly, OS was longer in patients with ALBI grades 1-2 than in those with grade 3 (8.3 ± 11.4 vs. 3.3 ± 5.0 months, p = 0.0008). Conclusions: Oncological efficacy of atezolizumab/bevacizumab is moderate in Child-Pugh class B patients, and the shorter PFS and OS associated with the greater likelihood of experiencing treatment-related adverse events observed in these patients suggest great caution and individualization of treatment, possibly with the support of the ALBI grade.

Artificial intelligence in liver cancer - new tools for research and patient management.

Liver cancer has high incidence and mortality globally. Artificial intelligence (AI) has advanced rapidly, influencing cancer care. AI systems are already approved for clinical use in some tumour types (for example, colorectal cancer screening). Crucially, research demonstrates that AI can analyse histopathology, radiology and natural language in liver cancer, and can replace manual tasks and access hidden information in routinely available clinical data. However, for liver cancer, few of these applications have translated into large-scale clinical trials or clinically approved products. Here, we advocate for the incorporation of AI in all stages of liver cancer management. We present a taxonomy of AI approaches in liver cancer, highlighting areas with academic and commercial potential, and outline a policy for AI-based liver cancer management, including interdisciplinary training of researchers, clinicians and patients. The potential of AI in liver cancer is immense, but effort is required to ensure that AI can fulfil expectations.

Automated graded prognostic assessment for patients with hepatocellular carcinoma using machine learning.

BACKGROUND: Accurate mortality risk quantification is crucial for the management of hepatocellular carcinoma (HCC); however, most scoring systems are subjective. PURPOSE: To develop and independently validate a machine learning mortality risk quantification method for HCC patients using standard-of-care clinical data and liver radiomics on baseline magnetic resonance imaging (MRI). METHODS: This retrospective study included all patients with multiphasic contrast-enhanced MRI at the time of diagnosis treated at our institution. Patients were censored at their last date of follow-up, end-of-observation, or liver transplantation date. The data were randomly sampled into independent cohorts, with 85% for development and 15% for independent validation. An automated liver segmentation framework was adopted for radiomic feature extraction. A random survival forest combined clinical and radiomic variables to predict overall survival (OS), and performance was evaluated using Harrell's C-index. RESULTS: A total of 555 treatment-naïve HCC patients (mean age, 63.8 years ± 8.9 [standard deviation]; 118 females) with MRI at the time of diagnosis were included, of which 287 (51.7%) died after a median time of 14.40 (interquartile range, 22.23) months, and had median followed up of 32.47 (interquartile range, 61.5) months. The developed risk prediction framework required 1.11 min on average and yielded C-indices of 0.8503 and 0.8234 in the development and independent validation cohorts, respectively, outperforming conventional clinical staging systems. Predicted risk scores were significantly associated with OS (p < .00001 in both cohorts). CONCLUSIONS: Machine learning reliably, rapidly, and reproducibly predicts mortality risk in patients with hepatocellular carcinoma from data routinely acquired in clinical practice. CLINICAL RELEVANCE STATEMENT: Precision mortality risk prediction using routinely available standard-of-care clinical data and automated MRI radiomic features could enable personalized follow-up strategies, guide management decisions, and improve clinical workflow efficiency in tumor boards. KEY POINTS: • Machine learning enables hepatocellular carcinoma mortality risk prediction using standard-of-care clinical data and automated radiomic features from multiphasic contrast-enhanced MRI. • Automated mortality risk prediction achieved state-of-the-art performances for mortality risk quantification and outperformed conventional clinical staging systems. • Patients were stratified into low, intermediate, and high-risk groups with significantly different survival times, generalizable to an independent evaluation cohort.

Automated MRI liver segmentation for anatomical segmentation, liver volumetry, and the extraction of radiomics.

OBJECTIVES: To develop and evaluate a deep convolutional neural network (DCNN) for automated liver segmentation, volumetry, and radiomic feature extraction on contrast-enhanced portal venous phase magnetic resonance imaging (MRI). MATERIALS AND METHODS: This retrospective study included hepatocellular carcinoma patients from an institutional database with portal venous MRI. After manual segmentation, the data was randomly split into independent training, validation, and internal testing sets. From a collaborating institution, de-identified scans were used for external testing. The public LiverHccSeg dataset was used for further external validation. A 3D DCNN was trained to automatically segment the liver. Segmentation accuracy was quantified by the Dice similarity coefficient (DSC) with respect to manual segmentation. A Mann-Whitney U test was used to compare the internal and external test sets. Agreement of volumetry and radiomic features was assessed using the intraclass correlation coefficient (ICC). RESULTS: In total, 470 patients met the inclusion criteria (63.9±8.2 years; 376 males) and 20 patients were used for external validation (41±12 years; 13 males). DSC segmentation accuracy of the DCNN was similarly high between the internal (0.97±0.01) and external (0.96±0.03) test sets (p=0.28) and demonstrated robust segmentation performance on public testing (0.93±0.03). Agreement of liver volumetry was satisfactory in the internal (ICC, 0.99), external (ICC, 0.97), and public (ICC, 0.85) test sets. Radiomic features demonstrated excellent agreement in the internal (mean ICC, 0.98±0.04), external (mean ICC, 0.94±0.10), and public (mean ICC, 0.91±0.09) datasets. CONCLUSION: Automated liver segmentation yields robust and generalizable segmentation performance on MRI data and can be used for volumetry and radiomic feature extraction. CLINICAL RELEVANCE STATEMENT: Liver volumetry, anatomic localization, and extraction of quantitative imaging biomarkers require accurate segmentation, but manual segmentation is time-consuming. A deep convolutional neural network demonstrates fast and accurate segmentation performance on T1-weighted portal venous MRI. KEY POINTS: • This deep convolutional neural network yields robust and generalizable liver segmentation performance on internal, external, and public testing data. • Automated liver volumetry demonstrated excellent agreement with manual volumetry. • Automated liver segmentations can be used for robust and reproducible radiomic feature extraction.

Community-Level Factors Associated with Hepatocellular Carcinoma Incidence and Mortality: An Observational Registry Study.

BACKGROUND: Hepatocellular carcinoma (HCC) incidence and outcomes vary across populations in the United States, but few studies evaluate local drivers of observed disparities. We measured HCC incidence at the community level and assessed community-level HCC risk factors with the goal of informing resource allocation to improve early case detection, which is associated with improved outcomes. METHODS: Clinical and demographic data including census tract of residence for all adults diagnosed with HCC in the Connecticut Tumor Registry between 2008 and 2019 were combined with publicly available U.S. Census and Centers for Disease Control and Prevention (CDC) data at the ZIP Code tabulation area (ZCTA) level. The average annual incidence of HCC was calculated for each ZCTA and associations between community-level characteristics, HCC incidence, stage at diagnosis, and survival were evaluated. RESULTS: Average annual HCC incidence during the study period was 8.9/100,000 adults and varied from 0 to 97.7 per 100,000 adults by ZCTA. At the community level, lower rates of high school graduation, higher rates of poverty, and rural community type were associated with higher HCC incidence. Persons with HCC living in the highest incidence ZCTAs were diagnosed at a younger age and were less likely to be alive at 1, 2, and 5 years after diagnosis. CONCLUSIONS: Community-level socioeconomic factors are strongly associated with HCC incidence and survival in Connecticut. IMPACT: This reproducible geo-localization approach using cancer registry, Census, and CDC data can be used to identify communities most likely to benefit from health system investments to reduce disparities in HCC.

Spectral top-down recovery of latent tree models.

Modeling the distribution of high-dimensional data by a latent tree graphical model is a prevalent approach in multiple scientific domains. A common task is to infer the underlying tree structure, given only observations of its terminal nodes. Many algorithms for tree recovery are computationally intensive, which limits their applicability to trees of moderate size. For large trees, a common approach, termed divide-and-conquer, is to recover the tree structure in two steps. First, separately recover the structure of multiple, possibly random subsets of the terminal nodes. Second, merge the resulting subtrees to form a full tree. Here, we develop spectral top-down recovery (STDR), a deterministic divide-and-conquer approach to infer large latent tree models. Unlike previous methods, STDR partitions the terminal nodes in a non random way, based on the Fiedler vector of a suitable Laplacian matrix related to the observed nodes. We prove that under certain conditions, this partitioning is consistent with the tree structure. This, in turn, leads to a significantly simpler merging procedure of the small subtrees. We prove that STDR is statistically consistent and bound the number of samples required to accurately recover the tree with high probability. Using simulated data from several common tree models in phylogenetics, we demonstrate that STDR has a significant advantage in terms of runtime, with improved or similar accuracy.

Advancing diagnosis and management of liver disease in adults through exome sequencing.

BACKGROUND: Whole-exome sequencing (WES) is an effective tool for diagnosis in patients who remain undiagnosed despite a comprehensive clinical work-up. While WES is being used increasingly in pediatrics and oncology, it remains underutilized in non-oncological adult medicine, including in patients with liver disease, in part based on the faulty premise that adults are unlikely to harbor rare genetic variants with large effect size. Here, we aim to assess the burden of rare genetic variants underlying liver disease in adults at two major tertiary referral academic medical centers. METHODS: WES analysis paired with comprehensive clinical evaluation was performed in fifty-two adult patients with liver disease of unknown etiology evaluated at two US tertiary academic health care centers. FINDINGS: Exome analysis uncovered a definitive or presumed diagnosis in 33% of patients (17/52) providing insight into their disease pathogenesis, with most of these patients (12/17) not having a known family history of liver disease. Our data shows that over two-thirds of undiagnosed liver disease patients attaining a genetic diagnosis were being evaluated for cholestasis or hepatic steatosis of unknown etiology. INTERPRETATION: This study reveals an underappreciated incidence and spectrum of genetic diseases presenting in adulthood and underscores the clinical value of incorporating exome sequencing in the evaluation and management of adults with liver disease of unknown etiology. FUNDING: S.V. is supported by the NIH/NIDDK (K08 DK113109 and R01 DK131033-01A1) and the Doris Duke Charitable Foundation Grant #2019081. This work was supported in part by NIH-funded Yale Liver Center, P30 DK34989.

Adherence to Tumor Board Recommendations in the Treatment of Patients with Hepatocellular Carcinoma.

Background: Hepatocellular carcinoma (HCC) is a heterogeneous disease that typically arises in the setting of chronic liver disease, making treatment selection complex. Multidisciplinary liver tumor boards (MDLTB) have been shown to improve outcomes in patients with HCC. However, in many cases, patients evaluated by MDLTBs ultimately do not receive the board's recommended treatment. Purpose: This study aims to assess adherence to MDLTB recommendations for the treatment of HCC, the reasons for non-adherence, and the survival of Barcelona Clinic Liver Cancer (BCLC) Stage A patients treated with curative treatment compared to palliative locoregional therapy. Patients and Methods: A single-site, retrospective cohort study was conducted of all patients with treatment-naïve HCC who were evaluated by an MDLTB at a tertiary care center in Connecticut between 2013 and 2016, of which 225 patients met inclusion criteria. Investigators conducted a chart review and recorded adherence to the MDLTB's recommendations, and in cases of discordance, evaluated and recorded the underlying cause; investigators assessed MDLTB recommendations' compliance with BCLC guidelines. Survival data was accrued through February 1st of 2022 and analyzed via Kaplan-Meier analysis and multivariate Cox regression. Results: Treatment adherent to MDLTB recommendations occurred in 85.3% of patients (n=192). The majority of non-adherence occurred in the management of BCLC Stage A disease. In cases where adherence was possible but the recommendation was not followed, most discrepancies were whether to treat with curative or palliative intent (20/24), with almost all discrepancies occurring in patients (19/20) with BCLC Stage A disease. For patients with Stage A unifocal HCC, those who received curative therapy lived significantly longer than patients who received palliative locoregional therapy (5.55 years vs 4.26 years, p=0.037). Conclusion: Most forms of non-adherence to MDLTB recommendations were unavoidable; however, treatment discordance in the management of patients with BCLC Stage A unifocal disease may present an opportunity for clinically significant quality improvement.

Immune checkpoint inhibitors in malignancies after liver transplantation: A systematic review and pooled analysis.

BACKGROUND AND AIMS: Treatment of de novo malignancies and recurrent hepatocellular carcinoma with immune checkpoint inhibitors (ICI) in liver transplant recipients (LT) is an attractive strategy that is infrequently pursued because of the lack of strong evidence regarding their safety and efficacy. In this systematic review with pooled analysis, we aimed to assess safety and efficacy of ICI therapy following LT. METHODS: We performed a systematic search of case reports and series published until January 2022. We included 31 publications reporting a total of 52 patients treated with ICIs after LT and assessed in a pooled analysis the risk of graft rejection and the outcome of ICI therapy. RESULTS: Acute graft rejection occurred in 15 patients (28.8%) and 7 patients (13.4% of the total cohort) died because of graft loss. Rejection was associated with shorter overall survival (OS) (17.2 months, confidence interval [CI] 12.1-22.2 vs. 3.5 months, CI 1.6-5.4, p < 0.001). Disease control rate was 44.2% (n = 23), and in these patients, OS was longer than in non-responders (26.4 months, CI 20.8-32.0 vs. 3.4 months, CI 2.1-4.7, p < 0.001). CONCLUSIONS: Observational, off-label experience suggests that treatment with ICI for advanced malignancies in LT recipients might not be discarded a priori. This notwithstanding, ICI treatment in these patients is associated with a substantial risk of graft rejection and mortality. Prospective studies are needed to provide adequate safety and efficacy figures of ICI treatment in this fragile population.

Holistic management of hepatocellular carcinoma: The hepatologist's comprehensive playbook.

Hepatocellular carcinoma (HCC) is a common complication in patients with chronic liver disease and leads to significant morbidity and mortality. Liver disease and liver cancer are preventable by mitigating and managing common risk factors, including chronic hepatitis B and C infection, alcohol use, diabetes, obesity and other components of the metabolic syndrome. The management of patients with HCC requires treatment of the malignancy and adequate control of the underlying liver disease, as preserving liver function is critical for successful cancer treatment and may have a relevant prognostic role independent of HCC management. Hepatologists are the ideal providers to guide the care of patients with HCC as they are trained to identify patients at risk, apply appropriate surveillance strategies, assess and improve residual liver function, evaluate candidacy for transplant, provide longitudinal care to optimize and preserve liver function during and after HCC treatment, survey for cancer recurrence and manage its risk factors, and prevent and treat decompensating events. We highlight the need for a team-based holistic approach to the patient with liver disease and HCC and identify necessary gaps in current care and knowledge.

Super-resolution multi-reference alignment.

We study super-resolution multi-reference alignment, the problem of estimating a signal from many circularly shifted, down-sampled and noisy observations. We focus on the low SNR regime, and show that a signal in ℝ M is uniquely determined when the number L of samples per observation is of the order of the square root of the signal's length ( L = O ( M ) ). Phrased more informally, one can square the resolution. This result holds if the number of observations is proportional to 1/SNR3. In contrast, with fewer observations recovery is impossible even when the observations are not down-sampled (L = M). The analysis combines tools from statistical signal processing and invariant theory. We design an expectation-maximization algorithm and demonstrate that it can super-resolve the signal in challenging SNR regimes.

Optimization of the BCLC Staging System for Locoregional Therapy for Hepatocellular Carcinoma by Using Quantitative Tumor Burden Imaging Biomarkers at MRI.

Background Patients with intermediate- and advanced-stage hepatocellular carcinoma (HCC) represent a highly heterogeneous patient collective with substantial differences in overall survival. Purpose To evaluate enhancing tumor volume (ETV) and enhancing tumor burden (ETB) as new criteria within the Barcelona Clinic Liver Cancer (BCLC) staging system for optimized allocation of patients with intermediate- and advanced-stage HCC to undergo transarterial chemoembolization (TACE). Materials and Methods In this retrospective study, 682 patients with HCC who underwent conventional TACE or TACE with drug-eluting beads from January 2000 to December 2014 were evaluated. Quantitative three-dimensional analysis of contrast-enhanced MRI was performed to determine thresholds of ETV and ETB (ratio of ETV to normal liver volume). Patients with ETV below 65 cm3 or ETB below 4% were reassigned to BCLC Bn, whereas patients with ETV or ETB above the determined cutoffs were restratified or remained in BCLC Cn by means of stepwise verification of the median overall survival (mOS). Results This study included 494 patients (median age, 62 years [IQR, 56-71 years]; 401 men). Originally, 123 patients were classified as BCLC B with mOS of 24.3 months (95% CI: 21.4, 32.9) and 371 patients as BCLC C with mOS of 11.9 months (95% CI: 10.5, 14.8). The mOS of all included patients (including the BCLC B and C groups) was 15 months (95% CI: 12.3, 17.2). A total of 152 patients with BCLC C tumors were restratified into a new BCLC Bn class, in which the mOS was then 25.1 months (95% CI: 21.8, 29.7; P < .001). The mOS of the remaining patients (ie, BCLC Cn group) (n = 222; ETV ≥65 cm3 or ETB ≥4%) was 8.4 months (95% CI: 6.1, 11.2). Conclusion Substratification of patients with intermediate- and advanced-stage hepatocellular carcinoma according to three-dimensional quantitative tumor burden identified patients with a survival benefit from transarterial chemoembolization before therapy. © RSNA, 2022 Online supplemental material is available for this article.

Diabetes medications and risk of HCC.

Type 2 diabetes mellitus is a recognized risk factor for HCC in patients with liver disease, independent from the etiology of their liver disease. Hence, prevention and treatment of type 2 diabetes mellitus and its underlying cause, insulin resistance, should be considered a treatment target for patients with liver disease. The drug armamentarium for diabetes is wide and consists of agents with insulin-sensitizing activity, agents that stimulate insulin secretion, insulin itself, and agents that reduce gastrointestinal and urinary glucose absorption. From an endocrinology perspective, the main goal of treatment is the achievement of euglycemia; however, in patients at risk of, or with known underlying liver disease, the choice of diabetic medication as it relates to potential hepatic carcinogenesis remains complex and should be carefully considered. In the last decade, increasing evidence has suggested that metformin may reduce the risk of HCC, whereas evidence for other classes of diabetic medications, particularly some of the newer agents including the sodium glucose cotransporter-2 inhibitors and glucagon-like peptide-1 receptor agonists, is fewer and often inconsistent. In this review, we aim to summarize the current evidence on the potential effects of the most widely used diabetic agents on liver cancer tumorigenesis.

Improved performance and consistency of deep learning 3D liver segmentation with heterogeneous cancer stages in magnetic resonance imaging.

PURPOSE: Accurate liver segmentation is key for volumetry assessment to guide treatment decisions. Moreover, it is an important pre-processing step for cancer detection algorithms. Liver segmentation can be especially challenging in patients with cancer-related tissue changes and shape deformation. The aim of this study was to assess the ability of state-of-the-art deep learning 3D liver segmentation algorithms to generalize across all different Barcelona Clinic Liver Cancer (BCLC) liver cancer stages. METHODS: This retrospective study, included patients from an institutional database that had arterial-phase T1-weighted magnetic resonance images with corresponding manual liver segmentations. The data was split into 70/15/15% for training/validation/testing each proportionally equal across BCLC stages. Two 3D convolutional neural networks were trained using identical U-net-derived architectures with equal sized training datasets: one spanning all BCLC stages ("All-Stage-Net": AS-Net), and one limited to early and intermediate BCLC stages ("Early-Intermediate-Stage-Net": EIS-Net). Segmentation accuracy was evaluated by the Dice Similarity Coefficient (DSC) on a dataset spanning all BCLC stages and a Wilcoxon signed-rank test was used for pairwise comparisons. RESULTS: 219 subjects met the inclusion criteria (170 males, 49 females, 62.8±9.1 years) from all BCLC stages. Both networks were trained using 129 subjects: AS-Net training comprised 19, 74, 18, 8, and 10 BCLC 0, A, B, C, and D patients, respectively; EIS-Net training comprised 21, 86, and 22 BCLC 0, A, and B patients, respectively. DSCs (mean±SD) were 0.954±0.018 and 0.946±0.032 for AS-Net and EIS-Net (p<0.001), respectively. The AS-Net 0.956±0.014 significantly outperformed the EIS-Net 0.941±0.038 on advanced BCLC stages (p<0.001) and yielded similarly good segmentation performance on early and intermediate stages (AS-Net: 0.952±0.021; EIS-Net: 0.949±0.027; p = 0.107). CONCLUSION: To ensure robust segmentation performance across cancer stages that is independent of liver shape deformation and tumor burden, it is critical to train deep learning models on heterogeneous imaging data spanning all BCLC stages.

Spectral neighbor joining for reconstruction of latent tree Models.

A common assumption in multiple scientific applications is that the distribution of observed data can be modeled by a latent tree graphical model. An important example is phylogenetics, where the tree models the evolutionary lineages of a set of observed organisms. Given a set of independent realizations of the random variables at the leaves of the tree, a key challenge is to infer the underlying tree topology. In this work we develop Spectral Neighbor Joining (SNJ), a novel method to recover the structure of latent tree graphical models. Given a matrix that contains a measure of similarity between all pairs of observed variables, SNJ computes a spectral measure of cohesion between groups of observed variables. We prove that SNJ is consistent, and derive a sufficient condition for correct tree recovery from an estimated similarity matrix. Combining this condition with a concentration of measure result on the similarity matrix, we bound the number of samples required to recover the tree with high probability. We illustrate via extensive simulations that in comparison to several other reconstruction methods, SNJ requires fewer samples to accurately recover trees with a large number of leaves or long edges.

Detection of differentially abundant cell subpopulations in scRNA-seq data.

Comprehensive and accurate comparisons of transcriptomic distributions of cells from samples taken from two different biological states, such as healthy versus diseased individuals, are an emerging challenge in single-cell RNA sequencing (scRNA-seq) analysis. Current methods for detecting differentially abundant (DA) subpopulations between samples rely heavily on initial clustering of all cells in both samples. Often, this clustering step is inadequate since the DA subpopulations may not align with a clear cluster structure, and important differences between the two biological states can be missed. Here, we introduce DA-seq, a targeted approach for identifying DA subpopulations not restricted to clusters. DA-seq is a multiscale method that quantifies a local DA measure for each cell, which is computed from its k nearest neighboring cells across a range of k values. Based on this measure, DA-seq delineates contiguous significant DA subpopulations in the transcriptomic space. We apply DA-seq to several scRNA-seq datasets and highlight its improved ability to detect differences between distinct phenotypes in severe versus mildly ill COVID-19 patients, melanomas subjected to immune checkpoint therapy comparing responders to nonresponders, embryonic development at two time points, and young versus aging brain tissue. DA-seq enabled us to detect differences between these phenotypes. Importantly, we find that DA-seq not only recovers the DA cell types as discovered in the original studies but also reveals additional DA subpopulations that were not described before. Analysis of these subpopulations yields biological insights that would otherwise be undetected using conventional computational approaches.