Purpose Checkpoint inhibitors demonstrate salutary anti-cancer effects including long-term remissions. PD-L1 expression/amplification, high mutational burden and mismatch repair-deficiency correlate with response. We have, however, observed a subset of patients who appear to be “hyper-progressors,” with a greatly accelerated rate of tumor growth and clinical deterioration compared to pre-therapy, which was also recently reported by Institut Gustave Roussy. The current study investigated potential genomic markers associated with “hyper-progression” after immunotherapy. Method Consecutive stage IV cancer patients who received immunotherapies (CTLA-4, PD-1/PD-L1 inhibitors or other [investigational] agents) and had their tumor evaluated by next-generation sequencing were analyzed (N=155). We defined hyper-progression as time-to-treatment failure (TTF) <2 months, >50% increase in tumor burden compared to pre-immunotherapy imaging, and >2-fold increase in progression pace. Results Amongst 155 patients, TTF <2 months was seen in all six individuals with MDM2/MDM4 amplification. After anti-PD1/PDL1 monotherapy, four of these patients showed remarkable increases in existing tumor size (55% to 258%), new large masses, and significantly accelerated progression pace (2.3-, 7.1-, 7.2- and 42.3-fold compared to the two months before immunotherapy). In multivariate analysis, MDM2/MDM4 and EGFR alterations correlated with TTF<2 months. Two of 10 patients with EGFR alterations were also hyper-progressors (53.6% and 125% increase in tumor size; 35.7- and 41.7-fold increase). Conclusion Some patients with MDM2 family amplification or EGFR aberrations had poor clinical outcome and significantly increased rate of tumor growth after single-agent checkpoint (PD-1/PD-L1) inhibitors. Genomic profiles may help to identify patients at risk for progression on immunotherapy. Further investigation is urgently needed.

BackgroundPathologists use visual classification of glomerular lesions to assess samples from patients with diabetic nephropathy (DN). The results may vary among pathologists. Digital algorithms may reduce this variability and provide more consistent image structure interpretation.MethodsWe developed a digital pipeline to classify renal biopsies from patients with DN. We combined traditional image analysis with modern machine learning to efficiently capture important structures, minimize manual effort and supervision, and enforce biologic prior information onto our model. To computationally quantify glomerular structure despite its complexity, we simplified it to three components consisting of nuclei, capillary lumina and Bowman spaces; and Periodic Acid-Schiff positive structures. We detected glomerular boundaries and nuclei from whole slide images using convolutional neural networks, and the remaining glomerular structures using an unsupervised technique developed expressly for this purpose. We defined a set of digital features which quantify the structural progression of DN, and a recurrent network architecture which processes these features into a classification.ResultsOur digital classification agreed with a senior pathologist whose classifications were used as ground truth with moderate Cohen’s kappa κ = 0.55 and 95% confidence interval [0.50, 0.60]. Two other renal pathologists agreed with the digital classification with κ1 = 0.68, 95% interval [0.50, 0.86] and κ2 = 0.48, 95% interval [0.32, 0.64]. Our results suggest computational approaches are comparable to human visual classification methods, and can offer improved precision in clinical decision workflows. We detected glomerular boundaries from whole slide images with 0.93±0.04 balanced accuracy, glomerular nuclei with 0.94 sensitivity and 0.93 specificity, and glomerular structural components with 0.95 sensitivity and 0.99 specificity.ConclusionsComputationally derived, histologic image features hold significant diagnostic information that may augment clinical diagnostics.

Kidney failure is common in patients with Coronavirus Disease-19 (COVID-19) resulting in increased morbidity and mortality. In an international collaboration, 284 kidney biopsies were evaluated to improve understanding of kidney disease in COVID-19. Diagnoses were compared to five years of 63,575 native biopsies prior to the pandemic and 13,955 allograft biopsies to identify diseases increased in patients with COVID-19. Genotyping for APOL1 G1 and G2 alleles was performed in 107 African American and Hispanic patients. Immunohistochemistry for SARS-CoV-2 was utilized to assess direct viral infection in 273 cases along with clinical information at the time of biopsy. The leading indication for native biopsy was acute kidney injury (45.4%), followed by proteinuria with or without concurrent acute kidney injury (42.6%). There were more African American patients (44.6%) than patients of other ethnicities. The most common diagnosis in native biopsies was collapsing glomerulopathy (25.8%) which associated with high-risk APOL1 genotypes in 91.7% of cases. Compared to the five-year biopsy database, the frequency of myoglobin cast nephropathy and proliferative glomerulonephritis with monoclonal IgG deposits was also increased in patients with COVID-19 (3.3% and 1.7%, respectively), while there was a reduced frequency of chronic conditions (including diabetes mellitus, IgA nephropathy, and arterionephrosclerosis) as the primary diagnosis. In transplants, the leading indication was acute kidney injury (86.4%), for which rejection was the predominant diagnosis (61.4%). Direct SARS-CoV-2 viral infection was not identified. Thus, our multi-center large case series identified kidney diseases that disproportionately affect patients with COVID-19, demonstrated a high frequency of APOL1 high-risk genotypes within this group, with no evidence of direct viral infection within the kidney.

BackgroundInterstitial fibrosis, tubular atrophy (IFTA), and glomerulosclerosis are indicators of irrecoverable kidney injury. Modern machine learning (ML) tools have enabled robust, automated identification of image structures that can be comparable with analysis by human experts. ML algorithms were developed and tested for the ability to replicate the detection and quantification of IFTA and glomerulosclerosis that renal pathologists perform.MethodsA renal pathologist annotated renal biopsy specimens from 116 whole-slide images (WSIs) for IFTA and glomerulosclerosis. A total of 79 WSIs were used for training different configurations of a convolutional neural network (CNN), and 17 and 20 WSIs were used as internal and external testing cases, respectively. The best model was compared against the input of four renal pathologists on 20 new testing slides. Further, for 87 testing biopsy specimens, IFTA and glomerulosclerosis measurements made by pathologists and the CNN were correlated to patient outcome using classic statistical tools.ResultsThe best average performance across all image classes came from a DeepLab version 2 network trained at 40× magnification. IFTA and glomerulosclerosis percentages derived from this CNN achieved high levels of agreement with four renal pathologists. The pathologist- and CNN-based analyses of IFTA and glomerulosclerosis showed statistically significant and equivalent correlation with all patient-outcome variables.ConclusionsML algorithms can be trained to replicate the IFTA and glomerulosclerosis assessment performed by renal pathologists. This suggests computational methods may be able to provide a standardized approach to evaluate the extent of chronic kidney injury in situations in which renal-pathologist time is restricted or unavailable.

Cancer of unknown primary (CUP) is a rare malignancy that presents with metastatic disease and no identifiable site of origin. Most patients have unfavorable features and attempts to treat based on tissue-of-origin identification have not yielded a survival advantage compared with empiric chemotherapy. Next-generation sequencing has revealed genomic alterations that can be targeted in selected cases, suggesting that CUP represents a unique malignancy in which the genomic aberrations may be integral to the diagnosis. Recent trials focusing on tailored combination therapy matched to the genomic alterations in each cancer are providing new avenues of clinical investigation. Here, we discuss recent findings on molecular aberrations in CUP and how the genomic and immune landscape can be leveraged to optimize therapy. HighlightsCancer of unknown primary (CUP), by definition, is metastatic disease with an unidentifiable primary tumor.Patients with CUP are generally treated with empiric chemotherapies, such as taxanes and platinum-containing regimens; however, clinical outcomes remain poor.Recent studies with next-generation sequencing revealed that most CUP tumors harbored unique and complex genomic portfolios, with a mean of four to five alterations per tumor.CUP represents a unique cancer in which the genomic alterations may be the cornerstone of the diagnosis. Matched individualized combination targeted therapy in CUP merits prospective clinical investigation.