Alexander Spektor

Purpose: Clinical assessment of vertebral lesion quality (osteolytic, osteoblastic, mixed) remains subjective, with limited interobserver reliability. This study evaluated a novel application of 3D convolutional neural networks (3D-CNNs) for classifying lesion quality from CT volumes in metastatic cancer patients. Materials and Methods: This retrospective study used CT data from 151 cancer patients planned for radiotherapy for metastatic spine disease (September 2020-July 2024). Leveraging vertebra-level expert annotations, we introduced an unconventional U-Net-based strategy converting coarse voxel-wise predictions into vertebra-level lesion classifications. The final dataset comprised 2,125 vertebrae across four classes (no lesion, osteolytic, osteoblastic, mixed), split into a 3-fold cross-validation set and an independent holdout test set. Model performance was benchmarked against a DenseNet121 baseline and a musculoskeletal radiologist, with Cohen's kappa assessing inter-rater agreement. Results: The 3D model achieved an ensemble accuracy of 84.7%, outperforming DenseNet121 (72.1%), with substantial gains in F1 score, precision, and balanced accuracy. It showed high concordance with the radiologist (Cohen's kappa = 0.76) and comparable sensitivity and specificity across all lesion subtypes. We found both models and the radiologist to struggle with osteolytic lesions, reflecting the difficulty of distinguishing this class from age-related changes in vertebral bone density and architecture caused by benign bone lesions, age-related systemic skeletal disorders and cancer treatments. Conclusions: 3D-CNNs trained with vertebra-level labels can accurately and reliably classify vertebral metastatic lesion quality from CT scans, offering a scalable path toward automated characterization of metastatic spine disease to support clinical decision-making and large-scale radiomics research.

Breast cancer remains largely unresponsive to immunotherapy due to its immunologically “cold” nature, marked by low cytolytic activity and a paucity of neoantigens. To overcome this, we developed a novel therapeutic approach to activate the cGAS-STING pathway, a critical mediator of the type I interferon response essential for effective anti-tumor immunity. We demonstrate that combined inhibition of MCL1 and caspases robustly triggers a type I IFN response in breast cancer cells, effectively remodeling the tumor microenvironment by increasing immune cell infiltration and enhancing antigen presentation. In immunocompetent syngeneic mouse models, this combination therapy significantly suppressed tumor growth, an effect that was reversed upon blockade of the IFN signaling axis. Mechanistically, our findings reveal that co-targeting MCL1 and caspases reprograms the tumor microenvironment enhancing immune surveillance. Given the established safety profiles of both drug classes, this strategy offers a promising and rapidly deployable approach to sensitize breast tumors to immunotherapy.

Small-cell lung cancers (SCLCs) contain near-universal loss-of-function mutations in RB1 and TP53, compromising the G1–S checkpoint and leading to dysregulated E2F activity¹. Other cancers similarly disrupt the G1–S checkpoint through loss of CDKN2A or amplification of cyclin D or cyclin E, also resulting in excessive E2F activity^2,3. Although E2F activation is essential for cell cycle progression, hyperactivation promotes apoptosis^4,5,6,7,8,9, presenting a therapeutic vulnerability. Cyclin proteins use a conserved hydrophobic patch to bind to substrates bearing short linear RxL motifs^10,11,12,13. Cyclin A represses E2F through an RxL-dependent interaction^10,14, which, when disrupted, hyperactivates E2F¹⁵. However, this substrate interface has remained difficult to target. Here we developed cell-permeable, orally bioavailable macrocyclic peptides that inhibit RxL-mediated interactions of cyclins with their substrates. Dual inhibitors of cyclin A and cyclin B RxL motifs (cyclin A/Bi) selectively kill SCLC cells and other cancer cells with high E2F activity. Genetic screens revealed that cyclin A/Bi induces apoptosis through cyclin B- and CDK2-dependent spindle assembly checkpoint activation. Mechanistically, cyclin A/Bi hyperactivates E2F and cyclin B by blocking cyclin A–E2F and cyclin B–MYT1 RxL interactions. Notably, cyclin A/Bi promoted the formation of neomorphic cyclin B–CDK2 complexes, which drive spindle assembly checkpoint activation and mitotic cell death. Finally, orally administered cyclin A/Bi showed robust anti-tumour activity in chemotherapy-resistant SCLC patient-derived xenografts. These findings reveal gain-of-function mechanisms through which cyclin A/Bi triggers apoptosis and support their development for E2F-driven cancers.

Purpose Given the high prevalence of vertebral fractures post-radiotherapy in patients with metastatic spine disease, accurate and rapid muscle segmentation could support efforts to quantify muscular changes due to disease or treatment and enable biomechanical modeling for assessments of vertebral loading to improve personalized evaluation of vertebral fracture risk. This study presents a deep-learning approach for segmenting the complete volume of the trunk muscles from clinical CT images trained using sparsely annotated data. Materials and Methods we extracted 2,009 axial CT images at the midpoint of each vertebral level (T4 to L4) from clinical CT of 148 cancer patients. The key extensor and flexor muscles (up to 8 muscles per side) were manually contoured and labeled per image in the thoracic and lumbar regions. We first trained a 2D nnU-Net deep-learning model on these labels to segment key extensor and flexor muscles. Using these sparse annotations per spine, we trained the model to segment each muscle’s entire 3D volume Results The proposed method achieved comparable performance to manual segmentations, as assessed by expert radiologists, with a mean Dice score above 0.769. Significantly, the model drastically reduced segmentation time, from 4.3-6.5 hours for manual segmentation of 14 single axial CT images to approximately 1 minute for segmenting the complete thoracic-abdominal 3D volume. Conclusion The approach demonstrates high potential for automating 3D muscle segmentation, significantly reducing the manual intervention required for generating musculoskeletal models, and could be instrumental in enhancing clinical decision-making and patient care in radiation oncology. Summary A deep learning 2D nnU-Net model, trained on a sparse set of 2D muscle annotations, successfully segmented the entire volume of 20 thoracolumbar muscles from cancer patients’ clinical CT data. The model showed a remarkable increase in segmentation efficacy and generalizability, achieving comparable performance to manual segmentations in delineating each muscle anatomy. Key Points ▪ A deep learning model (2D nnU-Net), developed using a sparse set of single axial CT-slice at each mid-per vertebral level, containing manual image annotation of 20 thoracic and lumbar muscles, achieved comparable performance to manual segmentations, as assessed by expert radiologists, with a mean Dice score above 0.769. ▪ The model drastically reduced segmentation time, from 4.3-6.5 hours for manual segmentation of 14 single axial CT images to approximately 1 minute for segmenting the complete thoracic-abdominal 3D volume. ▪ Radiologist assessment based on a Likert scale (0-5) for clinical acceptability of the muscle anatomical segmentation showed strong model performance for a representative sample of clinical CT data (a (mean(SD) of 4.66 (0.73)) and external data (4.66 (0.73).

Pathologic vertebral fractures (PVF) are common and serious complications in patients with metastatic lesions affecting the spine. Accurate assessment of cancer patients’ PVF risk is an unmet clinical need. Load-to-strength ratios (LSRs) evaluated in vivo by estimating vertebral loading from biomechanical modeling and strength from computed tomography imaging (CT) have been associated with osteoporotic vertebral fractures in older adults. Here, for the first time, we investigate LSRs of thoracic and lumbar vertebrae of 135 spine metastases patients compared to LSRs of 246 healthy adults, comparable by age and sex, from the Framingham Heart Study under four loading tasks. Findings include: (1) Osteolytic vertebrae have higher LSRs than osteosclerotic and mixed vertebrae; (2). In patients’ vertebrae without CT observed metastases, LSRs were greater than healthy controls. (3) LSRs depend on the spinal region (Thoracic, Thoracolumbar, Lumbar). These findings suggest that LSRs may contribute to identifying patients at risk of incident PVF in metastatic spine disease patients. The lesion-mediated difference suggests that risk thresholds should be established based on spinal region, simulated task, and metastatic lesion type.

CDK1 has been known to be the sole cyclin-dependent kinase (CDK) partner of cyclin B1 to drive mitotic progression¹. Here we demonstrate that CDK5 is active during mitosis and is necessary for maintaining mitotic fidelity. CDK5 is an atypical CDK owing to its high expression in post-mitotic neurons and activation by non-cyclin proteins p35 and p39². Here, using independent chemical genetic approaches, we specifically abrogated CDK5 activity during mitosis, and observed mitotic defects, nuclear atypia and substantial alterations in the mitotic phosphoproteome. Notably, cyclin B1 is a mitotic co-factor of CDK5. Computational modelling, comparison with experimentally derived structures of CDK–cyclin complexes and validation with mutational analysis indicate that CDK5–cyclin B1 can form a functional complex. Disruption of the CDK5–cyclin B1 complex phenocopies CDK5 abrogation in mitosis. Together, our results demonstrate that cyclin B1 partners with both CDK5 and CDK1, and CDK5–cyclin B1 functions as a canonical CDK–cyclin complex to ensure mitotic fidelity.

53BP1 nucleates the anti-end resection machinery at DNA double-strand breaks, thereby countering BRCA1 activity. Loss of 53BP1 leads to DNA end processing and homologous recombination in BRCA1-deficient cells. Consequently, BRCA1-mutant tumors, typically sensitive to PARP inhibitors (PARPi), become resistant in the absence of 53BP1. Here, we demonstrate that the ‘leaky’ DNA end resection in the absence of 53BP1 results in increased micronuclei and cytoplasmic double-stranded DNA, leading to activation of the cGAS-STING pathway and pro-inflammatory signaling. This enhances CD8⁺ T cell infiltration, activates macrophages and natural killer cells, and impedes tumor growth. Loss of 53BP1 correlates with a response to immune checkpoint blockade (ICB) and improved overall survival. Immunohistochemical assessment of 53BP1 in two malignancies, high grade serous ovarian cancer and pancreatic ductal adenocarcinoma, which are refractory to ICBs, reveals that lower 53BP1 levels correlate with an increased adaptive and innate immune response. Finally, BRCA1-deficient tumors that develop resistance to PARPi due to the loss of 53BP1 are susceptible to ICB. Therefore, we conclude that 53BP1 is critical for tumor immunogenicity and underpins the response to ICB. Our results support including 53BP1 expression as an exploratory biomarker in ICB trials for malignancies typically refractory to immunotherapy.

Nuclear atypia, including altered nuclear size, contour, and chromatin organization, is ubiquitous in cancer cells. Atypical primary nuclei and micronuclei can rupture during interphase; however, the frequency, causes, and consequences of nuclear rupture are unknown in most cancers. We demonstrate that nuclear envelope rupture is surprisingly common in many human cancers, particularly glioblastoma. Using highly-multiplexed 2D and super-resolution 3D-imaging of glioblastoma tissues and patient-derived xenografts and cells, we link primary nuclear rupture with reduced lamin A/C and micronuclear rupture with reduced lamin B1. Moreover, ruptured glioblastoma cells activate cGAS-STING-signaling involved in innate immunity. We observe that local patterning of cell states influences tumor spatial organization and is linked to both lamin expression and rupture frequency, with neural-progenitor-cell-like states exhibiting the lowest lamin A/C levels and greatest susceptibility to primary nuclear rupture. Our study reveals that nuclear instability is a core feature of cancer, and links nuclear integrity, cell state, and immune signaling.

Genomic havoc from one fateful mistake: Many human tumors display scrambled genomes that arise from two distinct mutational processes. The first, the chromosome breakage-fusion-bridge (BFB) cycle, produces gene amplification and genomic instability. The second, chromothripsis, generates massive, clustered genomic rearrangements in one or a few chromosomes. Umbreit et al. hypothesized that these two processes are mechanistically related and tested this idea by recreating essential steps of the BFB cycle in cultured cells (see the Perspective by Paiano and Nussenzweig). They found that chromothripsis arises from a cascade of events that begins with aberrant chromosome bridge formation during mitosis, followed by chromosome fragmentation, DNA damage, chromosome missegregation, and the formation of micronuclei. They propose a model that explains how a single cell division error (chromosome bridge formation) can generate many hallmark features of cancer genomes. Science , this issue p. 240 ; see also p. eaba0712