Determining risk and predictors of head and neck cancer treatment-related lymphedema: A clinicopathologic and dosimetric data mining approach using interpretable machine learning and ensemble feature selection

Background and purpose

The ability to determine the risk and predictors of lymphedema is vital in improving the quality of life for head and neck (HN) cancer patients. However, selecting robust features is challenging due to the multicollinearity and high dimensionality of radiotherapy (RT) data. This study aims to overcome these challenges using an ensemble feature selection technique with machine learning (ML).

Materials and methods

Thirty organs-at-risk, including bilateral cervical lymph node levels, were contoured, and dose-volume data were extracted from 76 HN treatment plans. Clinicopathologic data was collected. Ensemble feature selection was used to reduce the number of features. Using the reduced features as input to ML and competing risk models, internal and external lymphedema prediction capability was evaluated with the ML models, and time to lymphedema event and risk stratification were estimated using the risk models.

Results

Two ML models, XGBoost and random forest, exhibited robust prediction performance. They achieved average F1-scores and AUCs of 84 ± 3.3 % and 79 ± 11.9 % (external lymphedema), and 64 ± 12 % and 78 ± 7.9 % (internal lymphedema). Predictive ML and risk models identified common predictors, including bulky node involvement, high dose to various lymph node levels, and lymph nodes removed during surgery. At 180 days, removing 0-25, 26-50, and > 50 lymph nodes increased external lymphedema risk to 72.1 %, 95.6 %, and 57.7 % respectively (p = 0.01).

Conclusion

Our approach, involving the reduction of HN RT data dimensionality, resulted in effective ML models for HN lymphedema prediction. Predictive dosimetric features emerged from both predictive and competing risk models. Consistency with clinicopathologic features from other studies supports our methodology.

Quality of Large Language Model Responses to Radiation Oncology Patient Care Questions

Importance

Artificial intelligence (AI) large language models (LLMs) demonstrate potential in simulating human-like dialogue. Their efficacy in accurate patient-clinician communication within radiation oncology has yet to be explored.

Objective

To determine an LLM's quality of responses to radiation oncology patient care questions using both domain-specific expertise and domain-agnostic metrics.

Design, Setting, and Participants

This cross-sectional study retrieved questions and answers from websites (accessed February 1 to March 20, 2023) affiliated with the National Cancer Institute and the Radiological Society of North America. These questions were used as queries for an AI LLM, ChatGPT version 3.5 (accessed February 20 to April 20, 2023), to prompt LLM-generated responses. Three radiation oncologists and 3 radiation physicists ranked the LLM-generated responses for relative factual correctness, relative completeness, and relative conciseness compared with online expert answers. Statistical analysis was performed from July to October 2023.

Main Outcomes and Measures

The LLM's responses were ranked by experts using domain-specific metrics such as relative correctness, conciseness, completeness, and potential harm compared with online expert answers on a 5-point Likert scale. Domain-agnostic metrics encompassing cosine similarity scores, readability scores, word count, lexicon, and syllable counts were computed as independent quality checks for LLM-generated responses.

Results

Of the 115 radiation oncology questions retrieved from 4 professional society websites, the LLM performed the same or better in 108 responses (94%) for relative correctness, 89 responses (77%) for completeness, and 105 responses (91%) for conciseness compared with expert answers. Only 2 LLM responses were ranked as having potential harm. The mean (SD) readability consensus score for expert answers was 10.63 (3.17) vs 13.64 (2.22) for LLM answers (P < .001), indicating 10th grade and college reading levels, respectively. The mean (SD) number of syllables was 327.35 (277.15) for expert vs 376.21 (107.89) for LLM answers (P = .07), the mean (SD) word count was 226.33 (191.92) for expert vs 246.26 (69.36) for LLM answers (P = .27), and the mean (SD) lexicon score was 200.15 (171.28) for expert vs 219.10 (61.59) for LLM answers (P = .24).

Conclusions and Relevance

In this cross-sectional study, the LLM generated accurate, comprehensive, and concise responses with minimal risk of harm, using language similar to human experts but at a higher reading level. These findings suggest the LLM's potential, with some retraining, as a valuable resource for patient queries in radiation oncology and other medical fields.

Dose prescription and reporting in stereotactic body radiotherapy: A multi-institutional study

BACKGROUND AND PURPOSE

Radiation dose prescriptions are foundational for optimizing treatment efficacy and limiting treatment-related toxicity. We sought to assess the lack of standardization of SBRT dose prescriptions across institutions.

MATERIALS & METHODS

Dosimetric data from 1298 patients from 9 academic institutions treated with IMRT and VMAT were collected. Dose parameters D100, D98, D95, D50, and D2 were used to assess dosimetric variability.

RESULTS

Disease sites included lung (48.3 %) followed by liver (29.7 %), prostate (7.5 %), spine (6.8 %), brain (4.1 %), and pancreas (2.5 %). The PTV volume in lung varied widely with bimodality into two main groups (22.0-28.7 cm3) and (48.0-67.1 cm3). A hot spot ranging from 120-150 % was noted in nearly half of the patients, with significant variation across institutions. A D50 ≥ 110 % was found in nearly half of the institutions. There was significant dosimetric variation across institutions.

CONCLUSIONS

The SBRT prescriptions in the literature or in treatment guidelines currently lack nuance and hence there is significant variation in dose prescriptions across academic institutions. These findings add greater importance to the identification of dose parameters associated with improved clinical outcome comparisons as we move towards more hypofractionated treatments. There is a need for standardized reporting to help institutions in adapting treatment protocols based on the outcome of clinical trials. Dosimetric parameters are subsequently needed for uniformity and thereby standardizing planning guidelines to maximize efficacy, mitigate toxicity, and reduce treatment disparities are urgently needed.

Abstract 2814: The mutational basis of cancer's vulnerability to ionizing radiation

Introduction: Large-scale sequencing efforts have established that cancer-associated genetic alterations are highly diverse, posing a challenge to the identification of variants that regulate complex phenotypes like radiation sensitivity. The impact of the vast majority of rare or common genetic variants on the sensitivity of cancers to radiotherapy remains largely unknown.

Methods: We developed a scalable gene editing and irradiation platform to assess the role of categories of variants in cells. Variants were prioritized based on genotype-phenotype associations from a previously completed large-scale cancer cell line radiation profiling study. Altogether, 488 alleles (396 unique SNVs) from 92 genes were generated and profiled in an immortalized lung cell line, BEAS-2B. We validated our results in other cell lines (TRT-HU1 and NCI-H520), in vivo via the use of both cell line and patient-derived murine xenografts, and in clinical cohorts.

Results: We show that resistance to radiation is characterized by substantial inter- and intra-gene allelic variation. Some genes (e.g. KEAP1) demonstrated significant intragenic allelic variation in the magnitude of conferred resistance and other genes (e.g. CTNNB1) displayed both resistance and sensitivity in a protein domain-dependent manner. We combined results from our platform with gene expression and metabolite features and identified the upregulation of amino acid transporters that facilitate oxidative reductive capacity and cell cycle deregulation as key regulators of radiation sensitivity.

Conclusions: Our results reveal new insights into the genetic determinants of tumor sensitivity to radiotherapy and nominate a multitude of cancer mutations that are predicted to impact treatment efficacy.

Citation Format: Priyanka Gopal, Mohamed E. Abazeed. The mutational basis of cancer's vulnerability to ionizing radiation [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 2814.

Image-Based Deep Neural Network for Individualizing Radiotherapy Dose Is Transportable Across Health Systems

PURPOSE

We developed a deep neural network that queries the lung computed tomography-derived feature space to identify radiation sensitivity parameters that can predict treatment failures and hence guide the individualization of radiotherapy dose. In this article, we examine the transportability of this model across health systems.

METHODS

This multicenter cohort-based registry included 1,120 patients with cancer in the lung treated with stereotactic body radiotherapy. Pretherapy lung computed tomography images from the internal study cohort (n = 849) were input into a multitask deep neural network to generate an image fingerprint score that predicts time to local failure. Deep learning (DL) scores were input into a regression model to derive iGray, an individualized radiation dose estimate that projects a treatment failure probability of < 5% at 24 months. We validated our findings in an external, holdout cohort (n = 271).

RESULTS

There were substantive differences in the baseline patient characteristics of the two study populations, permitting an assessment of model transportability. In the external cohort, radiation treatments in patients with high DL scores failed at a significantly higher rate with 3-year cumulative incidences of local failure of 28.5% (95% CI, 19.8 to 37.8) versus 10.2% (95% CI, 5.9 to 16.2; hazard ratio, 3.3 [95% CI, 1.74 to 6.49]; P < .001). A model that included DL score alone predicted treatment failures with a concordance index of 0.68 (95% CI, 0.59 to 0.77), which had a similar performance to a nested model derived from within the internal cohort (0.70 [0.64 to 0.75]). External cohort patients with iGray values that exceeded the delivered doses had proportionately higher rates of local failure (P < .001).

CONCLUSION

Our results support the development and implementation of new DL-guided treatment guidance tools in the image-replete and highly standardized discipline of radiation oncology.

The Mutational Landscape of Cancer's Vulnerability to Ionizing Radiation

PURPOSE

Large-scale sequencing efforts have established that cancer-associated genetic alterations are highly diverse, posing a challenge to the identification of variants that regulate complex phenotypes like radiation sensitivity. The impact of the vast majority of rare or common genetic variants on the sensitivity of cancers to radiotherapy remains largely unknown.

EXPERIMENTAL DESIGN

We developed a scalable gene editing and irradiation platform to assess the role of categories of variants in cells. Variants were prioritized on the basis of genotype-phenotype associations from a previously completed large-scale cancer cell line radiation profiling study. Altogether, 488 alleles (396 unique single-nucleotide variants) from 92 genes were generated and profiled in an immortalized lung cell line, BEAS-2B. We validated our results in other cell lines (TRT-HU1 and NCI-H520), in vivo via the use of both cell line and patient-derived murine xenografts, and in clinical cohorts.

RESULTS

We show that resistance to radiation is characterized by substantial inter- and intra-gene allelic variation. Some genes (e.g., KEAP1) demonstrated significant intragenic allelic variation in the magnitude of conferred resistance and other genes (e.g., CTNNB1) displayed both resistance and sensitivity in a protein domain-dependent manner. We combined results from our platform with gene expression and metabolite features and identified the upregulation of amino acid transporters that facilitate oxidative reductive capacity and cell-cycle deregulation as key regulators of radiation sensitivity.

CONCLUSIONS

Our results reveal new insights into the genetic determinants of tumor sensitivity to radiotherapy and nominate a multitude of cancer mutations that are predicted to impact treatment efficacy.

Multivalent state transitions shape the intratumoral composition of small cell lung carcinoma

Studies to date have not resolved how diverse transcriptional programs contribute to the intratumoral heterogeneity of small cell lung carcinoma (SCLC), an aggressive tumor associated with a dismal prognosis. Here, we identify distinct and commutable transcriptional states that confer discrete functional attributes in individual SCLC tumors. We combine an integrative approach comprising the transcriptomes of 52,975 single cells, high-resolution measurement of cell state dynamics at the single-cell level, and functional and correlative studies using treatment naïve xenografts with associated clinical outcomes. We show that individual SCLC tumors contain distinctive proportions of stable cellular states that are governed by bidirectional cell state transitions. Using drugs that target the epigenome, we reconfigure tumor state composition in part by altering individual state transition rates. Our results reveal new insights into how single-cell transition behaviors promote cell state equilibrium in SCLC and suggest that facile plasticity underlies its resistance to therapy and lethality.

Deterministic small-scale undulations of image-based risk predictions from the deep learning of lung tumors in motion

INTRODUCTION

Deep learning (DL) models that use medical images to predict clinical outcomes are poised for clinical translation. For tumors that reside in organs that move, however, the impact of motion (i.e., degenerated object appearance or blur) on DL model accuracy remains unclear. We examine the impact of tumor motion on an image-based DL framework that predicts local failure risk after lung stereotactic body radiotherapy (SBRT).

METHODS

We input pre-therapy free breathing (FB) computed tomography (CT) images from 849 patients treated with lung SBRT into a multitask deep neural network to generate an image fingerprint signature (or DL score) that predicts time-to-event local failure outcomes. The network includes a convolutional neural network encoder for extracting imaging features and building a task-specific fingerprint, a decoder for estimating handcrafted radiomic features, and a task-specific network for generating image signature for radiotherapy outcome prediction. The impact of tumor motion on the DL scores was then examined for a holdout set of 468 images from 39 patients comprising: (1) FB CT, (2) four-dimensional (4D) CT, and (3) maximum-intensity projection (MIP) images. Tumor motion was estimated using a 3D vector of the maximum distance traveled, and its association with DL score variance was assessed by linear regression.

FINDINGS

The variance and amplitude in 4D CT image-derived DL scores were associated with tumor motion (R²  = 0.48 and 0.46, respectively). Specifically, DL score variance was deterministic and represented by sinusoidal undulations in phase with the respiratory cycle. DL scores, but not tumor volumes, peaked near end-exhalation. The mean of the scores derived from 4D CT images and the score obtained from FB CT images were highly associated (Pearson r = 0.99). MIP-derived DL scores were significantly higher than 4D- or FB-derived risk scores (p < 0.0001).

INTERPRETATION

An image-based DL risk score derived from a series of 4D CT images varies in a deterministic, sinusoidal trajectory in a phase with the respiratory cycle. These results indicate that DL models of tumors in motion can be robust to fluctuations in object appearance due to movement and can guide standardization processes in the clinical translation of DL models for patients with lung cancer.

Fractionated Proton Radiation Therapy and Hearing Preservation for Vestibular Schwannoma: Preliminary Analysis of a Prospective Phase 2 Clinical Trial

BACKGROUND

Local management for vestibular schwannoma (VS) is associated with excellent local control with focus on preserving long-term serviceable hearing. Fractionated proton radiation therapy (FPRT) may be associated with greater hearing preservation because of unique dosimetric properties of proton radiotherapy.

OBJECTIVE

To investigate hearing preservation rates of FPRT in adults with VS and secondarily assess local control and treatment-related toxicity.

METHODS

A prospective, single-arm, phase 2 clinical trial was conducted of patients with VS from 2010 to 2019. All patients had serviceable hearing at baseline and received FPRT to a total dose of 50.4 to 54 Gy relative biological effectiveness (RBE) over 28 to 30 fractions. Serviceable hearing preservation was defined as a Gardner-Robertson score of 1 to 2, measured by a pure tone average (PTA) of ≤50 dB and a word recognition score (WRS) of ≥50%.

RESULTS

Twenty patients had a median follow-up of 4.0 years (range 1.0-5.0 years). Local control at 4 years was 100%. Serviceable hearing preservation at 1 year was 53% (95% CI 29%-76%), and primary end point was not yet reached. Median PTA and median WRS both worsened 1 year after FPRT (P < .0001). WRS plateaued after 6 months, whereas PTA continued to worsen up to 1 year after FPRT. Median cochlea D90 was lower in patients with serviceable hearing at 1 year (40.6 Gy [RBE] vs 46.9 Gy [RBE]), trending toward Wilcoxon rank-sum test statistical significance (P = .0863). Treatment was well-tolerated, with one grade 1 cranial nerve V dysfunction and no grade 2+ cranial nerve dysfunction.

CONCLUSION

FPRT for VS did not meet the goal of serviceable hearing preservation. Higher cochlea doses trended to worsening hearing preservation, suggesting that dose to cochlea correlates with hearing preservation independent of treatment modality.

Therapy-Induced Senescence: Opportunities to Improve Anticancer Therapy

Cellular senescence is an essential tumor suppressive mechanism that prevents the propagation of oncogenically activated, genetically unstable, and/or damaged cells. Induction of tumor cell senescence is also one of the underlying mechanisms by which cancer therapies exert antitumor activity. However, an increasing body of evidence from preclinical studies demonstrates that radiation and chemotherapy cause accumulation of senescent cells (SnCs) both in tumor and normal tissue. SnCs in tumors can, paradoxically, promote tumor relapse, metastasis, and resistance to therapy, in part, through expression of the senescence-associated secretory phenotype. In addition, SnCs in normal tissue can contribute to certain radiation- and chemotherapy-induced side effects. Because of its multiple roles, cellular senescence could serve as an important target in the fight against cancer. This commentary provides a summary of the discussion at the National Cancer Institute Workshop on Radiation, Senescence, and Cancer (August 10-11, 2020, National Cancer Institute, Bethesda, MD) regarding the current status of senescence research, heterogeneity of therapy-induced senescence, current status of senotherapeutics and molecular biomarkers, a concept of "one-two punch" cancer therapy (consisting of therapeutics to induce tumor cell senescence followed by selective clearance of SnCs), and its integration with personalized adaptive tumor therapy. It also identifies key knowledge gaps and outlines future directions in this emerging field to improve treatment outcomes for cancer patients.

Abstract 2258: Germline variants associated with poorer disease prognosis in nasopharyngeal carcinoma

Nasopharyngeal carcinoma (NPC) is an endemic disease prevalent in East Asia and germline variants that confer risk for NPC have previously been identified. However, NPC patients have contrasting phenotypes at presentation that are associated with disparate risks of relapse. We hypothesize that there exist germline variants that are associated with poorer NPC disease trajectories. We performed whole exome sequencing (100x; Illumina NovaSeq) on a cohort of 795 NPC patients and probed the germline genomic landscape in a predominantly Chinese population (N=723; 90.9%). We performed univariable and multivariable - adjusting for known prognostic clinical parameters - Cox regression analyses to discover germline variants that were associated with disease-free survival (DFS). We performed joint genotyping of germline variants following GATK's best practices workflow. We next filtered out common (≥5% allele frequency in East Asian population, or ≥1% in others) and benign variants, retaining only variants classified as pathogenic (P)/likely pathogenic (LP)/variant of unknown significance (VUS) in ClinVar or are predicted to be pathogenic by Combined Annotation Dependent Depletion or Rare Exome Variant Ensemble Learner. For improved statistical power, statistical analyses (uni- and multi-variable Cox regression) were conducted at the gene level, so that contributions across all variants were aggregated for each gene. To reduce false discovery, we focused our analysis only on genes of interest: hallmark-of-cancer genes (COSMIC gene census), DNA damage and mismatch repair genes, as well as genes known to be associated with NPC risk. Genes with variants in fewer than 20 unique patients were also excluded from analysis. We observed germline variants in hallmark-of-cancer genes (FAT1 [N=133; 16.7%] and NOTCH1 [N=39; 4.9%]) and DNA repair genes (SRSF6 [N=32; 4.0%] and POLD1 [N=32; 4.0%]). Additionally, 230 (28.9%) and 180 (22.6%) patients had at least one P/LP/VUS variants in HLA-A and HLA-DRB1, consistent with known literature that variants in HLA genes confer increased NPC risk. Likewise, we observed variants in genes associated with NPC risk such as KMT2C (N=207; 26.0%), SYNE1 (N=151; 19.0%), ARID1A (N=52; 6.5%) and MST1R (N=34; 4.3%). Univariable tests of association identified variants in MMP19 (HR 2.24 with 95% confidence interval [1.30-3.87], P=0.004), MST1R (HR 2.01 [1.14-3.54], P=0.016), TGFBR2 (HR 1.85 [1.09-3.14], P=0.023), NEIL3 (HR 2.08 [1.09-3.94], P=0.025), ETV6 (HR 2.13 [1.09-4.16], P=0.028), FANCD2 (HR 2.11 [1.09-4.13], P=0.030), DDX6 (HD 2.26 [1.06-4.82], P=0.035) POLD1 (HR 1.91 [1.04-3.53], P=0.038) and RECQL4 (HR 1.92 [1.04-3.54], P=0.037) that were significantly associated with inferior DFS. Among them, MMP19, TGFBR2, FANCD2 and RECQL4 remained significant in a clinico-molecular multivariable model that comprised of age, sex, cTN-categories, EBV DNA titer and treatment (C-index = 0.71).

Citation Format: Adelene Y.L. Sim, Enya H.w. Ong, Luo Huang, Kar Perng Low, Dewi Susanti, Terence W.k. Tan, Kam Weng Fong, Yoke Lim Soong, Gek San Tan, Tony K.H. Lim, Joseph T.s. Wee, Darren W.t. Lim, N. Gopalakrishna Iyer, Jacqueline S.G. Hwang, Mohamed E. Abazeed, Jin-Xin Bei, Melvin L.k. Chua. Germline variants associated with poorer disease prognosis in nasopharyngeal carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2258.

Targeting the Epidermal Growth Factor Receptor in EGFR-Mutated Lung Cancer: Current and Emerging Therapies

Epidermal growth factor receptor-targeting tyrosine kinase inhibitors (EGFR TKIs) are the standard of care for patients with EGFR-mutated metastatic lung cancer. While EGFR TKIs have initially high response rates, inherent and acquired resistance constitute a major challenge to the longitudinal treatment. Ongoing work is aimed at understanding the molecular basis of these resistance mechanisms, with exciting new studies evaluating novel agents and combination therapies to improve control of tumors with all forms of EGFR mutation. In this review, we first provide a discussion of EGFR-mutated lung cancer and the efficacy of available EGFR TKIs in the clinical setting against both common and rare EGFR mutations. Second, we discuss common resistance mechanisms that lead to therapy failure during treatment with EGFR TKIs. Third, we review novel approaches aimed at improving outcomes and overcoming resistance to EGFR TKIs. Finally, we highlight recent breakthroughs in the use of EGFR TKIs in non-metastatic EGFR-mutated lung cancer.

NRG Oncology HN006: Randomized phase II/III trial of sentinel lymph node biopsy versus elective neck dissection for early-stage oral cavity cancer

TPS6093

Background: Since patients with early-stage oral cavity cancer (OCC; T1-2N0M0; AJCC 8th ed) have a 20-30% rate of occult nodal metastases despite clinical and radiographic assessment, standard of care treatment includes elective neck dissection (END). Many patients have comprehensive surgical management of the regional cervical nodal basin even though the majority of those necks (70-80%) will not contain disease. Assessment of draining first echelon lymph nodes by sentinel lymph node (SLN) biopsy (Bx), a less invasive surgical procedure, may provide an alternative to END, while potentially reducing morbidity and cost. A decisive clinical trial comparing SLN Bx versus END can focus the HNC clinical and research community and resources on establishing the standard of care for management of the neck in early-stage OCC. Methods: In order to address the efficacy of SLN Bx in this population, we recently activated an international multi-institutional phase II/III prospective trial randomizing patients to two surgical arms: SLN Bx and END. PET/CT is an integral imaging biomarker in this trial. A node-negative PET/CT study with central read is required before randomization. Patients with a positive PET/CT central result will remain in a registry to compare imaging findings with final neck pathology. Given the current evidence available regarding morbidity for SLN Bx versus END, the phase II will determine if patient-reported neck and shoulder function and related QOL at 6 months after surgery using the Neck Dissection Impairment Index (NDII) shows a signal of superiority of SLN Bx compared to END. A total of 228 randomized patients with negative PET/CT for potential evaluation of shoulder-related morbidity with difference in 6-month NDII scores (minimum important difference ³7.5; one-sided a = 0.10; 90% power) will serve as the “Go/No-Go” decision to move forward into phase III. The phase III portion is a non-inferiority (NI) trial with disease-free survival (DFS) as the primary endpoint (NI margin hazard ratio 1.34 based on a 5% absolute difference in 2-year DFS; one-sided alpha 0.05; 80% power, and an interim look for efficacy at 67% of the events based on an O’Brien-Fleming boundary). The NDII at 6 months after surgery is a hierarchical co-primary endpoint for the phase III. Target accrual of phase III is 618 PET/CT negative patients, including those randomized in phase II (297 DFS events required for the final analysis). In addition to radiotherapy and imaging credentialing, quality assurance will include central pathology review of all negative SLN Bx cases and surgeon credentialing through an education course and SLN Bx and END case review by the surgical co-chairs. A surgical quality assurance working group will review all trial SLN Bx and END outcomes. As of 02/15/21, 7 patients have been screened and 6 of the planned 228 randomized patients in phase II have been enrolled. Clinical trial information: NCT04333537.

Multivalent state transitions regulate the intratumoral composition of small cell lung carcinoma

e20587

Background: Small cell lung carcinoma (SCLC) is an aggressive, tobacco-associated tumor with neuroendocrine features characterized by rapid growth, metastatic progression, and initial response followed by almost invariable resistance to therapy. Studies to date have not resolved the extent that diverse transcriptional programs drive SCLC and contribute to its lethality. Methods: We combined one of the largest and most diverse inventories of patient-derived xenograft models of SCLC with an ex vivo culture system that maintains transcriptional fidelity with matched primary SCLC tumor to identify transcriptional state heterogeneity. Using the expression of the Ascl1, NeuroD1, and Yap1 as markers of well-conserved SCLC states, we developed a state-of-the-art fluorescent platform that can directly measure single-cell state transitions in a multi-layered ecosystem using tandemly integrated reporters. We modeled population dynamics using a discrete time Markov chain and directly measure single-cell state transitions. Results: We show significant cell-state heterogeneity in several SCLC primary tumors, patient-derived xenografts (PDX), and ex vivo cultures. These states comprise distinct subpopulations marked by the master regulatory transcription factors (TFs) Ascl1, NeuroD1, and Yap1. Ex vivo, the 3 TFs are associated with suspension aggregates of small neuroendocrine cells, pre-suspension (loosely adherent) aggregates, and large mesenchymal cells with visible cytoplasm and spindle-like membrane extensions, respectively. We have observed equilibria in cell-state proportions in SCLC tumors both in vivo (PDX) and ex vivo. In addition, we have shown that the “elasticity” of SCLC responses, measured as the extent of clinical response during chemotherapy followed by the time to relapse from the end of therapy, is dependent on tumor TF levels. These observations suggest that mechanistic modeling of intra-tumoral state dynamics is of high clinical relevance. Conclusions: Our integrative approach is poised to formulate and validate a unified model of cellular states and program diversity in SCLC. If successful, the characterization of malignant cell ontogenic programs, their plasticity, and the advancement of new therapies designed to combat plasticity by epigenetic reprogramming will create a new scientific canvas for the study of this highly lethal disease.

The mutational landscape of the sensitivity of cancer to ionizing radiation

3129

Background: The impact of common or rare gene mutations on the sensitivity of cancers to ionizing radiation remains largely unknown. We conducted a systematic, arrayed (single variant per well) profiling effort to identify gene mutations that alter cellular sensitivity to radiation and validated some of our findings using a clinical cohort of patients who received thoracic radiotherapy alone. Methods: Candidate mutations were prioritized on the basis of genotype-phenotype associations from our previously completed large-scale cancer cell line irradiation profiling study (doi: 10.1038/ncomms11428), location within conserved protein domains, and functional impact (MutationAssessor). We used site-directed mutagenesis to generate mutant clones (2 clones per variant) and transferred the ORFs into lentiviral vectors in SV40 lung primary immortalized cells (BEAS2B). For clinical validation, an IRB-approved study was used to identify patients treated with lung radiotherapy alone. 197 patients with primary (stage I–IV) or recurrent lung cancer and patients with other cancer types and solitary metastases or oligometastases to the lung were included. Death without evidence of local failure was treated as a competing event, and Fine and Gray regression modeling was used to examine potential predictors of local failure. Results: Over 600 cancer variants were tested in ̃1200 experimental replicates, comprising 91 genes. We identified known and new radioresistant and radiosensitive variants involved in several cellular functional categories including cellular signaling, cytoskeleton, cell cycle, apoptosis, DNA methylation, and DNA repair. Variants that conferred resistance in BEAS2B cells were significantly more likely to confer resistance in TERT-HU1 and NCI-H520 cells, suggesting that most functional variants are cellular context indifferent. Variants under somatic oncogenic selection (hotspot mutants) were significantly more likely to confer resistance to radiation. Several infrequent cancer variants ( < 1% prevalence in cancer), including those in ERBB3, SMAD4, TGFBR1, VHL, CTNNB1, and MAP2K1, conferred radiation resistance. Some genes (e.g. KEAP1) demonstrated significant intragenic allelic variation in the magnitude of conferred resistance and other genes (e.g. CTNNB1) displayed both resistance and sensitivity in a protein domain-dependent manner. KRAS (resistant; HR 2.23; P= 0.02) and CTNNB1 exon 3 (sensitive; HR 0.3; P = 0.04) mutants conferred resistance and sensitivity, respectively, to radiotherapy in our clinical cohort. Conclusions: We report on a large-scale profiling effort to identify mutant alleles that govern radiation survival. Our results reveal new insights into potentially actionable determinants of tumor sensitivity to radiotherapy and accelerate clinical validation of common and rare gene mutations that impact radiation sensitivity.

RNA expression and risk of venous thromboembolism in lung cancer

BACKGROUND

The propensity to develop venous thromboembolism (VTE) on the basis of individual tumor biological features remains unknown.

OBJECTIVES

We conducted a whole transcriptome RNA sequencing strategy, focusing on a single cancer type (lung cancer), to identify biomarkers of cancer-associated VTE.

METHODS

Twelve propensity-matched patients, 6 each with or without VTE, were identified from a prospective institutional review board-approved registry at the Cleveland Clinic with available tissue from surgical excision of a primary lung mass between 2010 and 2015. Patients were propensity matched based on age, sex, race, history of prior cancer, date of cancer diagnosis, stage, histology, number of lines of chemotherapy, and length of follow-up. RNA sequencing was performed on tumor tissue, and gene set enrichment analysis (GSEA) was performed on differentially expressed genes.

RESULTS

We identified 1037 genes with differential expression. In patients with VTE, 869 genes were overexpressed and 168 were underexpressed compared to patients without VTE. Of these, 276 overexpressed and 35 underexpressed were significantly different (Q < 0.05). GSEA revealed upregulation of genes in complement, inflammation, and KRAS signaling pathways in tumors from patients with VTE.

CONCLUSIONS

These differentially expressed genes and associated pathways provide biologic insights into cancer-associated VTE and may provide insignts to develop new risk stratification schemes, prevention, or treatment strategies.

The Implications of Genetic Testing on Radiation Therapy Decisions: A Guide for Radiation Oncologists

The advent of affordable and rapid next-generation DNA sequencing technology, along with the US Supreme Court ruling invalidating gene patents, has led to a deluge of germline and tumor genetic variant tests that are being rapidly incorporated into clinical cancer decision-making. A major concern for clinicians is whether the presence of germline mutations may increase the risk of radiation toxicity or secondary malignancies. Because scarce clinical data exist to inform decisions at this time, the American Society for Radiation Oncology convened a group of radiation science experts and clinicians to summarize potential issues, review relevant data, and provide guidance for adult patients and their care teams regarding the impact, if any, that genetic testing should have on radiation therapy recommendations. During the American Society for Radiation Oncology workshop, several main points emerged, which are discussed in this manuscript: (1) variants of uncertain significance should be considered nondeleterious until functional genomic data emerge to demonstrate otherwise; (2) possession of germline alterations in a single copy of a gene critical for radiation damage responses does not necessarily equate to increased risk of radiation-induced toxicity; (3) deleterious ataxia-telangiesctasia gene mutations may modestly increase second cancer risk after radiation therapy, and thus follow-up for these patients after indicated radiation therapy should include second cancer screening; (4) conveying to patients the difference between relative and absolute risk is critical to decision-making; and (5) more work is needed to assess the impact of tumor somatic alterations on the probability of response to radiation therapy and the potential for individualization of radiation doses. Data on radiosensitivity related to specific genetic mutations is also briefly discussed.

Modeling Cellular Response in Large-Scale Radiogenomic Databases to Advance Precision Radiotherapy

Radiotherapy is integral to the care of a majority of patients with cancer. Despite differences in tumor responses to radiation (radioresponse), dose prescriptions are not currently tailored to individual patients. Recent large-scale cancer cell line databases hold the promise of unravelling the complex molecular arrangements underlying cellular response to radiation, which is critical for novel predictive biomarker discovery. Here, we present RadioGx, a computational platform for integrative analyses of radioresponse using radiogenomic databases. We fit the dose-response data within RadioGx to the linear-quadratic model. The imputed survival across a range of dose levels (AUC) was a robust radioresponse indicator that correlated with biological processes known to underpin the cellular response to radiation. Using AUC as a metric for further investigations, we found that radiation sensitivity was significantly associated with disruptive mutations in genes related to nonhomologous end joining. Next, by simulating the effects of different oxygen levels, we identified putative genes that may influence radioresponse specifically under hypoxic conditions. Furthermore, using transcriptomic data, we found evidence for tissue-specific determinants of radioresponse, suggesting that tumor type could influence the validity of putative predictive biomarkers of radioresponse. Finally, integrating radioresponse with drug response data, we found that drug classes impacting the cytoskeleton, DNA replication, and mitosis display similar therapeutic effects to ionizing radiation on cancer cell lines. In summary, RadioGx provides a unique computational toolbox for hypothesis generation to advance preclinical research for radiation oncology and precision medicine. SIGNIFICANCE: The RadioGx computational platform enables integrative analyses of cellular response to radiation with drug responses and genome-wide molecular data. GRAPHICAL ABSTRACT: http://cancerres.aacrjournals.org/content/canres/79/24/6227/F1.large.jpg.See related commentary by Spratt and Speers, p. 6076.

Cellular and Genetic Determinants of the Sensitivity of Cancer to α-Particle Irradiation

Targeted α-particle-emitting radionuclides have great potential for the treatment of a broad range of cancers at different stages of progression. A platform that accurately measures cancer cellular sensitivity to α-particle irradiation could guide and accelerate clinical translation. Here, we performed high-content profiling of cellular survival following exposure to α-particles emitted from radium-223 (²²³Ra) using 28 genetically diverse human tumor cell lines. Significant variation in cellular sensitivity across tumor cells was observed. ²²³Ra was significantly more potent than sparsely ionizing irradiation, with a median relative biological effectiveness of 10.4 (IQR: 8.4-14.3). Cells that are the most resistant to γ radiation, such as Nrf2 gain-of-function mutant cells, were sensitive to α-particles. Combining these profiling results with genetic features, we identified several somatic copy-number alterations, gene mutations, and the basal expression of gene sets that correlated with radiation survival. Activating mutations in PIK3CA, a frequent event in cancer, decreased sensitivity to ²²³Ra. The identification of cellular and genetic determinants of sensitivity to ²²³Ra may guide the clinical incorporation of targeted α-particle emitters in the treatment of several cancer types. SIGNIFICANCE: These findings address limitations in the preclinical guidance and prediction of radionuclide tumor sensitivity by identifying intrinsic cellular and genetic determinants of cancer cell survival following exposure to α-particle irradiation.See related commentary by Sgouros, p. 5479.

An image-based deep learning framework for individualizing radiotherapy dose

Background

Radiotherapy continues to be delivered uniformly without consideration of individual tumor characteristics. To advance toward more precise treatments in radiotherapy, we queried the lung computed tomography (CT)-derived feature space to identify radiation sensitivity parameters that can predict treatment failure and hence guide the individualization of radiotherapy dose.

Methods

We used a cohort-based registry of 849 patients with cancer in the lung treated with high dose radiotherapy using stereotactic body radiotherapy. We input pre-therapy lung CT images into a multi-task deep neural network, Deep Profiler, to generate an image fingerprint that primarily predicts time to event treatment outcomes and secondarily approximates classical radiomic features. We validated our findings in an independent study population (n = 95). Deep Profiler was combined with clinical variables to derive iGray, an individualized dose that estimates treatment failure probability to be <5%.

Findings

Radiation treatments in patients with high Deep Profiler scores fail at a significantly higher rate than in those with low scores. The 3-year cumulative incidences of local failure were 20.3% (95% CI: 16.0-24.9) and 5.7% (95% CI: 3.5-8.8), respectively. Deep Profiler independently predicted local failure (hazard ratio 1.65, 95% 1.02-2.66, p = 0.04). Models that included Deep Profiler and clinical variables predicted treatment failures with a concordance index of 0.72 (95% CI: 0.67-0.77), a significant improvement compared to classical radiomics or clinical variables alone (p = <0.001 and <0.001, respectively). Deep Profiler performed well in an external study population (n = 95), accurately predicting treatment failures across diverse clinical settings and CT scanner types (concordance index = 0.77 [95% CI: 0.69-0.92]). iGray had a wide dose range (21.1-277 Gy, BED), suggested dose reductions in 23.3% of patients and can be safely delivered in the majority of cases.

Interpretation

Our results indicate that there are image-distinct subpopulations that have differential sensitivity to radiotherapy. The image-based deep learning framework proposed herein is the first opportunity to use medical images to individualize radiotherapy dose.

Drug resistance profiling of a new triple negative breast cancer patient-derived xenograft model

Background

Triple-negative breast cancer (TNBC) represents an aggressive subtype with limited therapeutic options. Experimental preclinical models that recapitulate their tumors of origin can accelerate target identification, thereby potentially improving therapeutic efficacy. Patient-derived xenografts (PDXs), due to their genomic and transcriptomic fidelity to the tumors from which they are derived, are poised to improve the preclinical testing of drug-target combinations in translational models. Despite the previous development of breast and TNBC PDX models, those derived from patients with demonstrated health-disparities are lacking.

Methods

We use an aggressive TNBC PDX model propagated in SCID/Beige mice that was established from an African-American woman, TU-BcX-2 K1, and assess its metastatic potential and drug sensitivities under distinct in vitro conditions. Cellular derivatives of the primary tumor or the PDX were grown in 2D culture conditions or grown in mammospheres 3D culture. Flow cytometry and fluorescence staining was used to quantify cancer stem cell-like populations. qRT-PCR was used to describe the mesenchymal gene signature of the tumor. The sensitivity of TU-BcX-2 K1-derived cells to anti-neoplastic oncology drugs was compared in adherent cells and mammospheres. Drug response was evaluated using a live/dead staining kit and crystal violet staining.

Results

TU-BcX-2 K1 has a low propensity for metastasis, reflects a mesenchymal state, and contains a large burden of cancer stem cells. We show that TU-BcX-2 K1 cells have differential responses to cytotoxic and targeted therapies in 2D compared to 3D culture conditions insofar as several drug classes conferred sensitivity in 2D but not in 3D culture, or cells grown as mammospheres.

Conclusions

Here we introduce a new TNBC PDX model and demonstrate the differences in evaluating drug sensitivity in adherent cells compared to mammosphere, or suspension, culture.

Electronic supplementary material

The online version of this article (10.1186/s12885-019-5401-2) contains supplementary material, which is available to authorized users.

Transcriptomic and Protein Analysis of Small-cell Bladder Cancer (SCBC) Identifies Prognostic Biomarkers and DLL3 as a Relevant Therapeutic Target

PURPOSE

Transcriptomic profiling can shed light on the biology of small-cell bladder cancer (SCBC), nominating biomarkers, and novel therapeutic targets.

EXPERIMENTAL DESIGN

Sixty-three patients with SCBC had small-cell histology confirmed and quantified by a genitourinary pathologist. Gene expression profiling was performed for 39 primary tumor samples, 1 metastatic sample, and 6 adjacent normal urothelium samples (46 total) from the same cohort. Protein levels of differentially expressed therapeutic targets, DLL3 and PDL1, and also CD56 and ASCL1, were confirmed by IHC. A SCBC PDX model was utilized to assess in vivo efficacy of DLL3-targeting antibody-drug conjugate (ADC).

RESULTS

Unsupervised hierarchical clustering of 46 samples produced 4 clusters that correlated with clinical phenotypes. Patients whose tumors had the most "normal-like" pattern of gene expression had longer overall survival (OS) compared with the other 3 clusters while patients with the most "metastasis-like" pattern had the shortest OS (P = 0.047). Expression of DLL3, PDL1, ASCL1, and CD56 was confirmed by IHC in 68%, 30%, 52%, and 81% of tissue samples, respectively. In a multivariate analysis, DLL3 protein expression on >10% and CD56 expression on >30% of tumor cells were both prognostic of shorter OS (P = 0.03 each). A DLL3-targeting ADC showed durable antitumor efficacy in a SCBC PDX model.

CONCLUSIONS

Gene expression patterns in SCBC are associated with distinct clinical phenotypes ranging from more indolent to aggressive disease. Overexpression of DLL3 mRNA and protein is common in SCBC and correlates with shorter OS. A DLL3-targeted ADC demonstrated in vivo efficacy superior to chemotherapy in a PDX model of SCBC.

Case study: patient-derived clear cell adenocarcinoma xenograft model longitudinally predicts treatment response

There has been little progress in the use of patient-derived xenografts (PDX) to guide individual therapeutic strategies. In part, this can be attributed to the operational challenges of effecting successful engraftment and testing multiple candidate drugs in a clinically workable timeframe. It also remains unclear whether the ancestral tumor will evolve along similar evolutionary trajectories in its human and rodent hosts in response to similar selective pressures (i.e., drugs). Herein, we combine a metastatic clear cell adenocarcinoma PDX with a timely 3 mouse x 1 drug experimental design, followed by a co-clinical trial to longitudinally guide a patient's care. Using this approach, we accurately predict response to first- and second-line therapies in so far as tumor response in mice correlated with the patient's clinical response to first-line therapy (gemcitabine/nivolumab), development of resistance and response to second-line therapy (paclitaxel/neratinib) before these events were observed in the patient. Treatment resistance to first-line therapy in the PDX is coincident with biologically relevant changes in gene and gene set expression, including upregulation of phase I/II drug metabolism (CYP2C18, UGT2A, and ATP2A1) and DNA interstrand cross-link repair (i.e., XPA, FANCE, FANCG, and FANCL) genes. A total of 5.3% of our engrafted PDX collection is established within 2 weeks of implantation, suggesting our experimental designs can be broadened to other cancers. These findings could have significant implications for PDX-based avatars of aggressive human cancers.

Abstract 984: Systematic annotation of genetic variants that determine sensitivity to radiation: A pan-cancer encyclopedia

Cancer genome sequencing efforts have revealed tumor-associated gene variants that can direct therapy, functionally significant variants that do not currently direct therapy or variants that are of unknown functional significance. The impact of all of these categories of variants on the efficacy of radiotherapy remains largely unknown. To gain insight into the genetic landscape of radioresistance, we previously established a high-throughput platform that measures cell line survival to radiation. We then used this technology to analyze the radiosensitivity of more than 500 cancer cell lines annotated by the Cancer Cell Line Encyclopedia and discovered gene sets and pathways that regulate radiation survival. Here, we used the results from this study to identify gene variants (missense mutations, short inserts/deletions, and premature stop codons) that are associated with either radiosensitivity or radioresistance and functionally annotate a multitude of these variants.

Candidate variants from more than 100 ORFs were prioritized on the basis of their location within conserved protein domains (UniProt), predicted functional impact (MutationAssessor), and variant allele frequency. We used site-directed mutagenesis to generate mutant clones and transferred the ORFs into lentiviral vectors for stable expression under a PGK promoter in both SV40 and hTERT immortalized cell lines. For a subset of variants, expression of the endogenous gene was repressed by CRISPRi to model loss of heterozygosity. The effect of individual variants on radiation response was assessed using our high-throughput platform and benchmarked against the canonical, cyto- and radio-protective NFE2L2 E79K gain-of-function mutation and validated by colony forming assay.

We successfully adapted our high-throughput platform to measure the radiosensitivity of individual variants expressed in immortalized cell lines. Integral survival was significantly correlated between our high-throughput platform and clonogenic survival measurements. We discovered radioresistant conferring variants in genes involved in cell cycle transition, apoptosis (intrinsic), and MEK/ERK signaling. In contrast, variants associated with radiation sensitivity were identified in genes with established roles in DNA damage response pathways as well as in putative DNA repair factors.

Determining the impact of gene variants remains a major obstacle in the implementation of personalized radiotherapy. Here, we report on a large-scale profiling effort to identify and classify mutant alleles that govern radiation survival across multiple tumor lineages. Our results reveal new insights into the mechanisms of cellular survival to radiation and genome maintenance.

Citation Format: Brian D. Yard, Aaron P. Petty, Mohamed E. Abazeed. Systematic annotation of genetic variants that determine sensitivity to radiation: A pan-cancer encyclopedia [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 984.

Abstract 3681: A biologic basis for locoregional failure in patients with oral cavity cancers

BACKGROUND: The standard treatment for patients with oral cavity cancer (OCC) with intermediate risk pathologic variables after surgery is adjuvant radiotherapy. Despite this, one-third of patients experience locoregional failure (LRF). Clinicopathologic prognostic models have not been able to identify subsets of patients at higher risk of failure in whom treatment intensification with the addition of systemic chemotherapy should be considered. We posited that gene expression-derived tumor taxonomies can predict treatment failures and therefore guide more nuanced clinical decision making. Herein, we report on a score model based on OCC gene expression characteristics that can be incorporated into risk stratification and treatment decisions.

METHODS: Formalin-fixed paraffin-embedded (FFPE) tissue samples from patients with intermediate risk OCC treated with surgery followed by radiation alone were subjected to quantitative nuclease protection and next-generation sequencing to measure gene expression (HTG Molecular EdgeSeq™). A subset of samples that had corresponding frozen tumor samples were profiled by RNAseq to validate the FFPE results. Patients were divided into two groups based on LRF. Differentially expressed genes were identified using the R limma package. 98 genes were selected on the basis of unadjusted P values and predicted biological impact, as measured by gene set enrichment results (GSEA) and resultant biological pathway scores. The Cancer Genome Atlas (TCGA) HNSCC dataset (n=521) was used to validate the prognostic performance of our gene set.

RESULTS: Of the 78 patients included in the study, 35% of patients had LRF. GSEA of the 98 genes demonstrated a role for DNA repair, oxidative phosphorylation, hypoxia and p53 pathways, indicating radiobiologic plausibility for a significant subset of the genes that constitute the score. The mean composite score was 0.42 for patients with LRF, and -0.19 for patients without LRF (P = 0.0002). The Kaplan-Meier estimates of progression free survival at 3 years for the 1st (high risk) and 4th quartile (low risk) groups were 0.65 (0.47 to 0.89; 95% CI) and 0.93 (0.82 to 1; 95% CI), respectively. On multivariate analysis, the composite score was the strongest predictor of LRF (P = 0.0073). Composite scores also strongly predicted for overall survival in the TCGA HNSCC dataset (P &lt; 0.01) and the Kaplan-Meier estimates of overall survival at 2 years for the 1st and 4th quartile groups were 0.55 (0.45 to 0.68; 95% CI) and 0.73 (0.63 to 1; 84% CI), respectively. Composite scores performed the best in patients with OCC (P = 0.033).

CONCLUSIONS: We developed a gene signature that predicts LRF in patients with intermediate risk OCC treated with surgery and adjuvant radiotherapy. Further validation on larger datasets are needed. This biomarker can potentially identify higher risk patients who should be considered for intensification strategies with the addition of systemic therapy.

Citation Format: Elif I. Sarihan, Brian B. Burkey, Joseph Scharpf, Robert Lorenz, Eric D. Lamarre, Brandon Prendes, Jessica L. Geiger, David J. Adelstein, Shlomo A. Koyfman, Mohamed E. Abazeed. A biologic basis for locoregional failure in patients with oral cavity cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3681.

Abstract 3211: Therapeutic effects of radiotherapy on cancer cell lines using RadioGx computational platform

Purpose: Radiotherapy (RT) is the foundation of curative treatment regimens for many cancer types and is often delivered with drugs to produce synergistic effects. The “precision medicine” paradigm in which genomic biomarkers guide therapeutic decisions has eluded RT. To support preclinical discovery of biomarker-directed RT dose and drug combinations, we built a computational platform for integrating and interrogating radiogenomics data sets.

Methods: RadioGx integrates in vitro cancer cell line survival data with multimodal molecular profiles from the NCI-60 and Cancer Cell Line Encyclopedia (CCLE) cell line panels. We fit published radiation response data from both clonogenic and high-throughput viability assays to established radiobiologic models of clonogenic survival. We assessed the concordance of radiation response profiles across different survival endpoints (i.e., surviving fraction at 2Gy [SF2] and Area Under the Curve of the radiobiologic model [AUC]). By comparing radiation response with transcriptomic data, we examined indicators of radiosensitivity at the pathway level. We integrated drug response data from Cancer Therapeutics Response Portal (CTRP v2) using an interface between RadioGx and our previously published pharmacogenomics platform, PharmacoGx. Using the oxygen modification factor in established radiobiologic models, we identified pathways that are enriched under hypoxic conditions.

Results: RadioGx includes 600 radiation dose response data, 511 gene expression data sets, and 504 drug dose response data for chemotherapeutic compounds. Goodness-of-fit (given by R2 value) of the linear-quadratic radiobiologic model ranged from 0.927 to 0.99 (median=0.998, n=57) for the clonogenic assay and from 0.043 to 0.99 (median=0.952, n=535) for the viability assay. The concordance between SF2 and AUC using the clonogenic assay and the viability assay was 0.79 (n=12) and 0.82 (n=15), respectively. Among cell lines with survival data from both assays (n=28) across two different studies, the concordance of SF2 and AUC was 0.68 and 0.62, respectively. By interrogating the 504 cell lines that were profiled using the viability assay, we found drugs that behaved similar to RT were enriched in three pharmacologic classes, namely, drugs that impact mitosis, DNA replication, or cytoskeleton. Pathway analysis of radiation response under hypoxia revealed two key DNA repair pathways, namely, DNA double-strand break repair by NHEJ and DNA damage-induced 14-3-3, to be enriched compared to normoxic conditions.

Conclusions: Our work is a major step towards analyzing preclinical models of radiation response using in vitro survival and transcriptomic data. RadioGx coupled with PharmacoGx lays the foundation for the identification of synergistic drug-RT combinations, which can further push the boundaries of translational research towards precision radiation medicine.

Citation Format: Venkata S. Manem, Meghan Lambie, Petr Smirnov, Mark Freeman, Victor Kofia, Mohamed E. Abazeed, Scott V. Bratman, Benjamin Haibe-Kains. Therapeutic effects of radiotherapy on cancer cell lines using RadioGx computational platform [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3211.

Abstract 2194: Subclonal variation and evolutionary dynamics of BRAF mutations in cancer

Tumors have genetically distinct subclones that compete for space and resources and differentially resist efforts to make them extinct. We studied the intratumoral heterogeneity of BRAF mutations across several cancer types. We identified BRAF driver mutations as predominately clonal in some cancer types (e.g. melanoma) and subclonal in others (e.g. lung adenocarcinoma). Clonality corresponded to the amplification of BRAF and prevalence of V600 mutations in each cancer type. We mathematically and experimentally modeled the propagation and selection of tumors containing BRAF mutations and determined that the speed of clonal sweeps were associated with the extent of activation of MAPK signaling pathway and BRAF copy number. Consistent with these findings, tumors with "hard" sweeps were more likely to respond to BRAF and/or MEK inhibitors. Furthermore, some PDX models treated with cytotoxic therapy underwent BRAF mutant subclone expansion over time and this effect is mitigated by inhibitors of BRAF and/or MEK. Treatment strategies based on subclone composition resulted in significantly improved tumor control. Our study uncovers patterns of distinct BRAF clonal evolutionary trajectories and advances therapeutic strategies on the basis of BRAF mutation identity and subclone composition that merit testing in patients.

Citation Format: Priyanka Gopal, Elif I. Sarihan, Mohamed E. Abazeed. Subclonal variation and evolutionary dynamics of BRAF mutations in cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 2194.

Abstract 1056: Cancer avatars are sensitive diagnosticians of the pleural space

The identification of malignant cells in pleural fluid has critical prognostic and therapeutic implications but is frequently a diagnostic challenge. The first step in determining the cause of a suspected malignant pleural effusion, thoracentesis, has an unsatisfying low sensitivity (~60-70%). Patient-derived xenografts (PDX) retain the principal characteristics of the tumor of origin. However, there has been little progress in the application of these models to guide diagnostic and staging strategies. We assessed the feasibility, cellular yield and comparative diagnostic accuracy of pleural-derived xenografts in patients with effusions.

An institutional review board-approved single-institution prospective registry of patients was used to identify patients undergoing diagnostic and/or therapeutic thoracentesis. Cells isolated from the pleural fluid of these patients were injected into NSG mice. A diagnosis of the cancer in the xenograft was confirmed by a staff pathologist with expertise in the correspondent cancer type.

All of the cases that were clinically positive for tumor cells in the pleural space demonstrated tumor engraftment. Samples taken from patients with known benign etiologies did not result in tumor engraftment. The developed xenografts provided ample tissue that can be made available for extensive molecular testing, even in cases where the cytological samples consisted of very few cells. Critically, two of 10 cases resulted in engraftment of the tumor despite no evidence of malignant cells on cytological examination, indicating a 20% improvement in the statistical power of pleural-based xenografting for the detection of malignancy.

Our study indicates that pleural-derived xenografts can enhance the sensitivity of malignant cell detection, reflecting the inherent limitations of cytological analyses of a few malignant cells. Moreover, the amplification of cellular material provides ample source material that can be used to satisfy the increasing demands for tissue-based molecular testing. To our knowledge, this is the first effort to effectively use patient-derived xenografts for the purpose of cancer staging and diagnosis.

Citation Format: Gwendolyn B. Kuzmishin, Priyanka Gopal, Mohamed E. Abazeed. Cancer avatars are sensitive diagnosticians of the pleural space [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 1056.

Collateral sensitivity networks reveal evolutionary instability and novel treatment strategies in ALK mutated non-small cell lung cancer

Drug resistance remains an elusive problem in cancer therapy, particularly for novel targeted therapies. Much work is focused upon the development of an arsenal of targeted therapies, towards oncogenic driver genes such as ALK-EML4, to overcome the inevitable resistance that develops over time. Currently, after failure of first line ALK TKI therapy, another ALK TKI is administered, though collateral sensitivity is not considered. To address this, we evolved resistance in an ALK rearranged non-small cell lung cancer line (H3122) to a panel of 4 ALK TKIs, and performed a collateral sensitivity analysis. All ALK inhibitor resistant cell lines displayed significant cross-resistance to all other ALK inhibitors. We then evaluated ALK-inhibitor sensitivities after drug holidays of varying length (1-21 days), and observed dynamic patterns of resistance. This unpredictability led us to an expanded search for treatment options, where we tested 6 further anti-cancer agents for collateral sensitivity among resistant cells, uncovering possibilities for further treatment, including cross-sensitivity to standard cytotoxic therapies, as well as Hsp90 inhibitors. Taken together, these results imply that resistance to targeted therapy in non-small cell lung cancer is highly dynamic, and also one where there are many opportunities to re-establish sensitivities where there was once resistance. Drug resistance in cancer inevitably emerges during treatment; particularly with novel targeted therapies, designed to inhibit specific molecules. A clinically-relevant example of this phenomenon occurs in ALK-positive non-small cell lung cancer, where targeted therapies are used to inhibit the ALK-EML4 fusion protein. A potential solution to this may lie in finding drug sensitivities in the resistant population, termed collateral sensitivities, and then using these as second-line agents. This study shows how the evolution of resistance in ALK-positive lung cancer is a dynamic process through time, one in which patterns of drug resistance and collateral sensitivity change substantially, and therefore one where temporal regimens, such as drug cycling and drug holidays may have great benefit.

Abstract P5-06-02: Androgen receptor signaling regulates survival to ionizing radiation in breast cancers

This abstract was withdrawn by the authors.

The Vps13p-Cdc31p complex is directly required for TGN late endosome transport and TGN homotypic fusion

VPS13 proteins are widely conserved in eukaryotes and associated with human neurodegenerative and neurodevelopmental diseases. De et al. describe the lipid specificity and structure of yeast Vps13p, providing insight into its role in both TGN late endosome transport and TGN homotypic fusion.

Yeast VPS13 is the founding member of a eukaryotic gene family of growing interest in cell biology and medicine. Mutations in three of four human VPS13 genes cause autosomal recessive neurodegenerative or neurodevelopmental disease, making yeast Vps13p an important structural and functional model. Using cell-free reconstitution with purified Vps13p, we show that Vps13p is directly required both for transport from the trans-Golgi network (TGN) to the late endosome/prevacuolar compartment (PVC) and for TGN homotypic fusion. Vps13p must be in complex with the small calcium-binding protein Cdc31p to be active. Single-particle electron microscopic analysis of negatively stained Vps13p indicates that this 358-kD protein is folded into a compact rod-shaped density (20 × 4 nm) with a loop structure at one end with a circular opening ∼6 nm in diameter. Vps13p exhibits ATP-stimulated binding to yeast membranes and specific interactions with phosphatidic acid and phosphorylated forms of phosphatidyl inositol at least in part through the binding affinities of conserved N- and C-terminal domains.

A Histologic Basis for the Efficacy of SBRT to the lung

PURPOSE

Stereotactic body radiation therapy (SBRT) is the standard of care for medically inoperable patients with early-stage NSCLC. However, NSCLC is composed of several histological subtypes and the impact of this heterogeneity on SBRT treatments has yet to be established.

METHODS

We analyzed 740 patients with early-stage NSCLC treated definitively with SBRT from 2003 through 2015. We calculated cumulative incidence curves using the competing risk method and identified predictors of local failure using Fine and Gray regression.

RESULTS

Overall, 72 patients had a local failure, with a cumulative incidence of local failure at 3 years of 11.8%. On univariate analysis, squamous histological subtype, younger age, fewer medical comorbidities, higher body mass index, higher positron emission tomography standardized uptake value, central tumors, and lower radiation dose were associated with an increased risk for local failure. On multivariable analysis, squamous histological subtype (hazard ratio = 2.4 p = 0.008) was the strongest predictor of local failure. Patients with squamous cancers fail SBRT at a significantly higher rate than do those with adenocarcinomas or NSCLC not otherwise specified, with 3-year cumulative rates of local failure of 18.9% (95% confidence interval [CI]: 12.7-25.1), 8.7% (95% CI: 4.6-12.8), and 4.1% (95% CI: 0-9.6), respectively.

CONCLUSION

Our results demonstrate an increased rate of local failure in patients with squamous cell carcinoma. Standard approaches for radiotherapy that demonstrate efficacy for a population may not achieve optimal results for individual patients. Establishing the differential dose effect of SBRT across histological groups is likely to improve efficacy and inform ongoing and future studies that aim to expand indications for SBRT.

A genetic basis for the variation in the vulnerability of cancer to DNA damage

Radiotherapy is not currently informed by the genetic composition of an individual patient's tumour. To identify genetic features regulating survival after DNA damage, here we conduct large-scale profiling of cellular survival after exposure to radiation in a diverse collection of 533 genetically annotated human tumour cell lines. We show that sensitivity to radiation is characterized by significant variation across and within lineages. We combine results from our platform with genomic features to identify parameters that predict radiation sensitivity. We identify somatic copy number alterations, gene mutations and the basal expression of individual genes and gene sets that correlate with the radiation survival, revealing new insights into the genetic basis of tumour cellular response to DNA damage. These results demonstrate the diversity of tumour cellular response to ionizing radiation and establish multiple lines of evidence that new genetic features regulating cellular response after DNA damage can be identified.

Characterizing genomic alterations in cancer by complementary functional associations

Systematic efforts to sequence the cancer genome have identified large numbers of relevant mutations and copy number alterations in human cancers; however, elucidating their functional consequences, and their interactions to drive or maintain oncogenic states, is still a significant challenge. Here we introduce REVEALER, a computational method that identifies combinations of mutually exclusive genomic alterations correlated with functional phenotypes, such as the activation or gene-dependency of oncogenic pathways or the sensitivity to a drug treatment. We use REVEALER to uncover complementary genomic alterations associated with the transcriptional activation of β-catenin and NRF2, MEK-inhibitor sensitivity, and KRAS dependency. REVEALER successfully identified both known and new associations demonstrating the power of combining functional profiles with extensive characterization of genomic alterations in cancer genomes.

Integrative radiogenomic profiling of squamous cell lung cancer

Radiotherapy is one of the mainstays of anticancer treatment, but the relationship between the radiosensitivity of cancer cells and their genomic characteristics is still not well defined. Here, we report the development of a high-throughput platform for measuring radiation survival in vitro and its validation in comparison with conventional clonogenic radiation survival analysis. We combined results from this high-throughput assay with genomic parameters in cell lines from squamous cell lung carcinoma, which is standardly treated by radiotherapy, to identify parameters that predict radiation sensitivity. We showed that activation of NFE2L2, a frequent event in lung squamous cancers, confers radiation resistance. An expression-based, in silico screen nominated inhibitors of phosphoinositide 3-kinase (PI3K) as NFE2L2 antagonists. We showed that the selective PI3K inhibitor, NVP-BKM120, both decreased NRF2 protein levels and sensitized NFE2L2 or KEAP1-mutant cells to radiation. We then combined results from this high-throughput assay with single-sample gene set enrichment analysis of gene expression data. The resulting analysis identified pathways implicated in cell survival, genotoxic stress, detoxification, and innate and adaptive immunity as key correlates of radiation sensitivity. The integrative and high-throughput methods shown here for large-scale profiling of radiation survival and genomic features of solid-tumor-derived cell lines should facilitate tumor radiogenomics and the discovery of genotype-selective radiation sensitizers and protective agents.

Direct binding of the Kex2p cytosolic tail to the VHS domain of yeast Gga2p facilitates TGN to prevacuolar compartment transport and is regulated by phosphorylation

The VHS domains of yeast Gga1p and Gga2p bind sites (GBSs) in the Kex2p and Vps10p cytosolic tails. Phosphorylation of Ser-780 in the Kex2p GBS enhances Kex2p transport from the TGN to the PVC and is induced by cell wall damage. Kex2p GBS function is shown by direct binding, cell-free transport, and in vivo assays for Kex2 localization.

Human Golgi-localized, γ-ear–containing, ADP-ribosylation factor–binding proteins (Ggas) bind directly to acidic dileucine sorting motifs in the cytosolic tails (C-tails) of intracellular receptors. Despite evidence for a role in recruiting ubiquitinated cargo, it remains unclear whether yeast Ggas also function by binding peptide-sorting signals directly. Two-hybrid analysis shows that the Gga1p and Gga2p Vps27, Hrs, Stam (VHS) domains both bind a site in the Kex2p C-tail and that the Gga2p VHS domain binds a site in the Vps10p C-tail. Binding requires deletion of an apparently autoinhibitory sequence in the Gga2p hinge. Ser₇₈₀ in the Kex2p C-tail is crucial for binding: an Ala substitution blocks but an Asp substitution permits binding. Biochemical assays using purified Gga2p VHS–GGA and TOM1 (GAT) and glutathione S-transferase–Kex2p C-tail fusions show that Gga2p binds directly to the Kex2p C-tail, with relative affinities Asp₇₈₀ > Ser₇₈₀ > Ala₇₈₀. Affinity-purified antibody against a peptide containing phospho-Ser­₇₈₀ recognizes wild-type Kex2p but not S₇₈₀A Kex2p, showing that Ser₇₈₀ is phosphorylated in vivo; phosphorylation of Ser₇₈₀ is up-regulated by cell wall–damaging drugs. Finally, mutation of Ser₇₈₀ alters trafficking of Kex2p both in vivo and in cell-free trans-Golgi network (TGN)–prevacuolar compartment (PVC) transport. Thus yeast Gga adaptors facilitate TGN–PVC transport by direct binding of noncanonical phosphoregulated Gga-binding sites in cargo molecules.

Direct binding of the Kex2p cytosolic tail to the VHS domain of yeast Gga2p facilitates TGN to prevacuolar compartment transport and is regulated by phosphorylation

Human Golgi-localized, γ-ear–containing, ADP-ribosylation factor–binding proteins (Ggas) bind directly to acidic dileucine sorting motifs in the cytosolic tails (C-tails) of intracellular receptors. Despite evidence for a role in recruiting ubiquitinated cargo, it remains unclear whether yeast Ggas also function by binding peptide-sorting signals directly. Two-hybrid analysis shows that the Gga1p and Gga2p Vps27, Hrs, Stam (VHS) domains both bind a site in the Kex2p C-tail and that the Gga2p VHS domain binds a site in the Vps10p C-tail. Binding requires deletion of an apparently autoinhibitory sequence in the Gga2p hinge. Ser780in the Kex2p C-tail is crucial for binding: an Ala substitution blocks but an Asp substitution permits binding. Biochemical assays using purified Gga2p VHS–GGA and TOM1 (GAT) and glutathione S-transferase–Kex2p C-tail fusions show that Gga2p binds directly to the Kex2p C-tail, with relative affinities Asp780> Ser780> Ala780. Affinity-purified antibody against a peptide containing phospho-Ser­780recognizes wild-type Kex2p but not S780A Kex2p, showing that Ser780is phosphorylated in vivo; phosphorylation of Ser780is up-regulated by cell wall–damaging drugs. Finally, mutation of Ser780alters trafficking of Kex2p both in vivo and in cell-free trans-Golgi network (TGN)–prevacuolar compartment (PVC) transport. Thus yeast Gga adaptors facilitate TGN–PVC transport by direct binding of noncanonical phosphoregulated Gga-binding sites in cargo molecules.

Yeast Golgi-localized, gamma-Ear-containing, ADP-ribosylation factor-binding proteins are but adaptor protein-1 is not required for cell-free transport of membrane proteins from the trans-Golgi network to the prevacuolar compartment

Golgi-localized, gamma-Ear-containing, ADP-ribosylation factor-binding proteins (GGAs) and adaptor protein-1 (AP-1) mediate clathrin-dependent trafficking of transmembrane proteins between the trans-Golgi network (TGN) and endosomes. In yeast, the vacuolar sorting receptor Vps10p follows a direct pathway from the TGN to the late endosome/prevacuolar compartment (PVC), whereas, the processing protease Kex2p partitions between the direct pathway and an indirect pathway through the early endosome. To examine the roles of the Ggas and AP-1 in TGN-PVC transport, we used a cell-free assay that measures delivery to the PVC of either Kex2p or a chimeric protein (K-V), in which the Vps10p cytosolic tail replaces the Kex2p tail. Either antibody inhibition or dominant-negative Gga2p completely blocked K-V transport but only partially blocked Kex2p transport. Deletion of APL2, encoding the beta subunit of AP-1, did not affect K-V transport but partially blocked Kex2p transport. Residual Kex2p transport seen with apl2Delta membranes was insensitive to dominant-negative Gga2p, suggesting that the apl2Delta mutation causes Kex2p to localize to a compartment that precludes Gga-dependent trafficking. These results suggest that yeast Ggas facilitate the specific and direct delivery of Vps10p and Kex2p from the TGN to the PVC and that AP-1 modulates Kex2p trafficking through a distinct pathway, presumably involving the early endosome.

Rabies prophylaxis for pregnant women

Rabies Prophylaxis for Women

Cell-free transport from the trans-golgi network to late endosome requires factors involved in formation and consumption of clathrin-coated vesicles

Transport between the trans-Golgi network (TGN) and late endosome represents a conserved, clathrin-dependent sorting event that separates lysosomal from secretory cargo molecules and is also required for localization of integral membrane proteins to the TGN. Previously, we reported a cell-free reaction that reconstitutes transport from the yeast TGN to the late endosome/prevacuolar compartment (PVC) and requires the PVC t-SNARE Pep12p. Here, we report that factors required both for formation of clathrin-coated vesicles at the TGN (the Chc1p clathrin heavy chain and the Vps1p dynamin homolog) and for vesicle fusion at the PVC (the Vps21p rab protein and Vps45p SM (Sec1/Munc18) protein) are required for cell-free transport. The marker for TGN-PVC transport, Kex2p, is initially present in a clathrin-containing membrane compartment that is competent for delivery of Kex2p to the PVC. A Kex2p chimera containing the cytosolic tail (C-tail) of the vacuolar protein sorting receptor, Vps10p, is also efficiently transported to the PVC. Antibodies against the Kex2p and Vps10p C-tails selectively block transport of Kex2p and the Kex2-Vps10p chimera. The requirements for factors involved in vesicle formation and fusion, the identification of the donor compartment as a clathrin-containing membrane, and the need for accessibility of C-tail sequences argue that the TGN-PVC transport reaction involves selective incorporation of TGN cargo molecules into clathrin-coated vesicle intermediates. Further biochemical dissection of this reaction should help elucidate the molecular requirements and hierarchy of events in TGN-to-PVC sorting and transport.

Cell-free reconstitution of transport from the trans-golgi network to the late endosome/prevacuolar compartment

Vesicle-mediated transport between the trans-Golgi network (TGN) and the late endosome/prevacuolar compartment (PVC) is an essential step in lysosomal/vacuolar biogenesis. In addition, localization of integral membrane proteins to the TGN requires continual cycles of vesicular transport between the TGN and endosomal compartments. Genetic and biochemical analyses in yeast have identified a variety of proteins required for TGN-to-PVC transport. However, the precise mechanisms of vesicle formation, transport, and fusion have not been fully elucidated. To study the steps of TGN-to-PVC transport in mechanistic detail, we have developed a cell-free assay to monitor delivery of the processing protease Kex2p from the TGN to PVC compartments containing a Kex2p substrate. Transport is time-, temperature-, and ATP-dependent and requires the t-SNARE Pep12p. Moreover, cell-free delivery of Kex2p to the PVC results in the co-integration of Kex2p into PVC membranes containing the Kex2p substrate as determined by co-immunoisolation of Kex2p and the substrate using antibody against the Kex2p cytosolic tail. This work represents the first cell-free reconstitution and biochemical analysis of the essential vacuolar/lysosomal sorting step TGN to late endosome transport.

Bcl10 and MALT1, independent targets of chromosomal translocation in malt lymphoma, cooperate in a novel NF-kappa B signaling pathway

At least two distinct recurrent chromosomal translocations have been implicated in the pathogenesis of MALT lymphoma. The first, t(1;14), results in the transfer of the entire Bcl10 gene to chromosome 14 wherein Bcl10 expression is inappropriately stimulated by the neighboring Ig enhancer. The second, t(11;18), results in the synthesis of a novel fusion protein, API2-MALT1. Until now, no common mechanism of action has been proposed to explain how the products of these seemingly unrelated translocations may contribute to the same malignant process. We show here that Bcl10 and MALT1 form a strong and specific complex within the cell, and that these proteins synergize in the activation of NF-kappaB. The data support a mechanism of action whereby Bcl10 mediates the oligomerization and activation of the MALT1 caspase-like domain. This subsequently activates the IKK complex through an unknown mechanism, setting in motion a cascade of events leading to NF-kappaB induction. Furthermore, the API2-MALT1 fusion protein also strongly activates NF-kappaB and shows dependence upon the same downstream signaling factors. We propose a model whereby both the Bcl10.MALT1 complex and the API2-MALT1 fusion protein activate a common downstream signaling pathway that originates with the oligomerization-dependent activation of the MALT1 caspase-like domain.