Citations

Though specific growth rate (SGR) has potential prognostic value for oropharyngeal squamous cell carcinoma (OPSCC), there is sparse literature defining these rates. Our aims were to establish the SGRs of primary tumors (PTs) and lymph nodes (LNs) in OPSCC and to correlate SGR with oncologic outcome. A pilot study was designed with a retrospective analysis examining 54 patients from the University of California, Davis with OPSCC (diagnosed 2012–2019). Radiation oncology software and pretreatment serial CT scans were used to measure PT and LN volumes to calculate SGR and doubling time (DT). The mean PT-SGR was 1.2 ± 2.2%/day and the mean LN-SGR was 1.6 ± 1.9%/day. There was no statistically significant difference between slow-growing and fast-growing cohorts in terms of age, gender, smoking status, tumor subsite, HPV status (as determined with p16 staining), initial volume, or overall stage. SGR had no impact on 2-year overall survival, disease-free survival, or disease-specific survival. We found the average daily growth rates for OPSCC to be 1.2%/day and 1.6%/day. Our findings suggest PT- and LN-SGR are independent factors, not heavily influenced by known biomarkers and patient characteristics, without a statistical impact on prognosis. This information has value in patient counseling regarding tumor growth and in providing patients worried about fast-growing tumors the appropriate reassurance.

In this study, we investigated whether radiomics features from pre-treatment positron emission tomography (PET) images could be used to predict disease progression in patients with HPV-positive oropharyngeal cancer treated with definitive proton or x-ray radiotherapy. Machine learning models were built using a dataset from Mayo Clinic, Rochester, Minnesota (n = 72) and tested on a dataset from Mayo Clinic, Phoenix, Arizona (n = 22). A total of 71 clinical and radiomics features were considered. The Mann–Whitney U test was used to identify the top 2 clinical and top 20 radiomics features that were significantly different between progression and progression-free patients. Two dimensionality reduction methods were used to define two feature sets (manually filtered or machine-driven). A forward feature selection scheme was conducted on each feature set to build models of increased complexity (number of input features from 1 to 6) and evaluate model robustness and overfitting. The machine-driven features had superior performance and were less prone to overfitting compared to the manually filtered features. The four-variable Gaussian Naïve Bayes model using the ‘Radiation Type’ clinical feature and three machine-driven features achieved a training accuracy of 79% and testing accuracy of 77%. These results demonstrate that radiomics features can provide risk stratification beyond HPV-status to formulate individualized treatment and follow-up strategies.

AbstractBackgroundDeep learning auto‐segmentation (DLAS) models have been adopted in the clinic; however, they suffer from performance deterioration owing to the clinical practice variability. Some commercial DLAS software provide an incremental retraining function that enables users to train a custom model using their institutional data to account for clinical practice variability.PurposeThis study was performed to evaluate and implement the commercial DLAS software with the incremental retraining function for definitive treatment of patients with prostate cancer in a multi‐user environment.MethodsCT‐based target organs and organs‐at‐risk (OAR) delineation of 215 prostate cancer patients were utilized. The performance of three commercial DLAS software built‐in models was validated with 20 patients. A retrained custom model was developed using 100 patients and evaluated on the remaining data (n = 115). Dice similarity coefficient (DSC), Hausdorff distance (HD), mean surface distance (MSD), and surface DSC (SDSC) were utilized for quantitative evaluation. A multi‐rater qualitative evaluation was blindly performed with a five‐level scale. Visual inspection was performed in consensus and non‐consensus unacceptable cases to identify the failure modes.ResultsThree commercial DLAS vendor built‐in models achieved sub‐optimal performance in 20 patients. The retrained custom model had a mean DSC of 0.82 for prostate, 0.48 for seminal vesicles (SV), and 0.92 for rectum, respectively. This represents a significant improvement over the built‐in model with DSC of 0.73, 0.37, and 0.81 for the corresponding structures. Compared to the acceptance rate of 96.5% and consensus unacceptable rate (i.e., both reviewers rated as unacceptable) of 3.5% achieved by manual contours, the custom model achieved a 91.3% acceptance rate and 8.7% consensus unacceptable rate. The failure modes of retrained custom model were attributed to the following: cystogram (n = 2), hip prosthesis (n = 2), low dose rate brachytherapy seeds (n = 2), air in endorectal balloon(n = 1), non‐iodinated spacer (n = 2), and giant bladder(n = 1).ConclusionThe commercial DLAS software with the incremental retraining function was validated and clinically adopted for prostate patients in a multi‐user environment. AI‐based auto‐delineation of the prostate and OARs is shown to achieve improved physician acceptance, overall clinical utility, and accuracy.

7059 Background: TP53-mutated ( TP53mut) myeloid neoplasms (MN) are aggressive leukemia with poor survival. A subset of TP53mut MN is preceded by TP53mut CCUS that is conventionally considered to be high-risk premalignant condition, though evidence supporting the notion is lacking. We studied outcomes of TP53mut CCUS compared to TP53 wild-type ( TP53wt) CCUS. Methods: Mayo Clinic Molecular Hematology Database (N=7593) was queried to identify patients (pts) harboring pathogenic variant(s) in exons 4-11 of the TP53 gene with ≥2% variance allele frequency (VAF). CCUS was defined using the 5th edition of the WHO Classification for MN. Results: Of 457 pts with TP53mut, 29 (5.9%) had TP53mut CCUS. The median age at diagnosis was 67 years, 17 (59%) were female, and 20 (69%) had received cytotoxic therapies. One and ≥2 TP53mut were seen in 26 (89.6%) and 3 (10.3%) pts, respectively. Median TP53mut VAF was 9% (range 3-42), 3 had TP53mut VAF ≥20%, and 4 (14%) had non-MDS defining cytogenetic anomalies. Of 16 pts with available follow up evaluation (median interval 22 months, Table), 3 (10.3%) progressed to TP53mut MN at 5, 13, and 14 months from diagnosis. We next compared outcomes of TP53mut CCUS to TP53wt CCUS (n=138). The proportion of pts developing MN was not different between TP53wt and TP53mut CCUS (18.1 vs 10.3%, P= .42). Progression free survival of TP53mut CCUS was comparable to TP53wt CCUS at 1- (77 vs 83%), 2- (67 vs 67%), and 5-years (67 vs 59%, P=.9). Similarly, overall survival of TP53mut vs TP53wt CCUS at 1- (81 vs 89%), 2- (76 vs 76%), and 5-years (76 vs 67%) was comparable ( P= .94). Conclusions: In contrast to the prevalent notion, TP53mut was not associated with a higher risk of MN progression or mortality compared to TP53wt CCUS. Majority of TP53mut clones showed remarkable stability over years. [Table: see text]

AbstractETV6 mutations are rare but recurrent somatic events in myeloid neoplasms and are negatively prognostic in myelodysplastic syndrome. We set out to examine the clinical and molecular characteristics of patients undergoing investigation for myeloid neoplasms, found to have deleterious ETV6 mutations. ETV6 mutations occurred in 33 of 5793 (0.6%) cases investigated and predominantly in high‐risk disease entities including MDS with increased blasts, primary myelofibrosis and AML, myelodysplasia‐related. In three cases, isolated iso (17q) karyotype was concurrently detected, an otherwise rare karyotype in myeloid neoplasms. ETV6 mutations were frequently subclonal and never occurred as an isolated abnormality with ASXL1 (n = 22, 75%), SRSF2 (n = 14, 42%) and SETBP1 (n = 11, 33%) the predominant co‐mutations. Restricting to patients with MDS, higher rates of ASXL1, SETBP1, RUNX1 and U2AF1 mutations occurred in ETV6 mutated cases, relative to a consecutive control cohort with wild‐type ETV6. The median OS of the cohort was 17.5 months. This report highlights the clinical and molecular associations of somatic ETV6 mutations in myeloid neoplasms, suggests their occurrence as a later event, and proposes further translational research questions for their role in myeloid neoplasia.

Pathogenic variants in KMT5B , a lysine methyltransferase, are associated with global developmental delay, macrocephaly, autism, and congenital anomalies (OMIM # 617788). Given the relatively recent discovery of this disorder, it has not been fully characterized. Deep phenotyping of the largest ( n = 43) patient cohort to date identified that hypotonia and congenital heart defects are prominent features that were previously not associated with this syndrome. Both missense variants and putative loss-of-function variants resulted in slow growth in patient-derived cell lines. KMT5B homozygous knockout mice were smaller in size than their wild-type littermates but did not have significantly smaller brains, suggesting relative macrocephaly, also noted as a prominent clinical feature. RNA sequencing of patient lymphoblasts and Kmt5b haploinsufficient mouse brains identified differentially expressed pathways associated with nervous system development and function including axon guidance signaling. Overall, we identified additional pathogenic variants and clinical features in KMT5B -related neurodevelopmental disorder and provide insights into the molecular mechanisms of the disorder using multiple model systems.