A deep learning-informed interpretation of why and when dose metrics outside the PTV can affect the risk of distant metastasis in SBRT NSCLC patients.

PURPOSE: Recent papers suggested a correlation between the risk of distant metastasis (DM) and dose outside the PTV, though conclusions in different publications conflicted. This study resolves these conflicts and provides a compelling explanation of prognostic factors. MATERIALS AND METHODS: A dataset of 478 NSCLC patients treated with SBRT (IMRT or VMAT) was analyzed. We developed a deep learning model for DM prediction and explainable AI was used to identify the most significant prognostic factors. Subsequently, the prognostic power of the extracted features and clinical details were analyzed using conventional statistical methods. RESULTS: Treatment technique, tumor features, and dosiomic features in a 3 cm wide ring around the PTV (PTV3cm) were identified as the strongest predictors of DM. The Hazard Ratio (HR) for Dmean,PTV3cm was significantly above 1 (p < 0.001). There was no significance of the PTV3cm dose after treatment technique stratification. However, the dose in PTV3cm was found to be a highly significant DM predictor (HR > 1, p = 0.004) when analyzing only VMAT patients with small and spherical tumors (i.e., sphericity > 0.5). CONCLUSIONS: The main reason for conflicting conclusions in previous papers was inconsistent datasets and insufficient consideration of confounding variables. No causal correlation between the risk of DM and dose outside the PTV was found. However, the mean dose to PTV3cm can be a significant predictor of DM in small spherical targets treated with VMAT, which might clinically imply considering larger PTV margins for smaller, more spherical tumors (e.g., if IGTV > 2 cm, then margin ≤ 7 mm, else margin > 7 mm).

Improved outcome models with denoising diffusion.

PURPOSE: Radiotherapy outcome modelling often suffers from class imbalance in the modelled endpoints. One of the main options to address this issue is by introducing new synthetically generated datapoints, using generative models, such as Denoising Diffusion Probabilistic Models (DDPM). In this study, we implemented DDPM to improve performance of a tumor local control model, trained on imbalanced dataset, and compare this approach with other common techniques. METHODS: A dataset of 535 NSCLC patients treated with SBRT (50 Gy/5 fractions) was used to train a deep learning outcome model for tumor local control prediction. The dataset included complete treatment planning data (planning CT images, 3D planning dose distribution and patient demographics) with sparsely distributed endpoints (6-7 % experiencing local failure). Consequently, we trained a novel conditional 3D DDPM model to generate synthetic treatment planning data. Synthetically generated treatment planning datapoints were used to supplement the real training dataset and the improvement in the model's performance was studied. Obtained results were also compared to other common techniques for class imbalanced training, such as Oversampling, Undersampling, Augmentation, Class Weights, SMOTE and ADASYN. RESULTS: Synthetic DDPM-generated data were visually trustworthy, with Fréchet inception distance (FID) below 50. Extending the training dataset with the synthetic data improved the model's performance by more than 10%, while other techniques exhibited only about 4% improvement. CONCLUSIONS: DDPM introduces a novel approach to class-imbalanced outcome modelling problems. The model generates realistic synthetic radiotherapy planning data, with a strong potential to increase performance and robustness of outcome models.

Influence of relative binding affinity on efficacy in a panel of anti-CD3 scFv immunotoxins.

The in vitro cell killing potency of an immunotoxin reflects the aggregate of several independent biochemical properties. These include antigen binding affinity; internalization rate, intracellular processing and intrinsic toxin domain potency. This study examines the influence of antigen binding affinity on potency in various immunotoxin fusion proteins where target antigen binding is mediated by single chain antibody variable region fragments (scFv). Firstly, the relationship between affinity and potency was examined in a panel of four scFv immunotoxins generated from different anti-CD3 monoclonal antibodies fused to the 38 kDa fragment of Pseudomonas aeruginosa exotoxin A (PE38). Of these four scFv-PE38 immunotoxins, the one derived from the anti-CD3 monoclonal antibody UCHT1 has highest cell killing potency. Analysis of these four scFv-PE38 immunotoxins indicated a correlation between antigen binding affinity and immunotoxin potency in the cell killing assay with the exception of the scFvPE38 immunotoxin derived from the antibody BC3. However this scFv appeared to suffer a greater drop in affinity ( approximately 100x), relative to the parent Mab than did the other three scFvs used in this study (2-10x). Secondly, the scFv(UCHT1)-PE38 immunotoxin was then compared with a further panel of scFv(UCHT1)-derived immunotoxins including a divalent PE38 version and both monovalent and divalent Corynebacterium diphtheriae toxin (DT389) fusion proteins. When the scFv-UCHT1 domain was amino-terminally positioned relative to the toxin, as in the scFv(UCHT1)-PE38, an approximately 10-fold higher antigen-binding affinity was observed than with the C-terminal fusion, used in the DT389-scFv(UCHT1) molecule. Despite this lower antigen-binding activity, the DT389-scFv immunotoxin had a 60-fold higher potency in the T-cell-killing assay. Thirdly, a divalent form of the DT389-scFv construct, containing tandem scFv domains, had a 10-fold higher binding activity, which was exactly reflected in a 10-fold increase in potency. Therefore, when comparing immunotoxins in which scFvs from different antibodies are fused to the same toxin domain (DT or PE) a broad correlation appears to exist between binding affinity and immunotoxin potency. However, no correlation between affinity and potency appears to exist when different toxin domains are combined with the same scFv antibody domain.

A high-capacity alkaline phosphatase reporter system for the rapid analysis of specificity and relative affinity of peptides from phage-display libraries.

We describe a novel reporter enzyme cassette system which enables the analysis of large numbers of linear and cyclic peptides in terms of their binding to a specific target molecule. In this system, peptides selected for target binding from random peptide phage-display libraries are expressed as cloned fusion proteins with bacterial alkaline phosphatase. The binding specificity and relative affinity of each peptide-enzyme fusion protein is then evaluated in a target-specific ELISA. This strategy enables direct identification of the highest affinity peptides, specific for a given target, which can then be sequenced at the DNA level to derive their peptide sequences. This eliminates the need to sequence large numbers of clones to establish consensus sequences for binding peptides. This approach also eliminates the need for peptide synthesis or phage ELISA to determine relative binding affinities, which can be technically difficult. Identification of binding peptides based on specificity and relative affinity, rather than conforming to an amino acid consensus sequence, enables the rapid evaluation of hundreds of candidate peptides and identification of rare (non-consensus) binding peptides which may otherwise be missed.

Hepatitis C in the HIV (human immunodeficiency virus) Atlanta V.A. (Veterans Affairs Medical Center) Cohort Study (HAVACS): the effect of coinfection on survival.

To examine the prevalence of and survival rates for coinfection with human immunodeficiency virus (HIV) and hepatitis C virus (HCV), data were analyzed from all HIV-infected patients tested for HCV antibody from January 1992 until May 1997. The prevalence of HCV infection among 350 HIV-infected patients was 33%. By univariate analysis, HCV-positive (HCV+) patients were more likely to be older (P = .003), be positive for hepatitis B core antibody (P = .006), be black (P = .001), be intravenous drug users (P = .001), and have an abnormal level of aspartate aminotransferase (AST) (P = .001). In a logistic regression model, only intravenous drug abuse and abnormal AST level remained independently associated with HCV positivity. Length of survival, as determined by the Cox proportional hazards model, was similar for HCV+ vs. HCV- patients when analyzed for three different endpoints: time from diagnosis of HIV to diagnosis of AIDS, time from diagnosis of HIV to death, and time from diagnosis of AIDS to death. The prevalence of HCV infection in this population is high but does not appear to affect HIV progression or survival.

Optimization of the anti-(human CD3) immunotoxin DT389-scFv(UCHT1) N-terminal sequence to yield a homogeneous protein.

The production and regulatory approval processes for biopharmaceuticals require detailed characterization of potential products. Therapeutic proteins should preferably be homogeneous, although limited, reproducible, heterogeneity may be tolerated. A diphtheria toxin-based anti-(human CD3) immunotoxin, DT389-scFv(UCHT1), was expressed in Escherichia coli and purified following refolding [DT389 corresponds to amino acids 1-389 of diphtheria toxin, scFv is single-chain variable-region antibody fragment and UCHT1is an anti-(human CD3) monoclonal antibody]. Biochemical characterization of this molecule by MS and N-terminal sequencing by Edman degradation revealed that the protein was heterogeneous at the N-terminus, containing species both with (60%) and without (40%) the initiator methionine residue. In an attempt to generate an N-terminally homogeneous molecule, a panel of seven N-terminal variants was designed, based on the published specificity of bacterial methionine aminopeptidase. Following bacterial expression, partial purification and separation on SDS/PAGE, these proteins were subjected to N-terminal sequencing by Edman degradation. Three of the mutants yielded a 100% homogeneous amino acid sequence. By contrast, the original DT389-scFv(UCHT1) protein and four variant proteins yielded two sequences with varying ratios corresponding to species with and without methionine. The N-terminal sequences of the three homogeneous clones were MLADD and MLDD, where the methionine was completely retained, and SADD, where the methionine was completely removed. One of the homogeneous mutants (SADD) was expressed, refolded and purified and found to be equipotent with the parent immunotoxin. Thus, using a rational mutagenesis approach, three N-terminally homogeneous variants of DT389-scFv(UCHT1) have been identified, at least one of which is functionally indistinguishable from the parent immunotoxin. This approach is generally applicable to biopharmaceutical production and immunotoxin development in particular.

Expression, purification, and characterization of the human squalene synthase: use of yeast and baculoviral systems.

We have cloned and utilized a cDNA corresponding to the human squalene synthase gene to generate active enzyme from yeast and baculoviral expression systems. Expression of human squalene synthase in yeast resulted in production of active enzyme in cellular lysates. The presence of the active human enzyme, however, was insufficient to rescue growth of spores defective in yeast squalene synthase function, suggesting that structural differences in the yeast and human enzymes may affect localization or folding of the protein. Expression of the human enzyme in Sf-9 insect cells after infection with recombinant baculovirus encoding the human squalene synthase gene resulted in detection of substantial enzymatic activity in cell lysate preparations. Following extraction from the Sf-9 cells, the human enzyme was purified to near homogeneity utilizing a series of ion-exchange chromatography steps with an overall yield of purified protein of approximately 5 mg per liter of Sf-9 cell culture. The purified enzyme was characterized through steady-state kinetic and physical measurements and the kinetic constants are consistent with values observed for other squalene synthases. Zaragozic acid C was found to be a competitive inhibitor with respect to farnesyl pyrophosphate and has a Kis value of 250 pM (@ [NADPH] = 5 mM). Inhibition experiments with zaragozic acid C at low (approximately 0.5 x Km) and high (approximately 10 x Km) concentrations of NADPH indicated that the inhibitor does not bind in the enzyme's NADPH binding domain. These studies demonstrate that the human enzyme can be prepared from baculovirus-infected Sf-9 cells in a catalytically active configuration and in sufficient quantities to allow for further biochemical, kinetic, and structural characterization.