OP0023 PREDICTION OF REMISSION FOR EACH BIOLOGICS BASED ON PATIENT’S CLINICAL INFORMATION BEFORE STARTING BIOLOGICS USING EXPLAINABLE ARTIFICIAL INTELLIGENCE: DATA FROM THE KOREAN COLLEGE OF RHEUMATOLOGY BIOLOGICS REGISTRY

Background:

Many studies have identified predictive factors of response to biologics in patients wirh rheumatoid arthritis (RA). However, there is still a lack in using them in daily clinical practice. Therefore, it is necessary to develop a method that can assist the physician in selecting effective biologics.

Objectives:

The purpose of this study is to establish machine learning model that predicts remission in patients treated with biologics using data of RA patients from the Korean College of Rheumatology Biologics (KOBIO) registry, and to identify the important features that have the most influence on the response to biologics using explainable artificial intelligence (AI).

Methods:

A total of 1,527 patients who started with biologics such as etanercept, adalimumab, golimumab, infliximab, abatacept, and tocilizumab from December 2012 to June 2019 were enrolled. Remission was predicted using 46 variables corresponding to baseline profiles at the starting of each biologics. We used five machine learning methods such as lasso, ridge, SVM, random forest, and XGBoost. For explainability of those models, we used Shapley plot to interpret the feature importance for each biologics.

Results:

In all machine learning methods, the accuracy and the area under the receiver operating characteristic (AUROC) were 57.2%~74.5%, 0.547~0.747, respectively (Table 1). The accuracy and AUROC of each biologics were similar between machine learning methods. Figure 2 showed interpretation of feature importance with the Shapley plot for remission. The most important feature was age in adalimumab (younger were closer to remission), daily corticosteroid dose in etanercept, golimumab, and all TNF inhibitors (using fewer doses daily were closer to remission), baseline erythrocyte sedimentation rate in infliximab (lower ESR were closer to remission), disease duration in abatacept (longer disease durations showed difficulty determining remission), baseline c-reactive protein in tocilizumab (higher CRP were closer to remission).

Table.

Predicting remission for all biologics in various machine learning method.

MeasureLassoRidgeSVMRandom ForestXGBoostNo info rateSampleAbataceptAccuracy74.1%74.1%70.6%71.8%68.8%70.6%216AUROC0.7250.7420.7070.6770.6470.500AdalimumabAccuracy73.6%72.0%70.4%72.0%70.4%68.8%315AUROC0.7100.7290.7000.6750.6630.500EtanerceptAccuracy72.0%72.0%70.0%71.5%70.0%68.0%250AUROC0.7410.7470.7260.7190.7040.500GolimumabAccuracy71.3%68.5%66.7%68.5%68.5%68.5%138AUROC0.7460.7270.7010.6900.6550.500InfliximabAccuracy72.8%73.5%67.6%73.5%69.1%72.5%172AUROC0.6630.6830.6160.5970.5270.500TNF inhibitorsAccuracy73.9%74.5%73.9%74.2%73.6%70.3%875AUROC0.7390.7410.7260.7470.7240.500TocilizumabAccuracy62.4%63.6%62.4%59.5%57.2%59.5%436AUROC0.6330.6400.6330.6150.5470.500Figure 2.

Shapley plots and SHAP values for the feature importance from clinical information in patients with RA.

Conclusion:

We developed machine learning models for predicting remission as a response to each biologics in active RA patients based on their clinical profiles, and found important clinical features using explainable AI. This approach may support clinical decisions to improve treatment outcomes in patients with RA.

Disclosure of Interests:

None declared

PIN1 Attenuation Improves Midface Hypoplasia in a Mouse Model of Apert Syndrome

Premature fusion of the cranial suture and midface hypoplasia are common features of syndromic craniosynostosis caused by mutations in the FGFR2 gene. The only treatment for this condition involves a series of risky surgical procedures designed to correct defects in the craniofacial bones, which must be performed until brain growth has been completed. Several pharmacologic interventions directed at FGFR2 downstream signaling have been tested as potential treatments for premature coronal suture fusion in a mouse model of Apert syndrome. However, there are no published studies that have targeted for the pharmacologic treatment of midface hypoplasia. We used Fgfr2^S252W/+ knock-in mice as a model of Apert syndrome and morphometric analyses to identify causal hypoplastic sites in the midface region. Three-dimensional geometric and linear analyses of Fgfr2^S252W/+ mice at postnatal day 0 demonstrated distinct morphologic variance. The premature fusion of anterior facial bones, such as the maxilla, nasal, and frontal bones, rather than the cranium or cranial base, is the main contributing factor toward the anterior-posterior skull length shortening. The cranial base of the mouse model had a noticeable downward slant around the intersphenoid synchondrosis, which is related to distortion of the airway. Within a skull, the facial shape variance was highly correlated with the cranial base angle change along Fgfr2 S252W mutation-induced craniofacial anomalies. The inhibition of an FGFR2 downstream signaling enzyme, PIN1, via genetic knockdown or use of a PIN1 inhibitor, juglone, attenuated the aforementioned deformities in a mouse model of Apert syndrome. Overall, these results indicate that FGFR2 signaling is a key contributor toward abnormal anterior-posterior dimensional growth in the midface region. Our study suggests a novel therapeutic option for the prevention of craniofacial malformations induced by mutations in the FGFR2 gene.

Localized Metabolomic Gradients in Patient-Derived Xenograft Models of Glioblastoma

Glioblastoma (GBM) is increasingly recognized as a disease involving dysfunctional cellular metabolism. GBMs are known to be complex heterogeneous systems containing multiple distinct cell populations and are supported by an aberrant network of blood vessels. A better understanding of GBM metabolism, its variation with respect to the tumor microenvironment, and resulting regional changes in chemical composition is required. This may shed light on the observed heterogeneous drug distribution, which cannot be fully described by limited or uneven disruption of the blood-brain barrier. In this work, we used mass spectrometry imaging (MSI) to map metabolites and lipids in patient-derived xenograft models of GBM. A data analysis workflow revealed that distinctive spectral signatures were detected from different regions of the intracranial tumor model. A series of long-chain acylcarnitines were identified and detected with increased intensity at the tumor edge. A 3D MSI dataset demonstrated that these molecules were observed throughout the entire tumor/normal interface and were not confined to a single plane. mRNA sequencing demonstrated that hallmark genes related to fatty acid metabolism were highly expressed in samples with higher acylcarnitine content. These data suggest that cells in the core and the edge of the tumor undergo different fatty acid metabolism, resulting in different chemical environments within the tumor. This may influence drug distribution through changes in tissue drug affinity or transport and constitute an important consideration for therapeutic strategies in the treatment of GBM. SIGNIFICANCE: GBM tumors exhibit a metabolic gradient that should be taken into consideration when designing therapeutic strategies for treatment.See related commentary by Tan and Weljie, p. 1231.

Accurate Patient-Specific Machine Learning Models of Glioblastoma Invasion Using Transfer Learning

BACKGROUND AND PURPOSE

MR imaging-based modeling of tumor cell density can substantially improve targeted treatment of glioblastoma. Unfortunately, interpatient variability limits the predictive ability of many modeling approaches. We present a transfer learning method that generates individualized patient models, grounded in the wealth of population data, while also detecting and adjusting for interpatient variabilities based on each patient's own histologic data.

MATERIALS AND METHODS

We recruited patients with primary glioblastoma undergoing image-guided biopsies and preoperative imaging, including contrast-enhanced MR imaging, dynamic susceptibility contrast MR imaging, and diffusion tensor imaging. We calculated relative cerebral blood volume from DSC-MR imaging and mean diffusivity and fractional anisotropy from DTI. Following image coregistration, we assessed tumor cell density for each biopsy and identified corresponding localized MR imaging measurements. We then explored a range of univariate and multivariate predictive models of tumor cell density based on MR imaging measurements in a generalized one-model-fits-all approach. We then implemented both univariate and multivariate individualized transfer learning predictive models, which harness the available population-level data but allow individual variability in their predictions. Finally, we compared Pearson correlation coefficients and mean absolute error between the individualized transfer learning and generalized one-model-fits-all models.

RESULTS

Tumor cell density significantly correlated with relative CBV (r = 0.33, P < .001), and T1-weighted postcontrast (r = 0.36, P < .001) on univariate analysis after correcting for multiple comparisons. With single-variable modeling (using relative CBV), transfer learning increased predictive performance (r = 0.53, mean absolute error = 15.19%) compared with one-model-fits-all (r = 0.27, mean absolute error = 17.79%). With multivariate modeling, transfer learning further improved performance (r = 0.88, mean absolute error = 5.66%) compared with one-model-fits-all (r = 0.39, mean absolute error = 16.55%).

CONCLUSIONS

Transfer learning significantly improves predictive modeling performance for quantifying tumor cell density in glioblastoma.

Large-area fabrication of microlens arrays by using self-pinning effects during the thermal reflow process

Generally, the fabrication of curved structures such as microlens arrays has been regarded as an expensive and complicated process. Here, we propose a facile method to form a microlens array with controlled lens curvature by combining residue-free nanoimprint lithography (NIL) with V-shaped molds and the successive thermal reflow procedure of the printed polymeric structures. The V-shaped molds used in this study enable the bottom substrate to be exposed after the NIL process when the initial thickness is controlled. Then, we use the thermal reflow to realize hemi-cylindrical curved lenses by applying heat. The polymers are self-pinned on the exposed substrate, which is strong enough to fix the boundary to not dewet or be flattened in the broad temperature range of the reflow process, which is essential for a large-area fabrication. Furthermore, we demonstrate the modulation of the focal lengths of the lenses by controlling the initial polymer thickness coated on a substrate.

Antimicrobial activities of an oxygenated cyclohexanone derivative isolated from Amphirosellinia nigrospora JS-1675 against various plant pathogenic bacteria and fungi

AIMS

To evaluate the antimicrobial activities of an active compound isolated from the culture broth of Amphirosellinia nigrospora JS-1675 against various plant pathogenic bacteria and fungi.

METHODS AND RESULTS

While screening for bioactive secondary metabolites from endophytic fungi, we found that A. nigrospora JS-1675 showed strong in vitro antibacterial activity against Ralstonia solanacearum. One compound (1) was isolated and identified as (4S, 5S, 6S)-5,6-epoxy-4-hydroxy-3-methoxy-5-methyl-cyclohex-2-en-1-one. Growth of most of the tested phytopathogenic bacteria was inhibited by compound 1 and the ethyl acetate (EtOAc) layer except Pseudomonas syringae pv. lachrymans. Compound 1 also inhibited the mycelial growth of several plant pathogenic fungi. Both compound 1 and the EtOAc layer reduced bacterial leaf spot disease in detached peach leaves. They also suppressed the development of bacterial wilt on tomato seedlings quite effectively.

CONCLUSIONS

Amphirosellinia nigrospora JS-1675 showed antimicrobial activity against plant pathogenic bacteria and fungi by producing compound 1.

SIGNIFICANCE AND IMPACT OF THE STUDY

This is the first report on the occurrence of compound 1 in A. nigrospora JS-1675 and its efficacy against plant pathogenic bacteria and fungi. Their strong disease control efficacy against tomato bacterial wilt suggests that this fungus can be used as a microbial bactericide.

Utility of the laryngeal handshake method for identifying the cricothyroid membrane

BACKGROUND

The cricothyroid membrane is the most commonly accessed location for invasive surgical airway. Although the laryngeal handshake method is recommended for identifying the cricothyroid membrane, there is no clinical data regarding the utility of the laryngeal handshake method in cricothyroid membrane identification. The objective of this study was to compare the accuracy of cricothyroid membrane identification between the laryngeal handshake method and simple palpation.

METHODS

After anaesthesia induction, the otorhinolaryngology resident and anaesthesia resident identified and marked the needle insertion point for cricothyroidotomy using simple palpation and the laryngeal handshake method, respectively. The cricothyroid membrane was confirmed with ultrasonography. Identification was determined successful if the marked point was placed within the longitudinal area of the cricothyroid membrane and within 5 mm from midline transversely. The accuracy of cricothyroid membrane identification using the laryngeal handshake method and simple palpation was compared.

RESULTS

A total of 123 patients were enrolled. The cricothyroid membrane was correctly identified in 87 (70.7%, 95% confidence interval 61.8-78.6%) patients using the laryngeal handshake method compared to 78 (63.4%, 95% confidence interval 54.3-71.9%) patients using simple palpation (P = .188). The time required to identify the cricothyroid membrane was longer when using the laryngeal handshake method (15 [3-48] seconds vs 10.9 [3-55] seconds, P = .003).

CONCLUSION

The success rate of identifying the cricothyroid membrane was similar among the anesthesiologists who performed the laryngeal handshake method and also among otorhinolaryngologists who used simple palpation.

Small-Molecule Screen Identifies De Novo Nucleotide Synthesis as a Vulnerability of Cells Lacking SIRT3

Sirtuin 3 (SIRT3) is a NAD+-dependent deacetylase downregulated in aging and age-associated diseases such as cancer and neurodegeneration and in high-fat diet (HFD)-induced metabolic disorders. Here, we performed a small-molecule screen and identified an unexpected metabolic vulnerability associated with SIRT3 loss. Azaserine, a glutamine analog, was the top compound that inhibited growth and proliferation of cells lacking SIRT3. Using stable isotope tracing of glutamine, we observed its increased incorporation into de novo nucleotide synthesis in SIRT3 knockout (KO) cells. Furthermore, we found that SIRT3 KO cells upregulated the diversion of glutamine into de novo nucleotide synthesis through hyperactive mTORC1 signaling. Overexpression of SIRT3 suppressed mTORC1 and growth in vivo in a xenograft tumor model of breast cancer. Thus, we have uncovered a metabolic vulnerability of cells with SIRT3 loss by using an unbiased small-molecule screen.

Reversibility of atrophic gastritis and intestinal metaplasia after Helicobacter pylori eradication - a prospective study for up to 10 years

BACKGROUND

Atrophic gastritis and intestinal metaplasia are premalignant conditions for gastric cancer. Their reversibility by Helicobacter pylori eradication remains controversial.

AIM

To evaluate the reversibility of atrophic gastritis and intestinal metaplasia by H. pylori eradication with long-term follow-up.

METHODS

598 subjects were prospectively enrolled and followed for up to 10 years. They were categorised as H. pylori-negative (n = 65), H. pylori non-eradicated (n = 91), and H. pylori-eradicated (n = 442). Histological assessment was performed for antrum and corpus by Sydney classification.

RESULTS

Histological follow-up was performed regularly at 1, 2, 3-4 and ≥5 years, with mean follow-up of 1.07 ± 0.21, 2.29 ± 0.83, 3.93 ± 1.02, and 6.45 ± 1.28 years, respectively. Atrophic gastritis in antrum and corpus gradually and significantly (both P < .05 for all timepoints) improved only in the H. pylori-eradicated group compared to that at baseline. Significant difference in atrophic gastritis between H. pylori-eradicated and H. pylori-negative groups disappeared from 1-year follow-up. Similarly, intestinal metaplasia in antrum and corpus improved significantly (both P < .05 for all timepoints) only in the H. pylori-eradicated group in comparison with that at baseline. Significant difference in intestinal metaplasia between H. pylori-eradicated and H. pylori-negative groups disappeared from ≥5 years of follow-up in the antrum and from 3 years of follow-up in the corpus.

CONCLUSION

H. pylori eradication may be a preventative strategy for intestinal-type gastric cancer by regression of atrophic gastritis and intestinal metaplasia.

mTOR and HDAC Inhibitors Converge on the TXNIP/Thioredoxin Pathway to Cause Catastrophic Oxidative Stress and Regression of RAS-Driven Tumors

Although agents that inhibit specific oncogenic kinases have been successful in a subset of cancers, there are currently few treatment options for malignancies that lack a targetable oncogenic driver. Nevertheless, during tumor evolution cancers engage a variety of protective pathways, which may provide alternative actionable dependencies. Here, we identify a promising combination therapy that kills NF1-mutant tumors by triggering catastrophic oxidative stress. Specifically, we show that mTOR and HDAC inhibitors kill aggressive nervous system malignancies and shrink tumors in vivo by converging on the TXNIP/thioredoxin antioxidant pathway, through cooperative effects on chromatin and transcription. Accordingly, TXNIP triggers cell death by inhibiting thioredoxin and activating apoptosis signal-regulating kinase 1 (ASK1). Moreover, this drug combination also kills NF1-mutant and KRAS-mutant non-small cell lung cancers. Together, these studies identify a promising therapeutic combination for several currently untreatable malignancies and reveal a protective nodal point of convergence between these important epigenetic and oncogenic enzymes.Significance: There are no effective therapies for NF1- or RAS-mutant cancers. We show that combined mTOR/HDAC inhibitors kill these RAS-driven tumors by causing catastrophic oxidative stress. This study identifies a promising therapeutic combination and demonstrates that selective enhancement of oxidative stress may be more broadly exploited for developing cancer therapies. Cancer Discov; 7(12); 1450-63. ©2017 AACR.This article is highlighted in the In This Issue feature, p. 1355.

Metabolic recycling of ammonia via glutamate dehydrogenase supports breast cancer biomass

Ammonia is a ubiquitous by-product of cellular metabolism; however, the biological consequences of ammonia production are not fully understood, especially in cancer. We found that ammonia is not merely a toxic waste product but is recycled into central amino acid metabolism to maximize nitrogen utilization. In our experiments, human breast cancer cells primarily assimilated ammonia through reductive amination catalyzed by glutamate dehydrogenase (GDH); secondary reactions enabled other amino acids, such as proline and aspartate, to directly acquire this nitrogen. Metabolic recycling of ammonia accelerated proliferation of breast cancer. In mice, ammonia accumulated in the tumor microenvironment and was used directly to generate amino acids through GDH activity. These data show that ammonia is not only a secreted waste product but also a fundamental nitrogen source that can support tumor biomass.