Prediction of long-term mortality by using machine learning models in Chinese patients with connective tissue disease-associated interstitial lung disease

Background The exact risk assessment is crucial for the management of connective tissue disease-associated interstitial lung disease (CTD-ILD) patients. In the present study, we develop a nomogram to predict 3‑ and 5-year mortality by using machine learning approach and test the ILD-GAP model in Chinese CTD-ILD patients. Methods CTD-ILD patients who were diagnosed and treated at the First Affiliated Hospital of Zhengzhou University were enrolled based on a prior well-designed criterion between February 2011 and July 2018. Cox regression with the least absolute shrinkage and selection operator (LASSO) was used to screen out the predictors and generate a nomogram. Internal validation was performed using bootstrap resampling. Then, the nomogram and ILD-GAP model were assessed via likelihood ratio testing, Harrell’s C index, integrated discrimination improvement (IDI), the net reclassification improvement (NRI) and decision curve analysis. Results A total of 675 consecutive CTD-ILD patients were enrolled in this study, during the median follow-up period of 50 (interquartile range, 38–65) months, 158 patients died (mortality rate 23.4%). After feature selection, 9 variables were identified: age, rheumatoid arthritis, lung diffusing capacity for carbon monoxide, right ventricular diameter, right atrial area, honeycombing, immunosuppressive agents, aspartate transaminase and albumin. A predictive nomogram was generated by integrating these variables, which provided better mortality estimates than ILD-GAP model based on the likelihood ratio testing, Harrell’s C index (0.767 and 0.652 respectively) and calibration plots. Application of the nomogram resulted in an improved IDI (3- and 5-year, 0.137 and 0.136 respectively) and NRI (3- and 5-year, 0.294 and 0.325 respectively) compared with ILD-GAP model. In addition, the nomogram was more clinically useful revealed by decision curve analysis. Conclusions The results from our study prove that the ILD-GAP model may exhibit an inapplicable role in predicting mortality risk in Chinese CTD-ILD patients. The nomogram we developed performed well in predicting 3‑ and 5-year mortality risk of Chinese CTD-ILD patients, but further studies and external validation will be required to determine the clinical usefulness of the nomogram.

Statistical Modelling of Population-Level Exonic Variant Frequency Data with an Emphasis on Rare Variants

Introduction & Objective: Rare variants with allele frequency smaller than 1% are postulated to be associated with disease susceptibility. Since allele frequencies vary globally, the use of population control data that does not match the study population can produce bias. The research question is to identify factors that explain variation in allele frequency across populations. The secondary question is to evaluate the potential bias in using population as control data when studying variants. We use data from gnomAD (Genome Aggregation Database) to answer these questions. Methods: We apply each of three model formulations: Linear, Logistic, and Poisson to explain how the frequency or count of variants depends on population subgroup/ancestry, functional annotation, sex, and disease status. We also evaluate interactions between population subgroups and functional annotation. Results: For very rare variants (allele frequency < 0.1%), likelihood ratio testing (LRT) provides evidence that allele frequencies vary with functional annotation and population in all three model formulations. By LRT, interactions of population and functional annotation are significant in the Logistic model and the Poisson model. The goodness-of-fit statistics show a better fit in the linear model compared to low frequency variants. Conclusion: We observe that population & functional annotation affect variant frequencies, and conclude that detection of differences across populations and annotations is model scale-dependent, especially for different degrees of rareness. Therefore, statisticians need to carefully consider the potential for bias when using gnomAD as control data. Moreover, gnomAD is a great resource for studies dealing with rare variants.

Selective tweeting of COVID-19 articles: Does title or abstract positivity influence dissemination?

Background: Previous research has shown that articles may be cited more frequently on the basis of title or abstract positivity. Whether a similar selective sharing practice exists on Twitter is not well understood. The objective of this study was to assess if COVID-19 articles with positive titles or abstracts were tweeted more frequently than those with non-positive titles or abstracts. Methods: COVID-19 related articles published between January 1st and April 14th, 2020 were extracted from the LitCovid database and all articles were screened for eligibility. Titles and abstracts were classified using a list of positive and negative words from a previous study. A negative binomial regression analysis controlling for confounding variables (2018 impact factor, open access status, continent of the corresponding author, and topic) was performed to obtain regression coefficients, with the p values obtained by likelihood ratio testing. Results: A total of 3752 COVID-19 articles were included. Of the included studies, 44 titles and 112 abstracts were positive; 1 title and 7 abstracts were negative; and 3707 titles and 627 abstracts were neutral. Articles with positive titles had a lower tweet rate relative to articles with non-positive titles, with a regression coefficient of -1.10 (P < .001), while the positivity of the abstract did not impact tweet rate (P = .2218). Conclusion: COVID-19 articles with non-positive titles are preferentially tweeted, while abstract positivity does not influence tweet rate.

Does anxiety moderate the effectiveness of mirtazapine in patients with treatment-resistant depression? A secondary analysis of the MIR trial

Background: There is a lack of evidence to guide treatment of comorbid depression and anxiety. Preliminary evidence suggests mirtazapine may be effective in treating patients with both depression and anxiety symptoms. Methods: We undertook a secondary analysis of mirtazapine (MIR): a placebo-controlled trial of the addition of mirtazapine to a selective serotonin reuptake inhibitor or serotonin–norepinephrine reuptake inhibitor in treatment-resistant depression (TRD) in primary care. We subdivided participants into three groups by baseline generalized anxiety disorder score (GAD-7): severe (GAD-7 ⩾ 16), moderate (GAD-7 = 11–15), no/mild (GAD-7 ⩽ 10). We used linear regression including likelihood-ratio testing of interaction terms to assess how baseline anxiety altered the response of participants to mirtazapine as measured by 12-week GAD-7 and Beck Depression Inventory II (BDI-II) scores. Results: Baseline generalized anxiety moderated mirtazapine’s effect as measured by GAD-7 ( p = 0.041) and BDI-II ( p = 0.088) at 12 weeks. Participants with severe generalized anxiety receiving mirtazapine had lower 12-week GAD-7 score (adjusted difference between means (ADM) −2.82, 95% confidence interval (CI) −0.69 to −4.95) and larger decreases in BDI-II score (ADM −6.36, 95% CI −1.60 to −10.84) than placebo. Conversely, there was no anxiolytic benefit (ADM 0.28, 95% CI −1.05 to 1.60) or antidepressant benefit (ADM −0.17, 95% CI −3.02 to 2.68) compared with placebo in those with no/mild generalized anxiety. Conclusions: These findings extend the evidence for the effectiveness of mirtazapine to reduce generalized anxiety in TRD in primary care. These results may inform targeted prescribing in depression based on concurrent anxiety symptoms, although these conclusions are constrained by the post-hoc nature of this analysis.

THU0484 FIBROMYALGIA AND MULTIPLE SWITCHING OF BIOLOGICS IN SPONDYLOARTHRITIS

Background:Fibromyalgia (FM) is a condition characterized by chronic widespread pain, tender points, fatigue and disturbed sleep rhythm. Some of these symptoms such as fatigue, tender points and diffuse pain seen in patients with spondylarthritis (SpA). Moreover, FM and SpA can coexist creating a diagnostic challenge, particularly in early disease course and influence clinical disease activity assessment.Objectives:With this cross-sectional study, we aim to estimate the prevalence of FM in SpA and to elaborate its effect on biological treatments.Methods:FM was identified according to the ACR 2010 diagnostic criteria. SpA patients identified according to rheumatologist using various SpA subsets criteria. A review of the electronic medical files for SpA patients attending the rheumatology outpatient clinic and infusion unit at a major tertiary hospital during the period from June to December 2018 were included. Patients’ demographics, socioeconomics, disease characteristics, activity, HLA status and abnormal MRI sacroiliac were explored. Regarding SpA medications, number, frequency and dose of DMARDs and biological agents were obtained.Continuous variables were reported by their mean and standard deviation (SD) and qualitative variables by frequency and percentage. Statistical significance was set at p <0.05. Statistical analysis was performed using SPSS version 23.Results:Of the 305 enrolled SpA patients, 43 (14.1%) had FM. Females represents 57.4% of the patients, mean age was 44.07 ± 11.85 years. Arab ethnicity represents most of our cohort 84.9%, the majority were Emirati 64.6%. Smokers were 8.2% and ex-smokers were 3.3%. Axial SpA represents 38.4% while peripheral SpA 61.6% of our cohort according to ASAS classification.HLA B27 tested in a sample of 180 patients; it was positive in only 17.8%. CRP found to be elevated in 20.3% of the patients at baseline. Abnormal MRI SIJ bone marrow edema changes were found in 10.8%, while other SIJ changes was seen in additional 20.6%. The prevalence of FM showed no statistically significant difference between axial and peripheral SpA. Patients SpA and FM have longer disease duration than SpA alone, P= 0.034. Table.1 show demographics, socioeconomics and clinical data of our cohort.Regarding medication, the use of biologics among SpA patients with FM is more frequent than SpA patients without FM (74.4% vs 51.5 % respectively), P= 0.005. Interestingly, the likelihood ratio testing showed that SpA patient with Fibromyalgia switch more frequently to another biologics than SpA without fibromyalgia, P= 0.015.Cramer’s V test showed that there is a high statistically significant (P= 0.002) and very strong association (> 0.25) between presence of Fibromyalgia and multiple switching of biologics in SpA.There was no difference in the exposure to prednisolone nor conventional DMARDs between SpA patients with or without FM, P= 0.64 & 1 respectively.Gender, Female, n (%)175 (57.4)Age, mean ± SD (min- max), years44.07 ± 11.85 (18- 78)Type of A, n (%)AxialPeripheral117 (38.4)188 (61.6)Fibromyalgia, n (%)FM in axial SpAFM in Peripheral SpA43 (14.1)18 (41.9)25 (58.1)SpA Disease duration (months)FM+, mean ±SDFM-, mean ±SD107.7± 50.486± 57.9Elevated CRP, n (%)62 (20.3)HLA B27 in180 patients, n (%)PositiveNegative32 (17.8)148 (82.2)Abnormal MRI SIJ, n (%)Bone marrow edemaSubchondral sclerosisFatty transformation of bone marrowErosion92 (30.2)33 (10.8)21 (6.9)5 (1.6)2 (0.7)Number of conventional DMARDs ever tired, n (%)NoneOneTwoThree81 (26.6)166 (54.4)46 (15.1)12 (3.9)Frequency of DMARDs usage, n, (%)Conventional DMARDsPrednisoloneBiologic DNARDs224 (73.4)56 (18.4)164 (53.8)Conclusion:FM coexistence with SpA might impact clinical evaluation of disease activity and possibly negatively affect self-measurement of treatment response. In our study, SPA patients exposed to more biologics if they have coexisting FM; Moreover, they are more frequent switchers among biologics including TNFi and IL17i.Acknowledgments:N Elsidig, A Al Marzooqi, N Zamani, A HossainiDisclosure of Interests: :None declared

Hypomagnesemia is associated with an increased risk of early clinical failure in patients with Burkitt lymphoma.

e20048 Background: No literature exists regarding whether hypomagnesemia at the time of diagnosis in Burkitt Lymphoma (BL) is associated with inferior survival. Methods: Patients with new diagnosis of BL who were seen at Mayo Clinic, MN from 2000-2019 with a serum magnesium level available prior to chemotherapy were included. Patients were allocated to two groups; Abnormal Magnesium Group (AMG), defined as a magnesium level < 1.7 mg/dL and Normal Magnesium Group (NMG), defined as ≥ 1.7 – 2.3mg/dL. Two-sided Wilcoxon rank sum test and Chi square/Fischer’s exact test were used to compare the continuous and categorical variables, respectively. Kaplan-Meier and log-rank tests were used to perform all time to event analysis which were done from time of treatment. Hazard ratios (HR) with confidence intervals (CI) were calculated using Cox-proportional hazards. Results: Of 90 patients with a diagnosis of BL, 42 patients had a magnesium level at or before time of diagnosis. The Table lists the baseline characteristics and significant findings of the AMG/NMG groups. The median follow-up was 30 months. Hypomagnesemia was predictive for inferior EFS at 30 months - 34.3% (95% CI: 8.7–74.0) for AMG and 79.3% (95% CI: 60.8-90.4) for NMG (p = 0.038). OS at 30 months for AMG was 42.9% (95% CI: 14.4-77.0) and for NMG was 93.7% (95% CI: 78.1-98.4%) (p = 0.002). Other parameters significant on univariate analysis included the Charlson Comorbidity Index (p = 0.034), creatinine (p = 0.009), and number of treatments (p = 0.048). In the multivariate analysis, creatinine (p = 0.005) and number of treatments (p = 0.004) were significant. Likelihood ratio testing of predictors associated with inferior OS were significant for hypomagnesemia (likelihood ratio 3.99, p = 0.046). Conclusions: Patients with hypomagnesemia at the time of BL diagnosis have inferior outcomes compared to BL patients with a normal magnesium level. Prospective studies are needed to confirm this finding and test Mg replacement strategies to mitigate the effects of hypomagnesemia. [Table: see text]

HER2 status and prediction of extended endocrine benefit with breast cancer index (BCI) in HR+ patients in the adjuvant tamoxifen: To offer more? (aTTom) trial.

522 Background: BCI is a validated gene expression-based assay that stratifies patients based on risk of overall (0-10y) and late (post-5y) distant recurrence (DR) and predicts likelihood of benefit from extended endocrine therapy (EET). The Trans-aTTom study established Level1B validation for BCI (H/I) to predict benefit from EET.1 In this updated Trans-aTTom analysis including HER2 status, BCI (H/I) and prediction of endocrine benefit were further characterized. Methods: Centralized HER2 was determined for all cases according to current ASCO/CAP guidelines. Kaplan-Meier and Cox proportional hazards regression were conducted to assess primary and secondary endpoints of Recurrence-Free Interval (RFI) and Disease-Free Interval (DFI), respectively. A three-way interaction using likelihood ratio testing, which included treatment, BCI (H/I) and HER2, was performed to assess the effect of HER2 on BCI (H/I) prediction of EET benefit. Results: Of 789 N+ patients, 90% (N = 711) and 9% (N = 72) were HR+/HER2- and HR+/HER2+, respectively. In the HER2- subset, BCI (H/I)-High (48%) showed significant benefit from 10y vs. 5y of tamoxifen (9.4% RFI: HR = 0.35 [95% CI 0.15-0.81]; P = 0.047) while BCI (H/I)-Low patients did not (-2.1% RFI; HR = 1.15 [95% CI 0.78-1.69]; P = 0.491). For DFI, BCI (H/I)-High patients also showed significant benefit (10.3% DFI; HR = 0.41 [95% CI 0.18-0.91]; P = 0.047) while BCI (H/I)-Low patients did not (-1.7% DFI; HR = 1.10 [95% CI 0.75-1.62] P = 0.612). As demonstrated in the overall N+ cohort, significant interaction between BCI (H/I) and treatment was shown in the HER2- subset (RFI P = 0.045; DFI P = 0.044). Notably, three-way interaction evaluating BCI (H/I), treatment and HER2 status was not statistically significant (P = 0.85), indicating the ability of BCI (H/I) to predict benefit of EET activity was not significantly affected by HER2 status. Conclusions: In this updated Trans-aTTom analysis with HER2 data, BCI (H/I) showed similar predictive performance for EET response in the HER2- subset when compared to the overall N+ cohort. These data further support the clinical utility of BCI (H/I) as a predictive biomarker for informing EET benefit in HR+/HER2- and HR+/HER2+ disease. Clinical trial information: NCT00003678 . [Table: see text]