scholarly journals Automated Identification of Patients with Immune-related Adverse Events from Clinical Notes using Word embedding and Machine Learning

2020 ◽  
Author(s):  
Samir Gupta ◽  
Anas Belouali ◽  
Neil J Shah ◽  
Michael B Atkins ◽  
Subha Madhavan
Keyword(s):  
Machine Learning ◽  
Adverse Events ◽  
Checkpoint Inhibitors ◽  
High Accuracy ◽  
Automated Identification ◽  
Real World Data ◽  
Clinical Notes ◽  
Safety Research ◽  
Advanced Malignancies ◽  
Treatment Safety

Immune Checkpoint Inhibitors (ICIs) have substantially improved survival in patients with advanced malignancies. However, ICIs are associated with a unique spectrum of side effects termed Immune-Related Adverse Events (irAEs). To ensure treatment safety, research efforts are needed to comprehensively detect and understand irAEs from real world data (RWD). The goal of this work is to evaluate a Machine Learning-based phenotyping approach that can identify patients with irAEs from a large volume of retrospective clinical notes representing RWD. Evaluation shows promising results with an average F1-score=0.75 and AUC-ROC=0.78. While the extraction of any available irAEs in charts achieves high accuracy, individual irAEs extraction has room for further improvement.

scholarly journals Automated Identification of Patients With Immune-Related Adverse Events From Clinical Notes Using Word Embedding and Machine Learning

2021 ◽  
pp. 541-549
Author(s):  
Samir Gupta ◽  
Anas Belouali ◽  
Neil J. Shah ◽  
Michael B. Atkins ◽  
Subha Madhavan
Keyword(s):  
Machine Learning ◽  
Adverse Events ◽  
Language Processing ◽  
Checkpoint Inhibitors ◽  
Characteristic Curve ◽  
Word Embeddings ◽  
Clinical Notes ◽  
Patient Level ◽  
Safety Research ◽  
Electronic Health

PURPOSE Although immune checkpoint inhibitors (ICIs) have substantially improved survival in patients with advanced malignancies, they are associated with a unique spectrum of side effects termed immune-related adverse events (irAEs). To ensure treatment safety, research efforts are needed to comprehensively detect and understand irAEs. Retrospective analysis of data from electronic health records can provide knowledge to characterize these toxicities. However, such information is not captured in a structured format within the electronic health record and requires manual chart review. MATERIALS AND METHODS In this work, we propose a natural language processing pipeline that can automatically annotate clinical notes and determine whether there is evidence that a patient developed an irAE. Seven hundred eighty-one cases were manually reviewed by clinicians and annotated for irAEs at the patient level. A dictionary of irAEs keywords was used to perform text reduction on clinical notes belonging to each patient; only sentences with relevant expressions were kept. Word embeddings were then used to generate vector representations over the reduced text, which served as input for the machine learning classifiers. The output of the models was presence or absence of any irAEs. Additional models were built to classify skin-related toxicities, endocrine toxicities, and colitis. RESULTS The model for any irAE achieved an average F1-score = 0.75 and area under the receiver operating characteristic curve = 0.85. This outperformed a basic keyword filtering approach. Although the classifier of any irAEs achieved good accuracy, individual irAE classification still has room for improvement. CONCLUSION We demonstrate that patient-level annotations combined with a machine learning approach using keywords filtering and word embeddings can achieve promising accuracy in classifying irAEs in clinical notes. This model may facilitate annotation and analysis of large irAEs data sets.

scholarly journals Multiple immune-related adverse events and anti-tumor efficacy: real-world data on various solid tumors

2020 ◽  
Author(s):  
Keitaro Shimozaki ◽  
Yasutaka Sukawa ◽  
Noriko Beppu ◽  
Isao Kurihara ◽  
Shigeaki Suzuki ◽  
...  
Keyword(s):  
Retrospective Study ◽  
Adverse Events ◽  
Real World ◽  
Proportional Hazards ◽  
Proportional Hazards Model ◽  
Checkpoint Inhibitors ◽  
Cox Proportional Hazards ◽  
Cox Proportional Hazards Model ◽  
University Hospital ◽  
Real World Data

Abstract Background Immune checkpoint inhibitors have been approved for various types of cancer; however, they cause a broad spectrum of immune-related adverse events (irAEs). The association between the development of irAEs and the clinical benefit remains uncertain. We aimed to evaluate the association of irAEs and the treatment efficacy in the real-world practice. Methods We conducted a retrospective study on patients with recurrent or metastatic non-small cell lung cancer, melanoma, renal cell carcinoma, or gastric cancer who received anti-PD-1/PD-L1 antibodies (nivolumab, pembrolizumab, or atezolizumab) at the Keio University Hospital between September 2014 and January 2019. We recorded treatment-related AEs from medical records and graded them using the Common Terminology Criteria for Adverse Events version 4. We performed an overall survival (OS) analysis using a Cox proportional hazards model. Results Among 212 patients eligible for this study, 108 experienced irAEs and 42 developed multiple irAEs. OS in patients with multiple irAEs was significantly longer than that in patients with single irAE (42.3 months vs. 18.8 months; hazard ratio [HR], 0.48; 95% confidence interval [CI], 0.25–0.93; P = 0.03). Moreover, OS from the development of a second irAE in those with multiple irAEs was longer than that from the development of the first irAE in patients with single irAEs (median OS, 26.9 months vs. 17.7 months, respectively; HR, 0.59; 95% CI, 0.30–1.14; P = 0.11). Conclusions Our single-center retrospective study revealed a remarkable tendency associating the development of multiple irAEs with favorable prognoses.

Impact of radiotherapy on risk of adverse events in patients receiving immunotherapy: A U.S. Food and Drug Administration pooled analysis.

2020 ◽  
Vol 38 (15_suppl) ◽  
pp. 3018-3018
Author(s):  
Mitchell Steven Anscher ◽  
Shaily Arora ◽  
Chana Weinstock ◽  
Rachael Lubitz ◽  
Anup Amatya ◽  
...  
Keyword(s):  
Adverse Events ◽  
Inflammatory Responses ◽  
Checkpoint Inhibitors ◽  
Pooled Analysis ◽  
Hematologic Toxicity ◽  
Pooled Data ◽  
Advanced Malignancies ◽  
Prospective Trials ◽  
Grade 3 ◽  
The Impact

3018 Background: Immune checkpoint inhibitors (ICIs) are widely used in the treatment of multiple advanced malignancies. Radiotherapy (RT) has been used in combination with ICIs to activate tumor-specific T cell responses, and RT also promotes non-specific acute and chronic inflammatory responses both locally and systemically. More than 50% of patients receive RT at some point during their course of cancer therapy, and relatively little information is available pertaining to the impact of RT, if any, on the risk of adverse events (AEs) in patients receiving ICIs. Methods: Pooled data from prospective trials of ICIs submitted to the FDA in initial or supplemental BLAs or NDAs through 12/2019 were included (N=66). Trials from applications that were withdrawn or not approved were not included. Patients were subdivided by whether or not radiotherapy was administered at any time during the course of their cancer treatment. AEs common to both ICI treatment and RT were identified to focus on the following reactions: neutropenia, thrombocytopenia, colitis, hepatitis, pneumonitis, and myocarditis. Descriptive statistics were used to examine AEs associated with the use of radiation and ICIs. Results: A total of 25,836 patients were identified, of which 9087 (35%) received RT and 16,749 (65%) did not. Radiation was associated with similar rates of AEs overall with numerically higher hematologic toxicities and pneumonitis and numerically lower colitis, hepatitis and myocarditis (Table). Patients receiving RT were more likely to experience Grade 3-5 hematologic toxicities compared to those not receiving RT. Conclusions: To our knowledge, this is the largest report of AE risk associated with the use of radiation and ICIs. Our results show that the incidence of hematologic toxicity and pneumonitis in patients receiving RT may be slightly higher. Analysis to determine comparability of baseline demographic characteristics, comprehensive AE profile, and timing of RT is underway. [Table: see text]

Safety of immunotherapy in Peruvian patients with diverse type of malignancies: Experience at a private center.

2020 ◽  
Vol 38 (15_suppl) ◽  
pp. e19258-e19258
Author(s):  
Alfredo Aguilar ◽  
Michelle Mariñas ◽  
Leny Bravo ◽  
Jenny Zavaleta ◽  
Joseph Pinto ◽  
...  
Keyword(s):  
Adverse Events ◽  
Malignant Tumors ◽  
Descriptive Analysis ◽  
Squamous Cell Carcinomas ◽  
First Line ◽  
Real World Data ◽  
Lung Cancers ◽  
Advanced Malignancies ◽  
Private Center ◽  
Third Line

e19258 Background: Immunotherapy has changed the landscape of cancer treatment. The aim of this work was to describe the adverse events related to immunotherapy treatment in diverse type of malignant tumors using real-world data. Methods: This is a retrospective review of patients with diverse type of advanced malignancies treated with immunotherapy at Oncosalud-AUNA (Lima-Peru) during the period 2016-2018. We present a descriptive analysis of the clinicopathological and treatment features of the patients, as well as data of safety of immunotherapeutic agents. Results: In total, 37 patients were included in the study. The median age was 67 years (38 to 84 years); 64.9% of patients were male; 54.1% were smoker/former smoker and 45.9% non-smokers. Regarding to the primary tumor, 75.7% were lung cancers (82.1%, adenocarcinomas and 17.9%, squamous cell carcinomas), 16.2% were melanomas, 5.4, head and neck cancers and 2.7%, were bladder cancers. Central nervous system metastases were present in 29.7% of patients. Immunotherapy was given after a first line in 43.2% of cases, 40.5% after the second line and 16.2%, after the third line of treatment. The types of immunotherapy were prembrolizumab in 54.1%, nivolumab in 40.5%, atezolizumab in 2.7% and avelumab in 2.7%. 27% of cases had combination of immuno with chemotherapy. Finally, regarding to adverse events, 94.6% had any adverse event; 48.6% fatigue and asthenia; 35.1%, nausea; 32.4, pruritus/rash; 27.7%, decreased appetite; 18.9%, hypo/hyperthyroidism; 13.5%, diarrhea/colitis; 10.8%, pneumonitis and 5.4%, infusion-related reactions. Conclusions: During the study period we had a slightly higher incidence of adverse events than reported by other works. It could be probably due to the age of patients and several prior lines of treatment.

Correlation of patient characteristics with clinical outcomes of camrelizumab combined with apatinib for unresectable hepatocellular carcinma (uHCC): An observational retrospective analysis.

2021 ◽  
Vol 39 (15_suppl) ◽  
pp. e16120-e16120
Author(s):  
Gang Liu ◽  
Liansheng Gong ◽  
Wenxuan Zhou ◽  
Xiaoli Li ◽  
Fei Wang ◽  
...  
Keyword(s):  
Adverse Events ◽  
Real World ◽  
Subgroup Analysis ◽  
Therapeutic Efficacy ◽  
Checkpoint Inhibitors ◽  
Objective Response ◽  
Therapeutic Effects ◽  
First Line ◽  
Real World Data ◽  
Baseline Serum

e16120 Background: There is limited data on clinical parameters to evaluate the therapeutic effects on immune checkpoint inhibitors (ICIs) combined with anti-angiogenic agent for uHCC. Here, we assessed efficacy and safety of camrelizumab combined with apatinib for uHCC from real-world data, and performed the retrospective subgroup analysis to investigate the potential factors related to therapy response and patients survival as well. Methods: We evaluated clinical data and outcome of 26 uHCC patients who received camrelizumab 200 mg intravenously every 2 weeks combined with apatinib 250 mg qd between May 2019 and Jul 2020. Objective response rate (ORR), disease control rate (DCR), progression free survival (PFS) and overall survival (OS) were evaluated using independent central review mRECIST and RECIST 1.1. Treatment related adverse events (TRAEs) and immune-related adverse events (irAEs) were evaluated. Results: The patients’ characteristics of our cohort are summarized in Table. Overall, our study shows that ORR was 57.7% (mRECIST), DCR was 84.62% (mRECIST), median PFS (mPFS) and OS (mOS) were 11 months and 18.2 months, respectively. For subgroup analysis, patients with first-line therapy (n=22) had dramatically better mPFS than non-first-line (15.0 vs. 4 months; p=0.01). Patients with baseline serum alpha-foetoprotein (AFP) > 400 ng/ml shows better therapeutic efficacy ( p<0.001). The patients with decreased AFP level after treatment had significantly longer mPFS (15.0 vs. 4 months; p<0.001) and mOS (NR vs. 5.7 months; p<0.001) than others. Overall, 14 (53.85%) patients had grade≥3 TRAEs, only 3 (11.54%) patients had grade≥2 irAEs. Conclusions: The first up-to-date real-world evidence indicates that both the baseline and post-treatment AFP level might be independent prognostic factors to evaluate the therapeutic efficacy and clinical outcome on the combination therapy of camrelizumab and apatinib. While larger sample sizes and longer follow-up study are needed to verify reliability of statistical results.[Table: see text]

Immune-related adverse events associated with immune-checkpoint inhibitors therapy in Mexican patients: Experience at a single center.

2021 ◽  
Vol 39 (15_suppl) ◽  
pp. e14580-e14580
Author(s):  
Iván Romarico Romarico Gonzalez Espinoza ◽  
Neil Cortés Escobar ◽  
Mariana Chiquillo-Domínguez ◽  
Gabriela Juárez Salazar ◽  
Julio Cesar Garibay Diaz ◽  
...  
Keyword(s):  
Adverse Events ◽  
Immune Checkpoint ◽  
Immune Checkpoint Inhibitors ◽  
Malignant Tumors ◽  
Checkpoint Inhibitors ◽  
Chemotherapeutic Agents ◽  
Real World Data ◽  
Non Melanoma Skin Cancer ◽  
Grade 3 ◽  
Patients Experience

e14580 Background: The use of immune-checkpoint inhibitors (ICIs) for solid malignancies is rapidly rising, and many new agents and treatment combinations are in development. However, ICIs have a unique side-effect profile of immune-related adverse events (irAEs) compared with chemotherapeutic agents or targeted therapies. The aim of this work was to describe the irAEs in diverse types of malignant tumors using real-world data. Methods: This is a retrospective and descriptive study of patients with diverse types of advanced malignancies treated with immunotherapy at Centro Oncológico Integral of the Hospital Ángeles in Puebla, México; during the period 2016-2020. Data about the primary neoplasm, ICIs, irAEs, organ system affected, grade and treatment was collected. Clinical and laboratory parameters were obtained by reviewing medical records. Results: A total of 117 patients were included, median age of 65 years, of which 63.2% were male and 36.8% were female. The most frequent neoplasms treated with ICIs were: lung (27.4%), kidney (16.2%), melanoma (12.8%), hepatocellular (9.4%), breast (8.5%), non-melanoma skin cancer (6.0%), mesothelioma (4.3%) and other tumors (15.3%). 39.3% of the patients had no metastases, 41.9% had metastases to at least 1 or 2 sites, and 18.8% to 3 or more sites. The types of ICIs were: nivolumab (35.0%), pembrolizumab (28.2%), atezolizumab (23.9%), ipilimumab + nivolumab (12.0%) and durvalumab (0.9%). The most frequent irAEs were: gastrointestinal (61.5%), neurologic (46.2%), pulmonary (38.5%), metabolic (32.5%) and hematologic (29.1%). 39.3% of the irAEs were reported as grade 1, 31.6% as grade 2, 14.5% as grade 3 and 2.6% as grade 4. Conclusions: Our work shows the incidence of irAEs in a poorly studied population and provides new data that complement that reported by other works, however, further prospective studies are necessary.[Table: see text]

scholarly journals Intestinal Microbiome Associated With Immune-Related Adverse Events for Patients Treated With Anti-PD-1 Inhibitors, a Real-World Study

2021 ◽  
Vol 12 ◽  
Author(s):  
Wenhui Liu ◽  
Fang Ma ◽  
Bao Sun ◽  
Yiping Liu ◽  
Haoneng Tang ◽  
...  
Keyword(s):  
Adverse Events ◽  
Checkpoint Inhibitors ◽  
Predictive Biomarkers ◽  
Intestinal Bacteria ◽  
Classification Model ◽  
Intestinal Microbiome ◽  
Fecal Samples ◽  
Microbial Biomarkers ◽  
Advanced Malignancies ◽  
Auc Value

AimImmune checkpoint inhibitors (ICIs) have updated the treatment landscape for patients with advanced malignancies, while their clinical prospect was hindered by severe immune-related adverse events (irAEs). The aim of this study was to research the association between gut microbiome diversity and the occurrence of ICI-induced irAEs.Patients and MethodWe prospectively obtained the baseline fecal samples and clinical data from patients treated with anti-PD-1 inhibitors as monotherapy or in combination with chemotherapy or antiangiogenesis regardless of treatment lines. The 16S rRNA V3-V4 sequencing was used to test the gene amplicons of fecal samples. The development of irAEs was evaluated and monitored from the beginning of therapy based on CTCAE V5.01.ResultsA total of 150 patients were included in the study and followed up for at least 6 months. A total of 90 (60%) patients developed at least one type of adverse effect, among which mild irAEs (grades 1–2) occurred in 65 patients (72.22%) and severe irAEs (grades 3–5) in 25 patients (27.78%). Patients with severe irAEs showed a visible higher abundance of Streptococcus, Paecalibacterium, and Stenotrophomonas, and patients with mild irAEs had a higher abundance of Faecalibacterium and unidentified_Lachnospiraceae. With the aid of a classification model constructed with 5 microbial biomarkers, patients without irAEs were successfully distinguished from those with severe irAEs (AUC value was 0.66).ConclusionCertain intestinal bacteria can effectively distinguish patients without irAEs from patients with severe irAEs and provide evidence of gut microbiota as an informative source for developing predictive biomarkers to predict the occurrence of irAEs.

scholarly journals Validation of a machine learning approach to estimate Clinical Disease Activity Index Scores for rheumatoid arthritis

RMD Open ◽  
2021 ◽  
Vol 7 (3) ◽  
pp. e001781
Author(s):  
Alison K. Spencer ◽  
Jigar Bandaria ◽  
Michelle B. Leavy ◽  
Benjamin Gliklich ◽  
Zhaohui Su ◽  
...  
Keyword(s):  
Rheumatoid Arthritis ◽  
Machine Learning ◽  
Disease Activity ◽  
Activity Index ◽  
Disease Activity Index ◽  
Learning Model ◽  
Real World Data ◽  
Clinical Notes ◽  
Machine Learning Model ◽  
R Value

ObjectiveDisease activity measures, such as the Clinical Disease Activity Index (CDAI), are important tools for informing treatment decisions and monitoring patient outcomes in rheumatoid arthritis (RA). Yet, documentation of CDAI scores in electronic medical records and other real-world data sources is inconsistent, making it challenging to use these data for research. The purpose of this study was to validate a machine learning model to estimate CDAI scores for patients with RA using clinical notes.MethodsA machine learning model was developed to estimate CDAI score values using clinical notes from a specific rheumatology visit. Data from the OM1 RA Registry were used to create a training cohort of 56 177 encounters and a separate validation cohort of 18 726 encounters, 11 985 of which passed a model-derived confidence filter; all included encounters had both a clinician-recorded CDAI score and a clinical note. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), positive predictive value (PPV) and negative predictive value (NPV), calculated using a binarised version of the outcome. The Spearman’s R and Pearson’s R values were also calculated.ResultsThe model had a PPV of 0.80, NPV of 0.84 and AUC of 0.88 when evaluating performance using the binarised version of the outcome. The model had a Spearman’s R value of 0.72 and a Pearson’s R value of 0.69 when evaluating performance using the continuous CDAI numeric scores.ConclusionA machine learning model estimates CDAI scores from clinical notes with good performance. Application of the model to real-world data sets may allow estimated CDAI scores to be used for research purposes.

Analysis of risk factors for immune-related adverse events in various solid tumors using real-world data

2021 ◽  
Author(s):  
Keitaro Shimozaki ◽  
Yasutaka Sukawa ◽  
Yasunori Sato ◽  
Sara Horie ◽  
Akihiko Chida ◽  
...  
Keyword(s):  
Risk Factors ◽  
Adverse Events ◽  
Serum Albumin ◽  
Checkpoint Inhibitors ◽  
Hazard Ratio ◽  
Type I ◽  
Hypersensitivity Reactions ◽  
Real World Data ◽  
History Of ◽  
Type I Hypersensitivity

The aim of this study was to determine the risk factors for immune-related adverse events (irAEs) induced by immune checkpoint inhibitors. The authors conducted a retrospective study in which patients with malignant melanoma, non-small-cell lung cancer, gastric cancer or renal cell carcinoma who received anti-PD-1/PD-L1 antibodies were included. Of 247 patients, 118 developed a total of 182 irAEs. In the multivariate Fine–Gray regression analysis, serum albumin level ≥3.6 g/dl (hazard ratio: 1.62; 95% CI: 1.10–2.39; p = 0.015) and history of Type I hypersensitivity reactions (hazard ratio: 1.48; 95% CI: 1.02–2.14; p = 0.037) were significantly associated with the development of irAEs. High serum albumin levels and history of Type I hypersensitivity reactions are risk factors for irAEs.

Single agent and combination checkpoint inhibitors therapy: A post marketing safety analysis.

2018 ◽  
Vol 36 (5_suppl) ◽  
pp. 125-125 ◽  
Author(s):  
Rawad Elias ◽  
Jennifer Rider ◽  
Xainming Tan ◽  
Osama E. Rahma
Keyword(s):  
Adverse Events ◽  
Older Patients ◽  
Checkpoint Inhibitors ◽  
Wide Spectrum ◽  
Descriptive Analysis ◽  
Single Agent ◽  
Adverse Event Reporting System ◽  
Age Groups ◽  
Older Individuals ◽  
Real World Data

125 Background: Immune Checkpoint Inhibitors (ICIs) are used to treat patients with a wide spectrum of malignancies. Individuals enrolled in clinical trials are usually more fit compared to patients seen in everyday practice. Therefore, it is important to evaluate the safety profile of these agents in real-world data. In addition, it is not clear if the safety of these agents is similar across all age groups. Methods: We reviewed the FDA Adverse Event Reporting System (FAERS) for adverse events associated with the use of PD-1 inhibitors (nivolumab and pembrolizumab); PD-L1 inhibitor (atezolizumab); and CTLA-4 inhibitor (ipilimumab). Our analysis was restricted to reports that included only an ICI as a suspect agent. For each agent, we performed a descriptive analysis of hospitalization (HO) and death (DE) outcomes, as well as select adverse events of special interest (AESI). We compared the distribution of each outcome within age groups ( < 65 years; 67-75; > 75) using Mantel-Haenszel chi -quare test for trend. Results: A total of 23,586 safety reports were included in our analysis. 415 for atezolizumab, 10,026 for nivolumab, 4,808 for pembrolizumab, 6,339 for ipilimumab, and 1,988 for the combination nivolumab plus ipilimumab. Increased age was associated with a statistically significant trend of more hospitalizations for all drugs except for the combination nivolumab plus ipilimumab where hospitalization rate was high ( > 80%) but similar across all age groups. Prevalence of any of the AESI was higher as age increased for all drugs (p < 0.0001) except for atezolizumab (p 0.12) and combination nivolumab plus ipilimumab (p 0.488). Proportion of older patients who experienced death was higher for pembrolizumab (p < 0.001), ipilimumab (p 0.002), and combination nivolumab plus ipilimumab (p < 0.001). Conclusions: Our analysis suggests that among patients treated with ICI, older individuals receiving pembrolizumab, Nivolumab or ipilimumab were more likely to develop immune related AEs, and to be hospitalized. An increased rate of death with higher age was seen with the use of pembrolizumab, ipilimumab, and combination nivolumab + ipilimumab. Older patients treated with ICIs should be monitored carefully for treatment-related AEs.

Sign in / Sign up

Export Citation Format

Share Document