Profile and outcome of cancer patients enrolled in contemporary phase I trials.

BACKGROUND: Phase I trials historically involved heavily pretreated patients (pts) with no more effective therapeutic options available and with poor expected outcomes. There are scare data regarding profile and outcomes of pts enrolled into modern phase I trials. Here, we sought to provide an overview of pts' profile and outcome into phase I trials at Gustave Roussy (GR). METHODS: This is a monocentric retrospective study, including all pts enrolled into phase I trials at GR from 2017 to 2021. Data regarding pts' demographics, tumour types, investigational treatments and survival outcomes were collected. RESULTS: In total, 9482 pts were referred for early phase trials; 2478 pts were screened, among which 449 (18.1%) failed screening; 1693 pts finally received at least one treatment dose as part of a phase I trial. Median age of pts was 59 years old (range, 18-88) and most common tumour types included gastrointestinal (25.3%), haematological (15%), lung (13.6%), genitourinary (10.5%) and gynaecologic cancers (9.4%). Amongst all pts treated and evaluable for response (1634 pts), objective response rate was 15.9% and disease control rate was 45.4%. Median progression-free survival and overall survival were, respectively, 2.6 months (95% confidence interval [95% CI], 2.3; 2.8) and 12.4 months (95% CI, 11.7; 13.6). CONCLUSION: As compared with historical data, our study shows that outcomes of pts included into modern phase I trials have improved and that these trials constitute nowadays a valid and safe therapeutic option. These updated data provide facts for adapting the methodology, role and place of phase I trials over the next years.

Natural Language Processing for Patient Selection in Phase I or II Oncology Clinical Trials.

PURPOSE: Early discontinuation affects more than one third of patients enrolled in early-phase oncology clinical trials. Early discontinuation is deleterious both for the patient and for the study, by inflating its duration and associated costs. We aimed at predicting the successful screening and dose-limiting toxicity period completion (SSD) from automatic analysis of consultation reports. MATERIALS AND METHODS: We retrieved the consultation reports of patients included in phase I and/or phase II oncology trials for any tumor type at Gustave Roussy, France. We designed a preprocessing pipeline that transformed free text into numerical vectors and gathered them into semantic clusters. These document-based semantic vectors were then fed into a machine learning model that we trained to output a binary prediction of SSD status. RESULTS: Between September 2012 and July 2020, 56,924 consultation reports were used to build the dictionary and 1,858 phase I or II inclusion reports were used to train (72%), validate (14%), and test (14%) a random forest model. Preprocessing could efficiently cluster words with semantic proximity. On the unseen test cohort of 264 consultation reports, the performances of the model reached: F1 score 0.80, recall 0.81, and area under the curve 0.88. Using this model, we could have reduced the screen fail rate (including dose-limiting toxicity period) from 39.8% to 12.8% (relative risk, 0.322; 95% CI, 0.209 to 0.498; P < .0001) within the test cohort. Most important semantic clusters for predictions comprised words related to hematologic malignancies, anatomopathologic features, and laboratory and imaging interpretation. CONCLUSION: Machine learning with semantic conservation is a promising tool to assist physicians in selecting patients prone to achieve SSD in early-phase oncology clinical trials.

ICD-10 hospital discharge diagnosis codes were sensitive for identifying pulmonary embolism but not deep vein thrombosis.

OBJECTIVE: To estimate the sensitivity of International Classification of Diseases, Tenth revision (ICD-10) hospital discharge diagnosis codes for identifying deep vein thrombosis (DVT) and pulmonary embolism (PE). STUDY DESIGN AND SETTING: We compared predefined ICD-10 discharge diagnosis codes with the diagnoses that were prospectively recorded for 1,375 patients with suspected DVT or PE who were enrolled at 25 hospitals in France. Sensitivity was calculated as the percentage of patients identified by predefined ICD-10 codes among positive cases of acute symptomatic DVT or PE confirmed by objective testing. RESULTS: The sensitivity of ICD-10 codes was 58.0% (159 of 274; 95% CI: 51.9, 64.1) for isolated DVT and 88.9% (297 of 334; 95% CI: 85.6, 92.2) for PE. Depending on the hospital, the median values for sensitivity were 57.7% for DVT (interquartile range, IQR, 48.6-66.7; intracluster correlation coefficient, 0.02; P=0.31) and 88.9% for PE (IQR, 83.3-96.3; intracluster correlation coefficient, 0.11; P=0.03). The sensitivity of ICD-10 codes was lower for surgical patients and for patients who developed PE or DVT while they were hospitalized. CONCLUSION: ICD-10 discharge diagnosis codes yield satisfactory sensitivity for identifying objectively confirmed PE. A substantial proportion of DVT cases are missed when using hospital discharge data for complication screening or research purposes.