Potential of ChatGPT and GPT-4 for Data Mining of Free-Text CT Reports on Lung Cancer.

Background The latest large language models (LLMs) solve unseen problems via user-defined text prompts without the need for retraining, offering potentially more efficient information extraction from free-text medical records than manual annotation. Purpose To compare the performance of the LLMs ChatGPT and GPT-4 in data mining and labeling oncologic phenotypes from free-text CT reports on lung cancer by using user-defined prompts. Materials and Methods This retrospective study included patients who underwent lung cancer follow-up CT between September 2021 and March 2023. A subset of 25 reports was reserved for prompt engineering to instruct the LLMs in extracting lesion diameters, labeling metastatic disease, and assessing oncologic progression. This output was fed into a rule-based natural language processing pipeline to match ground truth annotations from four radiologists and derive performance metrics. The oncologic reasoning of LLMs was rated on a five-point Likert scale for factual correctness and accuracy. The occurrence of confabulations was recorded. Statistical analyses included Wilcoxon signed rank and McNemar tests. Results On 424 CT reports from 424 patients (mean age, 65 years ± 11 [SD]; 265 male), GPT-4 outperformed ChatGPT in extracting lesion parameters (98.6% vs 84.0%, P < .001), resulting in 96% correctly mined reports (vs 67% for ChatGPT, P < .001). GPT-4 achieved higher accuracy in identification of metastatic disease (98.1% [95% CI: 97.7, 98.5] vs 90.3% [95% CI: 89.4, 91.0]) and higher performance in generating correct labels for oncologic progression (F1 score, 0.96 [95% CI: 0.94, 0.98] vs 0.91 [95% CI: 0.89, 0.94]) (both P < .001). In oncologic reasoning, GPT-4 had higher Likert scale scores for factual correctness (4.3 vs 3.9) and accuracy (4.4 vs 3.3), with a lower rate of confabulation (1.7% vs 13.7%) than ChatGPT (all P < .001). Conclusion When using user-defined prompts, GPT-4 outperformed ChatGPT in extracting oncologic phenotypes from free-text CT reports on lung cancer and demonstrated better oncologic reasoning with fewer confabulations. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Hafezi-Nejad and Trivedi in this issue.

Satisfaction with radiotherapy care among cancer patients treated in Germany-secondary analysis of a large multicenter study.

PURPOSE: Patient satisfaction with healthcare has been linked to clinical outcomes and regulatory agencies demand its regular assessment. Therefore, we aimed to investigate patient satisfaction with radiotherapy care and its determinants. METHODS: This is a secondary analysis of a multicenter prospective cross-sectional study. Eligible cancer patients anonymously completed questionnaires at the end of a course of radiotherapy. The outcome variable was overall patient satisfaction with radiotherapy care measured with a 10-point Likert scaled single-item. Given patient satisfaction was defined for patients scoring ≥ 8 points. Determinants of given patient satisfaction were assessed by univariable and multivariable analyses. A p-value < 0.05 was considered statistically significant. RESULTS: Out of 2341 eligible patients, 1075 participated (participation rate 46%). Data on patient satisfaction was provided by 1054 patients. There was a right-skewed distribution towards more patient satisfaction (mean = 8.8; SD = 1.68). Given patient satisfaction was reported by 85% (899/1054) of the patients. Univariable analyses revealed significant associations of lower patient satisfaction with tumor entity (rectal cancer), concomitant chemotherapy, inpatient care, treating center, lower income, higher costs, and lower quality of life. Rectal cancer as tumor entity, treating center, and higher quality of life remained significant determinants of patient satisfaction in a multivariable logistic regression. CONCLUSION: Overall patient satisfaction with radiotherapy care was high across 11 centers in Germany. Determinants of patient satisfaction were tumor entity, treating center, and quality of life. Although these data are exploratory, they may inform other centers and future efforts to maintain high levels of patient satisfaction with radiotherapy care.

[Design and Evaluation of a Clinical Investigator Training for Student-lead Prospective Multicentre Clinical Trials: a CHIR-Net SIGMA Research-based Learning Project]. / Design und Evaluation eines Prüf-Studierenden-Kurses für studentische prospektive Multicenterstudien ­ ein CHIR-Net-SIGMA-Projekt zum forschenden Lernen.

BACKGROUND: Scientific skills are not sufficiently taught during medical training, neither in medical school nor during postgraduate education. This results in a lack of clinician scientists. In order to counter this problem, the surgical study network (CHIR-Net) founded SIGMA (Student-initiated German Medical Audit). This paper describes the development, performance and evaluation of a Clinical Investigator Training (CIT) aiming to qualify students to autonomously conduct clinical trials. MATERIAL AND METHODS: Based on the Kern cycle, a curriculum was developed, composed of three parts: online tutorials, a workshop and a follow-up period. The educational objectives were defined according to Bloom's taxonomy of knowledge. The learning objectives were based on the requirements of the "Network of Coordinating Centers for Clinical Trials" and the German Medical Association as well as content relevant to clinical studies. A wide range of educational instruments and assessments were used. By including all relevant professional groups involved in clinical trials, an interconnected working environment for students was generated. The increase in knowledge was assessed by a multiple-choice pre/post exam. The satisfaction of participants was analysed by a 5-point Likert scale, on which 5 indicated full approval. RESULTS: The first SIGMA CIT was realised in 2018; the workshop took place in Heidelberg in February. Thirty-two medical students from thirteen different centres participated. On average, 53.8 ± 8.3% of questions were answered correctly in the pre-test, compared with 71.2 ± 7.2% in the post-test (p < 0.0001). The maximal individual improvement was 30%; the lowest difference compared to the pre-test was 5%. Subjective evaluation results were positive with an average result of 4.63 ± 0.34 on a 5-point Likert scale. CONCLUSION: It is feasible to teach medical students the fundamentals of clinical trials. A compact Clinical Investigator Training using modern principles of teaching is able to prepare students for an autonomous performance of clinical trials.

Prospective multicentre cohort study of patient-reported outcomes and complications following major abdominal neoplastic surgery (PATRONUS) - study protocol for a CHIR-Net student-initiated German medical audit study (CHIR-Net SIGMA study).

BACKGROUND: One of the most important aspects of designing a clinical trial is selecting appropriate outcomes. Patient-reported outcomes (PROs) can provide a personal assessment of the burden and impact of a malignant disease and its treatment. PROs comprise a wide range of outcomes including basic clinical symptom scores and complex metrics such as health-related quality of life (HRQoL). There is limited data on how postoperative complications following cancer surgery affect symptoms and HRQoL. For this reason the primary aim of the PATRONUS study is to investigate how perioperative complications affect cancer-related symptoms and HRQoL in patients undergoing abdominal cancer surgery. The PATRONUS study is designed and will be initiated and conducted by medical students under the direct supervision of clinician scientists based on the concept of inquiry-based learning. METHODS: PATRONUS is a non-interventional prospective multicentre cohort study. Patients undergoing elective oncological abdominal surgery will be recruited at regional centres of the clinical network of the German Surgical Society (CHIR-Net) and associated hospitals. A core set of 12 cancer associated symptoms will be assessed via the PRO version of the Common Terminology Criteria for Adverse Events. The cancer-specific HRQoL will be measured via the computerised adaptive testing version of the European Organisation for Research and Treatment of Cancer (EORTC) QLQ-C30. PROs will be measured eight times over a period of six months. The short-term clinical outcome measure is the rate of postoperative complications (grade II to V) within 30 days according to the Clavien-Dindo classification. The long-term clinical outcome is overall survival within six months postoperative. DISCUSSION: PATRONUS will provide essential insights into the patients' assessment of their well-being and quality of life in direct relation to clinical outcome parameters following abdominal cancer surgery. Furthermore, PATRONUS will investigate the feasibility of multicentre student-led clinical research. TRIAL REGISTRATION: German Clinical Trials Register: DRKS00013035 (registered on October 26, 2017). Universal Trial Number (UTN): U1111-1202-8863.

Pelvic Irradiation for Node-Positive Prostate Cancer After Prostatectomy: Long-Term Results of the Prospective PLATIN-4 and PLATIN-5 Trials.

PURPOSE: After radical prostatectomy (RP), adjuvant or salvage radiation treatment in node-positive prostate cancer is offered to prevent systemic disease. Prospective long-term survival and toxicity data on patients with radiation for nodal disease are still scarce. This study evaluates safety and feasibility of salvage radiation therapy to the pelvic lymph nodes in node-positive prostate cancer after RP. METHODS AND MATERIALS: Between 2009 and 2018, 78 patients with lymph node recurrence after RP (PLATIN-4 trial) or after RP and prostate bed radiation therapy (PLATIN-5 trial) were treated with salvage pelvic lymph node radiation therapy with boost to the involved nodes as field abutment (PLATIN-5) and boost to the prostate bed (PLATIN-4). Androgen deprivation therapy was started 2 months before radiation and recommended for 24 months. The primary endpoint was safety and feasibility of the intensity modulated radiation therapy-image guided radiation therapy technique based on the rate of treatment discontinuations and incidence of Common Terminology Criteria for Adverse Events grade 3+ toxicity. Secondary endpoints were progression-free survival and overall survival. RESULTS: No treatment discontinuations were reported in either trial. Median overall survival was not reached in PLATIN-4 and was 117 months in PLATIN-5. Median progression-free survival was 66 months in PLATIN-4 and 39 months in PLATIN-5. Late grade 3+ genitourinary and gastrointestinal toxicities were observed in 4% of patients at 24 months of follow-up. CONCLUSIONS: Salvage radiation therapy to the prostate bed and pelvic lymphatic drainage combined with long-term androgen deprivation therapy is a curative treatment option for patients with node-positive prostate cancer after RP, with excellent in-field disease control. Pelvic lymph node radiation therapy as field abutment after prostate bed radiation therapy is feasible with long-term survival and no high-grade toxicity.

Dosimetric benefit of online treatment plan adaptation in stereotactic ultrahypofractionated MR-guided radiotherapy for localized prostate cancer.

Background: Apart from superior soft tissue contrast, MR-guided stereotactic body radiation therapy (SBRT) offers the chance for daily online plan adaptation. This study reports on the comparison of dose parameters before and after online plan adaptation in MR-guided SBRT of localized prostate cancer. Materials and methods: 32 consecutive patients treated with ultrahypofractionated SBRT for localized prostate cancer within the prospective SMILE trial underwent a planning process for MR-guided radiotherapy with 37.5 Gy applied in 5 fractions. A base plan, derived from MRI simulation at an MRIdian Linac, was registered to daily MRI scans (predicted plan). Following target and OAR recontouring, the plan was reoptimized based on the daily anatomy (adapted plan). CTV and PTV coverage and doses at OAR were compared between predicted and adapted plans using linear mixed regression models. Results: In 152 out of 160 fractions (95%), an adapted radiation plan was delivered. Mean CTV and PTV coverage increased by 1.4% and 4.5% after adaptation. 18% vs. 95% of the plans had a PTV coverage ≥95% before and after online adaptation, respectively. 78% vs. 100% of the plans had a CTV coverage ≥98% before and after online adaptation, respectively. The D0.2cc for both bladder and rectum were <38.5 Gy in 93% vs. 100% before and after online adaptation. The constraint at the urethra with a dose of <37.5 Gy was achieved in 59% vs. 93% before and after online adaptation. Conclusion: Online adaptive plan adaptation improves target volume coverage and reduces doses to OAR in MR-guided SBRT of localized prostate cancer. Online plan adaptation could potentially further reduce acute and long-term side effects and improve local failure rates in MR-guided SBRT of localized prostate cancer.

Patient-Relevant Costs for Organ Preservation versus Radical Resection in Locally Advanced Rectal Cancer.

Total neoadjuvant therapy (TNT) is an evolving treatment schedule for locally advanced rectal cancer (LARC), allowing for organ preservation in a relevant number of patients in the case of complete response. Patients who undergo this so-called "watch and wait" approach are likely to benefit regarding their quality of life (QoL), especially if definitive ostomy could be avoided. In this work, we performed the first cost-effectiveness analysis from the patient perspective to compare costs for TNT with radical resection after neoadjuvant chemoradiation (CRT) in the German health care system. Individual costs for patients insured with a statutory health insurance were calculated with a Markov microsimulation. A subgroup analysis from the prospective "FinTox" trial was used to calibrate the model's parameters. We found that TNT was less expensive (-1540 EUR) and simultaneously resulted in a better QoL (+0.64 QALYs) during treatment and 5-year follow-up. The average cost for patients under TNT was 4711 EUR per year, which was equivalent to 3.2% of the net household income. CRT followed by resection resulted in higher overall costs for ostomy care, medication and greater loss of earnings. Overall, TNT appeared to be more efficacious and cost-effective from a patient's point of view in the German health care system.

Risk-adjusted Screening for Prostate Cancer-Defining the Low-risk Group by Data from the PROBASE Trial.

BACKGROUND: Risk-adjusted screening for prostate cancer (PCa) aims to reduce harms by less frequent retesting, especially in men at a low risk of PCa. Definitions of low risk are based mainly on studies in men starting screening at age 55-60 yr. OBJECTIVE: To identify men at age 45 yr with a low risk of PCa. DESIGN, SETTING, AND PARTICIPANTS: A population-based, risk-adjusted PCa screening trial was conducted in Germany using baseline prostate-specific antigen (PSA) starting in young men (PROBASE). INTERVENTION: PSA measurements starting at the age of 45 yr. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: The incidence of PCa within 5 yr was assessed in men with screen-negative baseline PSA <1.5 ng/ml compared with those with PSA 1.5-≤3.0 ng/ml. RESULTS AND LIMITATIONS: Of 23301 men who received a first PSA test at age 45 yr, 0.79% had a screen-positive PSA value of ≥3 ng/ml. Among the 89% of men who had a screen-negative baseline PSA value of <1.5 ng/ml, only 0.45% received a positive PSA test ≥3 ng/ml upon retesting after 5 yr. By contrast, for those with a screen-negative baseline PSA value of 1.5-3 ng/ml, 13% surpassed 3 ng/ml upon biennial testing within the next 4 yr. The incidence of PCa in subsequent screening rounds increased with increasing baseline PSA levels, from 0.13 per 1000 person-years for men with initial PSA level of <1.5 ng/ml to 8.0 per 1000 person-years for those with PSA levels of 1.5-3.0 ng/ml. A limitation is a follow-up time of only 5 yr, so far. CONCLUSIONS: Men with baseline PSA <1.5 ng/ml at age 45 yr are at a very low risk of PCa over the next 5 yr. PATIENT SUMMARY: The PROBASE study showed that men with baseline prostate-specific antigen (PSA) <1.5 ng/ml at age 45 yr have a very low prostate cancer detection rate over 5 yr and do not need PSA retesting during this time.

Comorbidity in limited disease small-cell lung cancer: Age-adjusted Charlson comorbidity index and its association with overall survival following chemoradiotherapy.

Background: Combined, platinum-based thoracic chemoradiotherapy (TCR) is the current state-of-the-art treatment for patients with limited disease (LD) small-cell lung cancer (SCLC). There is only limited data available regarding the effect of comorbidities on survival following TRC. The purpose of this study is to assess the age-adjusted Charlson comorbidity index (ACCI) as a predictor of overall survival in LD-SCLC patients undergoing TCR. Patients and methods: We retrospectively analyzed 367 SCLC patients diagnosed with LD-SCLC who received TCR between 2003 and 2017. We evaluated the ACCI (n = 348) as a predictor of overall survival (OS). In this cohort, 322 patients (88%) received platinum-based TCR (either cisplatin or carboplatin), and 37 (10%) patients received vincristine based TCR. Median radiation dose was 60 Gy (range 24-66 Gy). Additionally, 83% of patients (n = 303) received prophylactic cranial irradiation (PCI, 30 Gy in 2 Gy fractions). Kaplan-Meier survival analysis was performed for OS. For comparison of survival curves, Log-rank (Mantel-Cox) test was used. Univariate and multivariate Cox proportional-hazards ratios (HRs) were used to assess the influence of cofactors on OS. Results: Patients with an ACCI > 6 had a significantly shorter OS compared with patients with an ACCI ≤ 6 (median 11 vs. 20 months; p = 0.005). Univariate analysis for OS revealed a statistically significant effect for ACCI > 6 (HR 1.7; 95% CI 1.2-2.4; p = 0.003), PCI (HR 0.5; 95% CI 0.3-0.7; p < 0.001), and Karnofsky performance status ≤ 70% (KPS) (HR 1.4; 95% CI 1.1-1.90; p = 0.015). In multivariate analysis, OS was significantly associated with PCI (HR 0.6; 95% CI 0.4-0.9; p = 0.022) and ACCI > 6 (HR 1.5; 95% CI 1.0-2.1; p = 0.049). Conclusion: Comorbidity is significantly associated with survival in patients with LD-SCLC undergoing TCR. The ACCI may be a valuable tool to identify patients with a shorter survival and thus might be used for risk stratification and oncological decision making.

Financial toxicity in cancer patients undergoing radiotherapy in a universal health care system - A prospective multicenter study of 1075 patients.

PURPOSE: To establish and confirm prevalence as well as risk factors of financial toxicity in a large national cohort of cancer patients undergoing radiotherapy in a universal health care system. METHODS: We conducted a prospective cross-sectional study offering a patient-reported questionnaire to all eligible cancer patients treated with radiotherapy in 11 centers in Germany during 60 consecutive days. The four-point subjective financial distress question of the EORTC QLQ-C30 was used as a surrogate for financial toxicity. Confirmatory hypothesis testing evaluated the primary study outcomes: overall prevalence of financial toxicity and its association with predefined risk factors. P-values < 0.05 were considered statistically significant. RESULTS: Of 2341 eligible patients, 1075 (46%) participated. The prevalence of subjective financial distress (=any grade higher than not present) was 41% (438/1075) exceeding the hypothesized range of 26.04-36.31%. Subjective financial distress was felt "A little" by 26% (280/1075), "Quite a bit" by 11% (113/1075) and "Very much" by 4% (45/1075) of the patients. Lower household income, lower global health status/ quality of life, higher direct costs and higher loss of income significantly predicted higher subjective financial distress per ordinal regression and confirmed these risk factors. Higher psychosocial distress and lower patient satisfaction were significantly associated with higher subjective financial distress in an exploratory ordinal regression model. CONCLUSION: The overall prevalence of financial toxicity was higher than anticipated, although reported at low or moderate degrees by most affected patients. As we confirmed risk factors associated with financial toxicity, patients at risk should be addressed early for potential support.

Psychosocial distress in cancer patients undergoing radiotherapy: a prospective national cohort of 1042 patients in Germany.

PURPOSE: Psychosocial distress is common among cancer patients in general, but those undergoing radiotherapy may face specific challenges. Therefore, we investigated the prevalence and risk factors for distress in a large national cohort. METHODS: We performed a secondary analysis of a multicenter prospective cross-sectional study which surveyed cancer patients at the end of a course of radiotherapy using a patient-reported questionnaire. Distress was measured with the distress thermometer (DT), using a cut-off of ≥ 5 points for clinically significant distress. Univariate analyses and multivariate multiple regression were used to assess associations of distress with patient characteristics. A two-sided p-value < 0.05 was considered statistically significant. RESULTS: Out of 2341 potentially eligible patients, 1075 participated in the study, of which 1042 completed the DT. The median age was 65 years and 49% (511/1042) of patients were female. The mean DT score was 5.2 (SD = 2.6). Clinically significant distress was reported by 63% (766/1042) of patients. Of the patient characteristics that were significantly associated with distress in the univariate analysis, a lower level of education, a higher degree of income loss, lower global quality of life, and a longer duration of radiotherapy in days remained significantly associated with higher distress in the multivariate analysis. Yet effect sizes of these associations were small. CONCLUSION: Nearly two in three cancer patients undergoing radiotherapy reported clinically significant distress in a large multicenter cohort. While screening and interventions to reduce distress should be maintained and promoted, the identified risk factors may help to raise awareness in clinical practice. TRIAL REGISTRY IDENTIFIER: DRKS: German Clinical Trial Registry identifier: DRKS00028784.

Heidelberger Interprofessionelle Ausbildungsstation (HIPSTA): a practice- and theory-guided approach to development and implementation of Germany's first interprofessional training ward.

Background: Deficits in care and impaired patient-safety have been linked to inefficient interprofessional collaborative practice. Interprofessional training wards (IPTW) are an interprofessional educational intervention which aim to enable students and trainees from different health professions to work self-responsibly in order to manage the medical treatment and rehabilitation of real-life patients together as an interprofessional team. We aimed to develop and implement Germany´s first IPTW at the department of Surgery at Heidelberg University Hospital. Methods: The Kern cycle was used to develop an ITPW curriculum. Practical as well as theoretical considerations guided the design of the IPTW. Common project management tools including blueprinting and RASCI (Responsibility, Approval, Support, Consultation, Information) matrix were applied. Results: Since April 2017, 7 cohorts of students and trainees have had four-week long placements on HIPSTA. They run the IPTW in early and late shifts. Nursing and medical facilitators are supporting the IP team as needed. Learning objectives are operationalized as EPAs (entrustable professional activities) and interprofessional learning goals. Since initiation only minor modifications to the curriculum have been necessary and satisfaction of students/trainees, facilitators and patients is high. Conclusion: IPTWs can be established and run in the German health care system even in a complex clinical setting. The early involvement of all professions in a steering group seems to be key to success. Nursing and medical facilitators are of utmost importance for daily routine. The experiences outlined here could help others aiming to implement IPTWs at their sites. IPTWs might address a number of hitherto unaddressed educational needs. Trial registration: Not applicable.

Encouraging student-driven clinical research in Germany: the CHIR-Net SIGMA network.

Evidence should define and guide modern clinical care, yet many relevant questions in surgical practice remain unconfirmed by substantial data. Evidence-based medicine requires both the implementation of its principles in day-to-day work and the acquisition of new evidence preferably by randomized controlled trials and systematic reviews. Meaningful clinical research, however, is challenging to conduct, and its overall infrastructure in Germany was, until recently, considered poor compared to other leading countries. Although this has been significantly improved after the establishment of the Study Center of the German Surgical Society (SDGC) and the surgical clinical trial network CHIR-Net, limited focus has been put on the training, teaching, and recruitment of medical students to become competent clinical researchers and clinician scientists. To ensure continuing comprehensive clinical research in surgery, CHIR-Net aims to establish a student-driven multicenter research network in Germany, which is embedded in both the national CHIR-Net and the pan-European and international frameworks. Student-Initiated German Medical Audits (SIGMA) is a product of the strong collaboration between clinical scientists and medical trainees, enabling students to contribute to high-quality clinical trials. Additionally, participants are offered extensive training to support the next generation of research-active clinicians. Starting on 2018, SIGMA will perform its first multicenter observational study in Germany.