JNCI Cancer Spectrum最新文献

Background: Benign breast disease (BBD) is an important risk factor for breast cancer (BC) development. In this study, we analyzed hematoxylin and eosin-stained whole slide images (WSIs) from diagnostic BBD biopsies using different deep learning (DL) approaches to predict those who subsequently developed breast cancer (cases) and those who did not (controls).

Methods: We randomly divided cases and controls from a nested case-control study of 946 women with BBD into training (331 cases, 331 controls) and test (142 cases, 142 controls) sets. We employed customized VGG-16 and AutoML models for image-only classification using WSIs; logistic regression for classification using only clinico-pathological characteristics; and a multimodal network combining WSIs and clinico-pathological characteristics for classification.

Results: Both image-only (area under the receiver operating characteristic curve, AUROCs of 0.83 (standard error, SE: 0.001) and 0.78 (SE: 0.001) for customized VGG-16 and AutoML, respectively)) and multimodal (AUROC of 0.89 (SE: 0.03)) networks had high discriminatory accuracy for BC. The clinico-pathological characteristics only model had the lowest AUROC of 0.54 (SE: 0.03). Additionally, compared to the customized VGG-16 which performed better than AutoML, the multimodal network had improved accuracy, 0.89 (SE: 0.03) vs 0.83 (SE: 0.02), sensitivity, 0.93 (SE: 0.04) vs 0.83 (SE: 0.003), and specificity, namely 0.86 (SE: 0.03) vs 0.84 (SE: 0.003).

Conclusion: This study opens promising avenues for BC risk assessment in women with benign breast disease. Integrating whole slide images and clinico-pathological characteristics through a multimodal approach significantly improved predictive model performance. Future research will explore DL techniques to understand BBD progression to invasive BC.

Background: The PRO-CTCAE Measurement System was designed to enhance the quality of the standard toxicity evaluation in clinical trials. We developed a sub-study within NRG BR004, a phase III clinical trial in patients with newly documented HER2-positive metastatic breast cancer (MBC), to examine the added value and feasibility of frequent PRO-CTCAE data collection.

Methods: Patients were asked to complete 23 PRO-CTCAE items assessing 12 symptoms. Electronic PRO (ePRO) reporting was preferred; however, paper administration was allowed. The data on items assessed before treatment initiation, then weekly during Cycles 1-2 (12 weeks) are presented herein. Feasibility of frequent assessment with ePRO reporting was assessed using these data and was pre-defined as ≥ 25% of patients being compliant (submitted ≥75% of scheduled assessments). We also examined PRO-CTCAE and clinician-reported CTCAE data for key symptoms using maximum toxicity grade and the toxicity index (TI).

Results: Overall, 80% of patients (82 of 103) were compliant with expected weekly assessments (90% CI : 0.72-0.86). For all symptoms, the median maximum grade (TI value) of clinician-reported CTCAE was lower than the median maximum score (TI value) of patient-reported PRO-CTCAE. The differences in the data trend for weekly vs less frequent assessment were more apparent when data were evaluated using the TI vs the maximum score.

Conclusions: Weekly assessments within the first two chemotherapy cycles were feasible in this trial of MBC patients. As expected, patients reported greater severity of symptoms than clinicians. Demonstrating the feasibility of frequent assessment could have implications for future research and clinical practice.

Clinicaltrials.gov: NCT03199885 (https://clinicaltrials.gov/study/NCT03199885).

Background: Early Phase Clinical Trials (EP-CTs) continue to have significant difficulty with enrolling real-world populations with many minorities being underrepresented. Reasons for this include patient or clinician perception as well as cultural, linguistic or social barriers. In Australia, there is currently no prospective data in the EP-CT space regarding recruitment of priority populations.

Methods: PEARLER (Patient divErsity in eARLy phasE clinical tRials) was a multi-centre, prospective, cohort study involving two major EP-CT centres in Sydney, Australia. All participants who were consented to an EP-CT between August 2023 and August 2024 were enrolled. Participants completed a baseline demographic survey which included cultural and linguistic status, sexual orientation, socioeconomic status and regional diversity.

Results: One-hundred and fourteen participants were recruited. Median age was 63 years (25-83 years) with predominance for female participants (52%). No participant reported a non-binary gender. All participants reported their sexuality as heterosexual with no LGBTQIA+ participants recruited. 34 participants (30%) were identified as culturally diverse whilst 28 (25%) were linguistically diverse. One patient identified as indigenous Australian. 26% of participants were born overseas with 44% having at least one parent born overseas. The majority were living in households with family members with 8% of participants living alone.

Conclusion: PEARLER is the first prospective study that provides granular description of social, cultural, linguistic, economic and sexual diversity amongst EP-CT participants. Certain subgroups are under-represented including those with sexual diversity, gender diversity and indigenous backgrounds. Ongoing efforts to monitor and promote inclusion of diverse populations in clinical trials are vital.

Background: Patients' self-report of their symptoms can provide important data for the evaluation of treatment benefit and tolerability of oncology drugs. Contemporary treatment approaches, including immunotherapy and molecular targeted therapies, have unique toxicities based on their novel mechanisms of action. This scoping review aimed to summarize evidence from existing reviews and clinical practice guidelines to examine the type and prevalence of toxicities including symptomatic adverse events (sympAEs) for adult cancer patients to inform clinical care and therapeutic trials.

Methods: A systematic search of PubMed, Web of Science, and Embase was performed using predefined eligibility criteria. Thirty-one literature reviews and 3 clinical practice guidelines met inclusion criteria and were selected for review and data abstraction.

Results: Findings from this scoping review demonstrated several leading sympAEs that were reported across immunotherapy and targeted therapy drugs, including fatigue, diarrhea and rash. In addition to these more prevalent sympAEs, there were some less frequently reported class-specific sympAEs, which had potential for significant harm or disability to the patient if not properly identified and treated. Many studies reported toxicities as AEs or syndromes solely using data reported by clinicians without additional self-report from patients.

Conclusion: We identified several core sympAEs experienced by patients participating in oncology trials using immunotherapy and targeted therapy agents, which has implications for future trial design and drug labeling. Future cancer trials should assess patient-reported sympAEs based on identified drug mechanism to inform the tolerability of these newer agents and enhance patient safety during trial participation and clinical care.

Background: Data on prospective associations of accelerometer-measured physical activity (PA), sedentary behavior (SB), and mortality among cancer survivors are lacking. Our study examined accelerometer-measured daily PA (including light, moderate-to-vigorous PA [MVPA], total PA, and steps), SB (sitting time and mean bout duration), and mortality among cancer survivors in the Women's Health Accelerometry Collaboration (WHAC).

Methods: Postmenopausal women in WHAC who reported a cancer diagnosis ≥1 year prior to wearing an ActiGraph GT3X+ on the hip for ≥4 of 7 days from 2011-2015 were included. Outcomes included all-cause, cancer, and cardiovascular disease (CVD) mortality. Covariate-adjusted Cox regression estimated hazard ratios (HRs) and 95% confidence intervals (CIs) for each PA and SB measure in association with mortality.

Results: Overall, 2,479 cancer survivors (mean [SD] age, 74.2 [6.7] years) were followed for 8.3 years. For all-cause mortality (n = 594 cases), every 78.1 min/day in light PA, 96.5 min/day in total PA, 102.2 min/day in sitting time, and 4.8 min in sitting bout duration had HRs (95%CIs) of 0.92 (0.84-1.01), 0.89 (0.80-0.98), 1.12 (1.02-1.24) and 1.04 (0.96-1.12), respectively. Linear associations for cancer mortality (n = 168) and CVD mortality (n = 109) were not statistically significant except for steps (HR per 2,469 steps/day=0.66, 95%CI: 0.45-0.96) and sitting time (HR = 1.30, 95%CI: 1.02-1.67) for CVD mortality. Nonlinear associations showed benefits of MVPA (for all-cause and CVD mortality) and steps (all-cause mortality only) maximized around 60 min/day and 5,000 steps/day, respectively.

Conclusions: Among postmenopausal cancer survivors, higher PA and lower SB was associated with reduced hazards of all-cause and CVD mortality.

Objective: We use real-world data to develop a lung cancer screening (LCS) eligibility mechanism that is both accurate and free from racial bias.

Methods: Our data came from the Prostate, Lung, Colorectal, and Ovarian (PLCO) cancer screening trial. We built a systematic fairness-aware machine learning framework by integrating a Group and Intersectional Fairness and Threshold (GIFT) strategy with an easy ensemble classifier- (EEC-) or logistic regression- (LR-) based model. The best LCS eligibility mechanism EEC-GIFT* and LR-GIFT* were applied to the testing dataset and their performances were compared to the 2021 US Preventive Services Task Force (USPSTF) criteria and PLCOM2012 model. The equal opportunity difference (EOD) of developing lung cancer between Black and White smokers was used to evaluate mechanism fairness.

Results: The fairness of LR-GIFT* or EEC-GIFT* during training was notably greater than that of the LR or EEC models without greatly reducing their accuracy. During testing, the EEC-GIFT* (85.16% vs 78.08%, P < .001) and LR-GIFT* (85.98% vs 78.08%, P < .001) models significantly improved sensitivity without sacrificing specificity compared to the 2021 USPSTF criteria. The EEC-GIFT* (0.785 vs 0.788, P = .28) and LR-GIFT* (0.785 vs 0.788, P = .30) showed similar area under receiver operating characteristic curve (AUC) values compared to the PLCOM2012 model. While the average EODs between Blacks and Whites were significant for the 2021 USPSTF criteria (0.0673, P < .001), PLCOM2012 (0.0566, P < .001), and LR-GIFT* (0.0081, P < .001), the EEC-GIFT* model was unbiased (0.0034, P = .07).

Conclusion: Our EEC-GIFT* LCS eligibility mechanism can significantly mitigate racial biases in eligibility determination without compromising its predictive performance.

Background: Addressing critical gaps in precision medicine initiatives in colorectal cancer (CRC) requires building larger collaborative studies.

Methods: The Latino Colorectal Cancer Consortium (LC3) is a resource that harmonizes data collected in observational studies with data from individuals who identify as Hispanic/Latino with a diagnosis of primary colorectal adenocarcinoma. Data collected includes demographics, medical history, family history, and lifestyle risk factors from patient-completed surveys. Vital status, cause of death, treatment, and clinicopathological characteristics were obtained through medical chart abstraction, pathology reports and/or linkage to state cancer registries. Blood, saliva, or normal colonic tissues were used to extract and genotype germline DNA. Tumor tissue (snap frozen or formalin-fixed paraffin-embedded) were evaluated by pathologists for diagnosis, tissue content, tumor cellularity, necrosis, immune infiltration, and additional histopathologic characteristics. A centralized database with a virtual tumor repository was created to facilitate collaborative research.

Results: As of April 2024, LC3 assembled data from 2,210 patients (diagnosed 1994 to 2023). The mean age at diagnosis was 57 (range: 19-93) years; 54.3% of participants were male, and 62.0% had been diagnosed with colon cancer. Surveys were completed by 1,722 (77.8%) participants. Ongoing multi-omics profiling on up to 600 patients include: genome-wide germline genotyping, paired tumor/normal whole exome sequencing, bulk RNA-seq, T cell receptor immunosequencing, and multiplex immunofluorescence.

Conclusions: This consortium fills an important gap in research infrastructure in CRC as well as improving precision medicine initiatives for all individuals.

Background: Colorectal cancer survivors face increased risks of developing new primary cancers in colorectum and other anatomical sites. This systematic review aimed to estimate primary colorectal and extracolonic cancers risks following colorectal cancer.

Methods: Peer-reviewed articles published before January 2025 were screened across four databases to identify studies using population cancer registry reporting standardized incidence ratios (SIRs) of primary cancers following colorectal cancer, compared with the general population. A meta-analysis was conducted to summarize the SIRs, and age-specific cumulative risks of primary cancers following colorectal cancer were estimated using the summarized SIRs and age-, sex-, calendar-, region- and cancer-specific incidence data.

Results: Of 8254 articles identified, 57 were included in meta-analysis. The pooled SIRs (95% confidence interval) for any primary cancer, extracolonic cancer and colorectal cancer were 1.13 (1.06-1.20), 1.10 (1.03-1.17), and 1.55 (1.33-1.77), respectively. Increased risks were also observed for primary cancers of small intestine, ovary, uterus, testes, kidney, female breast, thyroid, and prostate overall, as well as for lung and urinary bladder cancer in recent studies. The cumulative risks of any primary cancer, extracolonic cancer, and colorectal cancer to age 75 years were: 38.5%, 31.6%, and 8.24% in Australasia; 33.8%, 30.9%, and 4.77% in North America; 27.4%, 25.6%, and 8.01% in East Asia; and 33.4%, 28.8%, and 4.68% in Europe.

Conclusion: Colorectal cancer survivors have an increased risk of subsequent primary cancers, both extracolonic and colorectal, when compared with the general population. These findings underscore the necessity for tailored surveillance and prevention strategies to effectively identify and manage subsequent primary cancers in this population.

Colorectal cancer (CRC) is a major global health challenge. Emerging research highlights the pivotal role of the gut microbiota in influencing CRC risk, progression, and treatment response. Metagenomic approaches, especially high-throughput shotgun sequencing, have provided unprecedented insights into the intricate connections between the gut microbiome and CRC. By enabling comprehensive taxonomic and functional profiling, metagenomics has revealed microbial signatures, activities, and biomarkers associated with colorectal tumorigenesis. Furthermore, metagenomics has shown a potential to guide patient stratification, predict treatment outcomes, and inform microbiome-targeted interventions. Despite remaining challenges in multi-omics data integration, taxonomic gaps, and validation across diverse cohorts, metagenomics has propelled our comprehension of the intricate gut microbiome-CRC interplay. This review underscores the clinical relevance of microbial signatures as potential diagnostic and prognostic tools in CRC. Furthermore, it discusses personalized treatment strategies guided by this omics' approaches.

Background: Adequate patient awareness and understanding of cancer clinical trials is essential for trial recruitment, informed decision making, and protocol adherence. Although large language models (LLMs) have shown promise for patient education, their role in enhancing patient awareness of clinical trials remains unexplored. This study explored the performance and risks of LLMs in generating trial-specific educational content for potential participants.

Methods: Generative Pretrained Transformer 4 (GPT4) was prompted to generate short clinical trial summaries and multiple-choice question-answer pairs from informed consent forms from ClinicalTrials.gov. Zero-shot learning was used for summaries, using a direct summarization, sequential extraction, and summarization approach. One-shot learning was used for question-answer pairs development. We evaluated performance through patient surveys of summary effectiveness and crowdsourced annotation of question-answer pair accuracy, using held-out cancer trial informed consent forms not used in prompt development.

Results: For summaries, both prompting approaches achieved comparable results for readability and core content. Patients found summaries to be understandable and to improve clinical trial comprehension and interest in learning more about trials. The generated multiple-choice questions achieved high accuracy and agreement with crowdsourced annotators. For both summaries and multiple-choice questions, GPT4 was most likely to include inaccurate information when prompted to provide information that was not adequately described in the informed consent forms.

Conclusions: LLMs such as GPT4 show promise in generating patient-friendly educational content for clinical trials with minimal trial-specific engineering. The findings serve as a proof of concept for the role of LLMs in improving patient education and engagement in clinical trials, as well as the need for ongoing human oversight.