JMIR Cancer最新文献

Background: Breast cancer subtyping is a crucial step in determining therapeutic options, but the molecular examination based on immunohistochemical staining is expensive and time-consuming. Deep learning opens up the possibility to predict the subtypes based on the morphological information from hematoxylin and eosin staining, a much cheaper and faster alternative. However, training the predictive model conventionally requires a large number of histology images, which is challenging to collect by a single institute.

Objective: We aimed to develop a data-efficient computational pathology platform, 3DHistoNet, which is capable of learning from z-stacked histology images to accurately predict breast cancer subtypes with a small sample size.

Methods: We retrospectively examined 401 cases of patients with primary breast carcinoma diagnosed between 2018 and 2020 at the Department of Pathology, National Cancer Center, South Korea. Pathology slides of the patients with breast carcinoma were prepared according to the standard protocols. Age, gender, histologic grade, hormone receptor (estrogen receptor [ER], progesterone receptor [PR], and androgen receptor [AR]) status, erb-B2 receptor tyrosine kinase 2 (HER2) status, and Ki-67 index were evaluated by reviewing medical charts and pathological records.

Results: The area under the receiver operating characteristic curve and decision curve were analyzed to evaluate the performance of our 3DHistoNet platform for predicting the ER, PR, AR, HER2, and Ki67 subtype biomarkers with 5-fold cross-validation. We demonstrated that 3DHistoNet can predict all clinically important biomarkers (ER, PR, AR, HER2, and Ki67) with performance exceeding the conventional multiple instance learning models by a considerable margin (area under the receiver operating characteristic curve: 0.75-0.91 vs 0.67-0.8). We further showed that our z-stack histology scanning method can make up for insufficient training data sets without any additional cost incurred. Finally, 3DHistoNet offered an additional capability to generate attention maps that reveal correlations between Ki67 and histomorphological features, which renders the hematoxylin and eosin image in higher fidelity to the pathologist.

Conclusions: Our stand-alone, data-efficient pathology platform that can both generate z-stacked images and predict key biomarkers is an appealing tool for breast cancer diagnosis. Its development would encourage morphology-based diagnosis, which is faster, cheaper, and less error-prone compared to the protein quantification method based on immunohistochemical staining.

Background: Medication adherence is crucial for improving clinical outcomes in the treatment of patients with cancer. The lack of adherence and adverse drug reactions can reduce the effectiveness of cancer therapy including the quality of life. The commonly used intervention methods for medication adherence continue to evolve, and the age of fifth-generation (5G) messaging has arrived.

Objective: In this study, we conducted a prospective, pilot randomized controlled trial to evaluate the effect of 5G messaging on medication adherence and clinical outcomes among patients with cancer in China.

Methods: The research population was patients with nonsmall cell lung cancer undergoing pemetrexed chemotherapy who require regular folic acid (FA) and vitamin B12 supplements. The intervention and control groups were assigned to 5G messaging and second-generation (2G) messaging, respectively. The patients' medication adherence and quality of life were assessed at baseline and 1-month and 3-month time points. Moreover, the chemotherapy-related hematologic or nonhematologic toxicities, as well as the serum levels of FA and vitamin B12, were measured.

Results: Of the 567 patients assessed for eligibility between January and May 2021, a total of 154 (27.2%) patients were included. Overall, 80 were randomized to the control group and 74 to the intervention group. The odds of adherence in the 5G messaging intervention group were significantly higher than the control group at the 1-month (62/69, 90% vs 56/74, 76%; adjusted odds ratio 2.67, 95% CI 1.02-7.71) and 3-month (50/60, 83% vs 48/64, 75%; adjusted odds ratio 2.36, 95% CI 1.00-5.23) time points. Correspondingly, the FA and vitamin B12 serum levels of patients in the 5G messaging group were higher than those of the control group. Regarding hematologic toxicities, only the incidence of leukopenia in the intervention group was lower than that in the control group (25/80, 31% in the control group vs 12/74, 16% in the intervention group; P=.04). There were no differences in nonhematologic toxicities and quality of life between the 2 groups.

Conclusions: In summary, we conclude that compared with conventional 2G text-based messaging, a 5G messaging intervention can better improve medication adherence and clinical outcome among patients with cancer.

Trial registration: Chinese Clinical Trial Registry ChiCTR2200058188; https://www.chictr.org.cn/showproj.html?proj=164489.

Background: Cancer is related to not only physical but also mental suffering. Notably, body image disturbances are highly relevant to cancer-related changes often persisting beyond recovery from cancer. Scalable and low-barrier interventions that can be blended with face-to-face psychotherapy for cancer survivors are highly warranted.

Objective: The aim of the study is to investigate whether smartphone-based bodily interventions are more effective to improve the mood of patients with cancer than smartphone-based fairy tale interventions (control intervention).

Methods: We recruited patients with cancer in 2 Swiss hospitals and conducted daily, fully automated smartphone-based interventions 6 times a week for 5 consecutive weeks, blended with weekly face-to-face group body psychotherapy. We applied 2 types of smartphone-based interventions using a within-subject design, randomly assigning patients daily to either bodily interventions or fairy tales. Each intervention type was presented 3 times a week. For this secondary analysis, 3-level mixed models were estimated with mood assessed by the 3 Multidimensional Mood Questionnaire subscales for good-bad mood, wakefulness, and calmness as key indicators. In addition, the effects on experience of presence, vitality, and burden assessed with visual analog scales were investigated.

Results: Based on the data from s=732 interventions performed by 36 participants, good-bad mood improved (β=.27; 95% CI 0.062-0.483), and participants became calmer (β=.98; 95% CI 0.740-1.211) following smartphone-based interventions. Wakefulness did not significantly change from pre- to postsmartphone-based intervention (β=.17; 95% CI -0.081 to 0.412). This was true for both intervention types. There was no interaction effect of intervention type with change in good-bad mood (β=-.01; 95% CI -0.439 to 0.417), calmness (β=.22; 95% CI -0.228 to 0.728), or wakefulness (β=.14; 95% CI -0.354 to 0.644). Experience of presence (β=.34; 95% CI 0.271-0.417) and vitality (β=.35; 95% CI 0.268-0.426) increased from pre- to postsmartphone-based intervention, while experience of burden decreased (β=-0.40; 95% CI -0.481 to 0.311). Again, these effects were present for both intervention types. There were no significant interaction effects of intervention type with pre- to postintervention changes in experience of presence (β=.14; 95% CI -0.104 to 0.384), experience of vitality (β=.06; 95% CI -0.152 to 0.265), and experience of burden (β=-.16; 95% CI -0.358 to 0.017).

Conclusions: Our results suggest that both smartphone-based audio-guided bodily interventions and fairy tales have the potential to improve the mood of cancer survivors.

Trial registration: ClinicalTrials.gov NCT03707548; https://clinicaltrials.gov/study/NCT03707548.

International registered report identifier (irrid): RR2-10.

Background: Appropriate annual screenings for colorectal cancer (CRC) are an essential preventive measure for the second-leading cause of cancer-related death in the United States. Studies have shown that CRC screening rates are influenced by various social determinants of health (SDOH) factors, including race, ethnicity, and geography. According to 2018 national data, participation in screening is lowest among Hispanic or Latinx individuals (56.1%). At an urban Federally Qualified Health Center, a quality improvement project was conducted to evaluate a texting program with a motivational fotonovela-a short narrative comic. Fotonovelas have previously been used in programs to improve knowledge of cervical cancer and human papillomavirus, vaccinations, and treatments for depression.

Objective: This study aimed to encourage compliance with fecal immunochemical test (FIT) screening. Patient engagement involved a texting program with fotonovelas informed by behavior change techniques. This study sought to understand the qualitative characteristics of patient motivation, intention, and barriers to completing their screening.

Methods: A total of 5241 English-speaking or Spanish-speaking Federally Qualified Health Center patients aged 50 to 75 years were randomized to either intervention (a 4-week tailored 2-way texting program with a fotonovela comic) or usual care (an SMS text message reminder and patient navigator phone call). The texting vendor used a proprietary algorithm to categorize patients in the intervention group into SDOH bands based on their home addresses (high impact=high social needs and low impact=low social needs). Over 4 weeks, patients were texted questions about receiving and returning their FIT, what barriers they may be experiencing, and their thoughts about the fotonovela.

Results: The SDOH index analysis showed that most of the patient population was in the SDOH band categories of high impact (555/2597, 21.37%) and very high impact (1416/2597, 54.52%). Patients sent 1969 total responses to the texting system. Thematic analysis identified 3 major themes in these responses: messages as a reminder, where patients reported that they were motivated to return the FIT and had already done so or would do so as soon as possible; increasing patients' understanding of screening importance, where patients expressed an increased knowledge about the purpose and importance of the FIT; and expressing barriers, where patients shared reasons for not completing the FIT.

Conclusions: The texting program and fotonovela engaged a subset of patients in each SDOH band, including the high and very high impact bands. Creating culturally tailored messages can encourage patient engagement for accepting the content of the messaging, confirming intentions to complete their FIT, and sharing insights about barriers to behavior change. To better supp

Background: Twitter has become a popular platform for individuals to broadcast their daily experiences and opinions on a wide range of topics and emotions. Tweets from patients with cancer could offer insights into their needs. However, limited research has been conducted using Twitter data to understand the needs of patients with cancer despite the substantial amount of health-related data posted on the platform daily.

Objective: This study aimed to uncover the potential of using Twitter data to understand the perspectives and experiences of patients with thyroid cancer at a global level.

Methods:  This retrospective descriptive study collected tweets relevant to thyroid cancer in 2020 using the Twitter scraping tool. Only English-language tweets were included, and data preprocessing was performed to remove irrelevant tweets, duplicates, and retweets. Both tweets and Twitter users were manually classified into various groups based on the content. Each tweet underwent sentiment analysis and was classified as either positive, neutral, or negative.

Results: A total of 13,135 tweets related to thyroid cancer were analyzed. The authors of the tweets included patients with thyroid cancer (3225 tweets, 24.6%), patient's families and friends (2449 tweets, 18.6%), medical journals and media (1733 tweets, 13.2%), health care professionals (1093 tweets, 8.3%), and medical health organizations (940 tweets, 7.2%), respectively. The most discussed topics related to living with cancer (3650 tweets, 27.8%), treatment (2891 tweets, 22%), diagnosis (1613 tweets, 12.3%), risk factors and prevention (1137 tweets, 8.7%), and research (953 tweets, 7.3%). An average of 36 tweets pertaining to thyroid cancer were posted daily. Notably, the release of a film addressing thyroid cancer and the public disclosure of a news reporter's personal diagnosis of thyroid cancer resulted in a significant escalation in the volume of tweets. From the sentiment analysis, 53.5% (7025/13,135) of tweets were classified as neutral statements and 32.7% (4299/13,135) of tweets expressed negative emotions. Tweets from patients with thyroid cancer had the highest proportion of negative emotion (1385/3225 tweets, 42.9%), particularly when discussing symptoms.

Conclusions:  This study provides new insights on using Twitter data as a valuable data source to understand the experiences of patients with thyroid cancer. Twitter may provide an opportunity to improve patient and physician engagement or apply as a potential research data source.

Background: Mobile health (mHealth) tools were developed during the past decades and are increasingly used by patients in cancer care too. Scientific research in the development of mHealth services is required in order to meet the various needs of patients and test usability.

Objective: The aim of this study is to assess patients' needs, preferences, and usability of an app (My University Clinic [MUC] app) developed by the Comprehensive Cancer Center Freiburg (CCCF) Germany.

Methods: Based on a qualitative cross-sectional approach, we conducted semistructured interviews with patients with cancer, addressing their needs, preferences, and usability of the designed MUC app. Patients treated by the CCCF were recruited based on a purposive sampling technique focusing on age, sex, cancer diagnoses, and treatment setting (inpatient, outpatient). Data analysis followed the qualitative content analysis according to Kuckartz and was performed using computer-assisted software (MAXQDA).

Results: For the interviews, 17 patients with cancer were selected, covering a broad range of sampling parameters. The results showed that patients expect benefits in terms of improved information about the disease and communication with the clinic staff. Demands for additional features were identified (eg, a list of contact persons and medication management). The most important concerns referred to data security and the potential restriction of personal contacts with health care professionals of the clinical departments of the CCCF. In addition, some features for improving the design of the MUC app with respect to usability or for inclusion of interacting tools were suggested by the patients.

Conclusions: The results of this qualitative study were discussed within the multidisciplinary team and the MUC app providers. Patients' perspectives and needs will be included in further development of the MUC app. There will be a second study phase in which patients will receive a test version of the MUC app and will be asked about their experiences with it.

Trial registration: Deutsches Register Klinischer Studien DRKS00022162; https://drks.de/search/de/trial/DRKS00022162.

Background: Patients with cancer require adequate preparation in self-management of treatment toxicities to reduce morbidity that can be achieved through well-designed digital technologies that are developed in co-design with patients and end users.

Objective: We undertook a user-centered co-design process in partnership with patients and other knowledge end users to develop and iteratively test an evidence-based and theoretically informed web-based cancer self-management program (I-Can Manage). The specific study aims addressed in 2 phases were to (1) identify from the perspective of patients with cancer and clinicians the desired content, features, and functionalities for an online self-management education and support (SMES) program to enable patient self-management of treatment toxicities (phase 1); (2) develop the SMES prototype based on human-centered, health literate design principles and co-design processes; and (3) evaluate usability of the I-Can Manage prototype through user-centered testing (phase 2).

Methods: We developed the I-Can Manage program using multiperspective data sources and based on humanistic and co-design principles with end users engaged through 5 phases of development. We recruited adult patients with lung, colorectal, and lymphoma cancer receiving systemic treatments from ambulatory clinics in 2 regional cancer programs for the qualitative inquiry phase. The design of the program was informed by data from qualitative interviews and focus groups, persona and journey mapping, theoretical underpinnings of social cognitive learning theory, and formalized usability testing using a cognitive think-aloud process and user satisfaction survey. A co-design team comprising key stakeholders (human design experts, patients/caregiver, clinicians, knowledge end users, and e-learning and digital design experts) was involved in the developmental process. We used a cognitive think-aloud process to test usability and participants completed the Post-Study System Usability Questionnaire (PSSUQ).

Results: In the initial qualitative inquiry phase, 16 patients participated in interviews and 19 clinicians participated in interviews or focus groups and 12 key stakeholders participated in a persona journey mapping workshop to inform development of the program prototype. The I-Can Manage program integrates evidence-based information and strategies for the self-management of treatment toxicities and health-promoting behaviors in 6 e-learning modules (lay termed "chapters"), starting with an orientation to self-management. Behavioral exercises, patient written and video stories, downloadable learning resources, and online completion of goals and action plans were integrated across chapters. Patient participants (n=5) with different cancers, gender, and age worked through the program in the human factors laboratory using a cognitive think-aloud process and all

Background: The treatment of non-small cell lung cancer (NSCLC) has evolved dramatically with the approval of immuno-oncology (IO) and targeted therapies (TTs). Insights on the patient experience with these therapies and their impacts are lacking. Health-related social media has been increasingly used by patients to share their disease and treatment experiences, thus representing a valuable source of real-world data to understand the patient's voice and uncover potential unmet needs.

Objective: This study aimed to describe the experiences of patients with NSCLC as reported in discussions posted on lung cancer-specific social media with respect to their disease symptoms and associated impacts.

Methods: Publicly available posts (2010-2019) were extracted from selected lung cancer- or NSCLC-specific websites. Social media users (patients and caregivers posting on these websites) were stratified by metastatic- and adjuvant-eligible subgroups and treatment received using natural language processing (NLP) and machine learning methods. Automated identification of symptoms was conducted using NLP. Qualitative data analysis (QDA) was conducted on random samples of posts mentioning pain-related, fatigue-related, respiratory-related, or infection-related symptoms to capture the patient experience with these and associated impacts.

Results: Overall, 1724 users (50,390 posts) and 574 users (4531 posts) were included in the metastatic group and adjuvant group, respectively. Among users in the metastatic group, pain, discomfort, and fatigue were the most commonly mentioned symptoms (49.7% and 39.6%, respectively), and in the QDA (258 posts from 134 users), the most frequent impacts related to physical impairments, sleep, and eating habits. Among users in the adjuvant group, pain, discomfort, and respiratory symptoms were the most commonly mentioned (44.8% and 23.9%, respectively), and impacts identified in the QDA (154 posts from 92 users) were mostly related to physical functioning.

Conclusions: Findings from this exploratory observational analysis of social media among patients and caregivers informed the lived experience of NSCLC in the era of novel therapies, shedding light on most reported symptoms and their impacts. These findings can be used to inform future research on NSCLC treatment development and patient management.

Background: Patients with blood cancer experience serious physical and emotional symptoms throughout their cancer journey.

Objective: Building on previous work, we aimed to develop an app designed to help patients with multiple myeloma and chronic lymphocytic leukemia self-manage symptoms and test it for acceptability and preliminary efficacy.

Methods: We developed our Blood Cancer Coach app with input from clinicians and patients. Our 2-armed randomized controlled pilot trial recruited participants from Duke Health and nationally in partnerships with the Association of Oncology Social Work, Leukemia and Lymphoma Society, and other patient groups. Participants were randomized to the attention control (Springboard Beyond Cancer website) arm or the Blood Cancer Coach app intervention arm. The fully automated Blood Cancer Coach app included symptom and distress tracking with tailored feedback, medication reminders and adherence tracking, multiple myeloma and chronic lymphocytic leukemia education resources, and mindfulness activities. Patient-reported data were collected at baseline, 4 weeks, and 8 weeks for both arms through the Blood Cancer Coach app. Outcomes of interest were global health (Patient Reported Outcomes Measurement Information System Global Health), posttraumatic stress (Posttraumatic Stress Disorder Checklist for DSM-5), and cancer symptoms (Edmonton Symptom Assessment System Revised). Among participants in the intervention arm, satisfaction surveys and usage data were used to evaluate acceptability.

Results: Among 180 patients who downloaded the app, 49% (89) of them consented to participate and 40% (72) of them completed baseline surveys. Of those who completed baseline surveys, 53% (38) of them completed week 4 surveys (16 intervention and 22 control) and 39% (28) of them completed week 8 surveys (13 intervention and 15 control). Most participants found the app at least moderately effective at helping manage symptoms (87%), feeling more comfortable seeking help (87%), increasing awareness of resources (73%), and reported being satisfied with the app overall (73%). Participants completed an average of 248.5 app tasks over the 8-week study period. The most used functions within the app were medication log, distress tracking, guided meditations, and symptom tracking. There were no significant differences between the control and intervention arms at week 4 or 8 on any outcomes. We also saw no significant improvement over time within the intervention arm.

Conclusions: The results of our feasibility pilot were promising in which most participants found the app to be helpful in managing their symptoms, reported satisfaction with the app, and that it was helpful in several important areas. We did not, however, find significantly reduced symptoms or improved global mental and physical health over 2 months. Recruitment and retent