Development and Validation of a Prognostic Survival Model With Patient-Reported Outcomes for Patients With Cancer

Machine Learning-Based Prediction of 1-Year Survival Using Subjective and Objective Parameters in Patients With Cancer

PURPOSE

Palliative care is recommended for patients with cancer with a life expectancy of <12 months. Machine learning (ML) techniques can help in predicting survival outcomes among patients with cancer and may help distinguish who benefits the most from palliative care support. We aim to explore the importance of several objective and subjective self-reported variables. Subjective variables were collected through electronic psycho-oncologic and palliative care self-assessment screenings. We used these variables to predict 1-year mortality.

MATERIALS AND METHODS

Between April 1, 2020, and March 31, 2021, a total of 265 patients with advanced cancer completed a patient-reported outcome tool. We documented objective and subjective variables collected from electronic health records, self-reported subjective variables, and all clinical variables combined. We used logistic regression (LR), 20-fold cross-validation, decision trees, and random forests to predict 1-year mortality. We analyzed the receiver operating characteristic (ROC) curve-AUC, the precision-recall curve-AUC (PR-AUC)-and the feature importance of the ML models.

RESULTS

The performance of clinical nonpatient variables in predictions (LR reaches 0.81 [ROC-AUC] and 0.72 [F1 score]) are much more predictive than that of subjective patient-reported variables (LR reaches 0.55 [ROC-AUC] and 0.52 [F1 score]).

CONCLUSION

The results show that objective variables used in this study are much more predictive than subjective patient-reported variables, which measure subjective burden. These findings indicate that subjective burden cannot be reliably used to predict survival. Further research is needed to clarify the role of self-reported patient burden and mortality prediction using ML.

Health-related quality of life and DNA methylation-based aging biomarkers among survivors of childhood cancer

BACKGROUND

Childhood cancer survivors are at high risk for morbidity and mortality and poor patient-reported outcomes, typically health-related quality of life (HRQOL). However, associations between DNA methylation-based aging biomarkers and HRQOL have not been evaluated.

METHODS

DNA methylation was generated with Infinium EPIC BeadChip on blood-derived DNA (median for age at blood draw = 34.5 years, range = 18.5-66.6 years), and HRQOL was assessed with age at survey (mean = 32.3 years, range = 18.4-64.5 years) from 2206 survivors in the St Jude Lifetime Cohort. DNA methylation-based aging biomarkers, including epigenetic age using multiple clocks (eg, GrimAge) and others (eg, DNAmB2M: beta-2-microglobulin; DNAmADM: adrenomedullin), were derived from the DNAm Age Calculator (https://dnamage.genetics.ucla.edu). HRQOL was assessed using the Medical Outcomes Study 36-Item Short-Form Health Survey to capture 8 domains and physical and mental component summaries. General linear models evaluated associations between HRQOL and epigenetic age acceleration (EAA; eg, EAA_GrimAge) or other age-adjusted DNA methylation-based biomarkers (eg, ageadj_DNAmB2M) after adjusting for age at blood draw, sex, cancer treatments, and DNA methylation-based surrogate for smoking pack-years. All P values were 2-sided.

RESULTS

Worse HRQOL was associated with greater EAA_GrimAge (physical component summaries: β = -0.18 years, 95% confidence interval [CI] = -0.251 to -0.11 years; P = 1.85 × 10-5; and 4 individual HRQOL domains), followed by ageadj_DNAmB2M (physical component summaries: β = -0.08 years, 95% CI = -0.124 to -0.037 years; P = .003; and 3 individual HRQOL domains) and ageadj_DNAmADM (physical component summaries: β = -0.082 years, 95% CI = -0.125 to -0.039 years; P = .002; and 2 HRQOL domains). EAA_Hannum (Hannum clock) was not associated with any HRQOL.

CONCLUSIONS

Overall and domain-specific measures of HRQOL are associated with DNA methylation measures of biological aging. Future longitudinal studies should test biological aging as a potential mechanism underlying the association between poor HRQOL and increased risk of clinically assessed adverse health outcomes.

Patient-reported predictors of postdischarge mortality after cardiac hospitalization

BACKGROUND

Adults hospitalized for cardiovascular events are at high risk for postdischarge mortality. Screening of psychosocial risk is prioritized by the Joint Commission. We tested whether key patient-reported psychosocial and behavioral measures could predict posthospitalization mortality in a cohort of adults hospitalized for a cardiovascular event.

METHODS

We conducted a prospective cohort study to test the prognostic utility of validated patient-reported measures, including health literacy, social support, health behaviors and disease management, and socioeconomic status. Cox survival analyses of mortality were conducted over a median of 3.5 years.

RESULTS

Among 2977 adults hospitalized for either acute coronary syndrome or acute decompensated heart failure, the mean age was 53 years, and 60% were male. After adjusting for demographic, clinical, and other psychosocial factors, mortality risk was greatest among patients who reported being unemployed (hazard ratio [HR]: 1.99, 95% confidence interval [CI]): 1.30-3.06), retired (HR: 2.14, 95% CI: 1.60-2.87), or unable to work due to disability (HR: 2.36, 95% CI: 1.73-3.21), as compared to those who were employed. Patient-reported perceived health competence (PHCS-2) and exercise frequency were also associated with mortality risk after adjusting for all other variables (HR: 0.86, 95% CI: 0.73-1.00 per four-point increase in PHCS-2; HR: 0.86, 95% CI: 0.77-0.96 per 3-day increase in exercise frequency, respectively).

CONCLUSIONS

Patient-reported measures of employment status, perceived health competence, and exercise frequency independently predict mortality after a cardiac hospitalization. Incorporating these brief, valid measures into hospital-based screening may help with prognostication and targeting patients for resources during post-discharge transitions of care.

Development and validation of risk of CPS decline (RCD): a new prediction tool for worsening cognitive performance among home care clients in Canada

BACKGROUND

To develop and validate a prediction tool, or nomogram, for the risk of a decline in cognitive performance based on the interRAI Cognitive Performance Scale (CPS).

METHODS

Retrospective, population-based, cohort study using Canadian Resident Assessment Instrument for Home Care (RAI-HC) data, collected between 2010 and 2018. Eligible home care clients, aged 18+, with at least two assessments were selected randomly for model derivation (75%) and validation (25%). All clients had a CPS score of zero (intact) or one (borderline intact) on intake into the home care program, out of a possible score of six. All individuals had to remain as home care recipients for the six months observation window in order to be included in the analysis. The primary outcome was any degree of worsening (i.e., increase) on the CPS score within six months. Using the derivation cohort, we developed a multivariable logistic regression model to predict the risk of a deterioration in the CPS score. Model performance was assessed on the validation cohort using discrimination and calibration plots.

RESULTS

We identified 39,292 eligible home care clients, with a median age of 79.0 years, 62.3% were female, 38.8% were married and 38.6% lived alone. On average, 30.3% experienced a worsening on the CPS score within the six-month window (i.e., a change from 0 or 1 to 2, 3, 4, 5, or 6). The final model had good discrimination (c-statistic of 0.65), with excellent calibration.

CONCLUSIONS

The model accurately predicted the risk of deterioration on the CPS score over six months among home care clients. This type of predictive model may provide useful information to support decisions for home care clinicians who use interRAI data internationally.

Patient-reported predictors of post-discharge mortality after cardiac hospitalization

Background

Adults hospitalized for cardiovascular events are at high risk for post-discharge mortality. Hospital-based screening of health-related psychosocial risk factors is now prioritized by the Joint Commission and the National Quality Forum to achieve equitable, high-quality care. We tested our hypothesis that key patient-reported psychosocial and behavioral measures could predict post-hospitalization mortality in a cohort of adults hospitalized for a cardiovascular event.

Methods

This was a prospective cohort of adults hospitalized at Vanderbilt University Medical Center. Validated patient-reported measures of health literacy, social support, disease self-management, and socioeconomic status were used as predictors of interest. Cox survival analyses of mortality were conducted over a median 3.5-year follow-up (range: 1.25 - 5.5 years).

Results

Among 2,977 adults, 1,874 (63%) were hospitalized for acute coronary syndrome and 1,103 (37%) were hospitalized for acute decompensated heart failure; 60% were male; and the mean age was 53 years. After adjusting for demographic, clinical, and other psychosocial factors, mortality risk was greatest among patients who reported being unable to work due to disability (Hazard Ratio (HR) 2.36, 95% Confidence Interval (CI): 1.73-3.21), who were retired (HR 2.14, 95% CI 1.60-2.87), and who reported unemployment (HR 1.99, 95% CI 1.30-3.06) as compared to those who were employed. Patient-reported measures of disease self-management, perceived health competence and exercise frequency, were also associated with mortality risk after full covariate adjustment (HR 0.86, 95% CI 0.73-1.00 per four-point increase), (HR 0.86, 95% CI 0.77-0.96 per three-day change), respectively.

Conclusions

Patient-reported measures of employment status independently predict post-discharge mortality after a cardiac hospitalization. Measure of disease self-management also have prognostic modest utility. Hospital-based screening of psychosocial risk is increasingly prioritized in legislative policy. Incorporating brief, valid measures of employment status and disease self-management factors may help target patients for psychosocial, financial, and rehabilitative resources during post-discharge transitions of care.

Behavioural economic interventions to embed palliative care in community oncology (BE-EPIC): study protocol for the BE-EPIC randomised controlled trial

INTRODUCTION

Palliative care (PC) is a medical specialty focusing on providing relief from the symptoms and stress of serious illnesses such as cancer. Early outpatient specialty PC concurrent with cancer-directed treatment improves quality of life and symptom burden, decreases aggressive end-of-life care and is an evidence-based practice endorsed by national guidelines. However, nearly half of patients with advanced cancer do not receive specialty PC prior to dying. The objective of this study is to test the impact of an oncologist-directed default PC referral orders on rates of PC utilisation and patient quality of life.

METHODS AND ANALYSIS

This single-centre two-arm pragmatic randomised trial randomises four clinician-led pods, caring for approximately 250 patients who meet guideline-based criteria for PC referral, in a 1:1 fashion into a control or intervention arm. Intervention oncologists receive a nudge consisting of an electronic health record message indicating a patient has a default pended order for PC. Intervention oncologists are given an opportunity to opt out of referral to PC. Oncologists in pods randomised to the control arm will receive no intervention beyond usual practice. The primary outcome is completed PC visits within 12 weeks. Secondary outcomes are change in quality of life and absolute quality of life scores between the two arms.

ETHICS AND DISSEMINATION

This study has been approved by the Institutional Review Board at the University of Pennsylvania. Study results will be disseminated in peer-reviewed journals and scientific conferences using methods that describe the results in ways that key stakeholders can best understand and implement.

TRIAL REGISTRATION NUMBER

NCT05365997.

myBeST-A Web-Based Survival Prognostic Tool for Women with Breast Cancer in Malaysia: Development Process and Preliminary Validation Study

Women with breast cancer are keen to know their predicted survival. We developed a new prognostic model for women with breast cancer in Malaysia. Using the model, this study aimed to design the user interface and develop the contents of a web-based prognostic tool for the care provider to convey survival estimates. We employed an iterative website development process which includes: (1) an initial development stage informed by reviewing existing tools and deliberation among breast surgeons and epidemiologists, (2) content validation and feedback by medical specialists, and (3) face validation and end-user feedback among medical officers. Several iterative prototypes were produced and improved based on the feedback. The experts (n = 8) highly agreed on the website content and predictors for survival with content validity indices ≥ 0.88. Users (n = 20) scored face validity indices of more than 0.90. They expressed favourable responses. The tool, named Malaysian Breast cancer Survival prognostic Tool (myBeST), is accessible online. The tool estimates an individualised five-year survival prediction probability. Accompanying contents were included to explain the tool's aim, target user, and development process. The tool could act as an additional tool to provide evidence-based and personalised breast cancer outcomes.

Comparing machine learning approaches to incorporate time-varying covariates in predicting cancer survival time

The Cox proportional hazards model is commonly used in evaluating risk factors in cancer survival data. The model assumes an additive, linear relationship between the risk factors and the log hazard. However, this assumption may be too simplistic. Further, failure to take time-varying covariates into account, if present, may lower prediction accuracy. In this retrospective, population-based, prognostic study of data from patients diagnosed with cancer from 2008 to 2015 in Ontario, Canada, we applied machine learning-based time-to-event prediction methods and compared their predictive performance in two sets of analyses: (1) yearly-cohort-based time-invariant and (2) fully time-varying covariates analysis. Machine learning-based methods-gradient boosting model (gbm), random survival forest (rsf), elastic net (enet), lasso and ridge-were compared to the traditional Cox proportional hazards (coxph) model and the prior study which used the yearly-cohort-based time-invariant analysis. Using Harrell's C index as our primary measure, we found that using both machine learning techniques and incorporating time-dependent covariates can improve predictive performance. Gradient boosting machine showed the best performance on test data in both time-invariant and time-varying covariates analysis.

Identification and Prognostication of End-of-Life State Using a Japanese Guideline-Based Diagnostic Method: A Diagnostic Accuracy Study

Purpose

Prognostic uncertainty can be a barrier to providing palliative care. Accurate prognostic estimation for patients at the end of life is challenging. This study aimed to evaluate the accuracy of end-of-life diagnosis using our unique diagnostic method.

Patients and Methods

A retrospective longitudinal observational study was conducted through collaboration among three medical facilities in a rural super-aged community in Japan. In 2007, we established a unique end-of-life diagnostic process comprising (1) physicians' judgement, (2) disclosure to patients, and (3) discussion at an end-of-life case conference (EOL-CC), based on Japanese end-of-life-related guidelines. Research subjects were consecutive patients discussed in EOL-CC between January 1, 2010, and September 30, 2017. The primary outcome was mortality within 6 months after the initial EOL-CC decision. Sensitivity, specificity, and diagnostic odds ratio were calculated using EOL-CC diagnosis (end-of-life or non-end-of-life) as an index test and overall survival (<6 months or ≥6 months) as a reference standard.

Results

In total, 315 patients were eligible for survival analysis (median age 89, range 54-107). The study population was limited to patients with severe conditions such as advanced cancer, organ failures, advanced dementia with severe deterioration in functioning. EOL-diagnosis by our methods was associated with much lower survival rate at 6 months after EOL-CC than non-EOL-diagnosis (6.9% vs 43.5%; P < 0.001). Of the patients, 297 were eligible for diagnostic accuracy analysis (median age 89, range 54-107). The EOL-diagnosis showed high sensitivity (0.95; 95% confidence interval [CI] 0.92-0.97) but low specificity (0.35; 95% CI 0.20-0.53) against the outcomes. It also showed a high diagnostic odds ratio (10.32; 95% CI 4.08-26.13).

Conclusion

The diagnostic process using the Japanese end-of-life guidelines had tolerable accuracy in identification and prognostication of end of life.

Development of Predictive Models for Survival among Women with Breast Cancer in Malaysia

Prediction of survival probabilities based on models developed by other countries has shown inconsistent findings among Malaysian patients. This study aimed to develop predictive models for survival among women with breast cancer in Malaysia. A retrospective cohort study was conducted involving patients who were diagnosed between 2012 and 2016 in seven breast cancer centres, where their survival status was followed until 31 December 2021. A total of 13 predictors were selected to model five-year survival probabilities by applying Cox proportional hazards (PH), artificial neural networks (ANN), and decision tree (DT) classification analysis. The random-split dataset strategy was used to develop and measure the models' performance. Among 1006 patients, the majority were Malay, with ductal carcinoma, hormone-sensitive, HER2-negative, at T2-, N1-stage, without metastasis, received surgery and chemotherapy. The estimated five-year survival rate was 60.5% (95% CI: 57.6, 63.6). For Cox PH, the c-index was 0.82 for model derivation and 0.81 for validation. The model was well-calibrated. The Cox PH model outperformed the DT and ANN models in most performance indices, with the Cox PH model having the highest accuracy of 0.841. The accuracies of the DT and ANN models were 0.811 and 0.821, respectively. The Cox PH model is more useful for survival prediction in this study's setting.

Trends in quality indicators of end-of-life care for women with gynecologic malignancies in Ontario, Canada

OBJECTIVE

A large body of research has validated several quality indicators of end-of-life (EOL) cancer care, but few have examined these in gynecologic cancer at a population-level. We examined patterns of EOL care quality in patients with gynecologic cancers across 13 years in Ontario, Canada.

METHODS

We conducted a population-based, retrospective cohort study of gynecologic cancer decedents in Ontario from 2006 to 2018 using linked administrative health care databases. Proportions of quality indices were calculated, including: emergency department (ED) use, hospital or intensive care unit (ICU) admission, chemotherapy ≤14 days of death, cancer-related surgery, tube or intravenous feeds, palliative home visits, and hospital death. We used multivariable logistic regression to examine factors associated with receipt of aggressive and supportive care.

RESULTS

There were 16,237 included decedents over the study period; hospital death rates decreased from 47% to 37%, supportive care use rose from 65% to 74%, and aggressive care remained stable (16%). Within 30 days of death, 50% were hospitalized, 5% admitted to ICU, and 67% accessed palliative homecare. Within 14 days of death, 31% visited the ED and 4% received chemotherapy. Patients with vulvovaginal cancers received the lowest rates of aggressive and supportive care. Using multivariable analyses, factors associated with increased aggressive EOL care use included younger age, shorter disease duration, lower income quintiles, and rural residence.

CONCLUSIONS

Over time, less women dying with gynecologic cancers in Ontario experienced death in hospital, and more accessed supportive care. However, the majority were still hospitalized and a significant proportion received aggressive care in the final 30 days of life.

Patient-reported outcomes predict survival and adverse events following anticancer treatment initiation in advanced HER2-positive breast cancer

BACKGROUND

The prognostic value of patient-reported outcomes (PROs) has been minimally explored in advanced breast cancer (BC), and their comparative prognostic performance against Eastern Cooperative Oncology Group performance status (ECOG PS) is largely unknown.

PATIENTS AND METHODS

This study pooled individual participant data from clinical trials CLEOPATRA, EMILIA, and MARIANNE. Pre-treatment PRO associations with overall survival (OS), progression-free survival (PFS), and grade ≥3 adverse events were evaluated via Cox proportional hazards regression. Prognostic performance was assessed with the C-statistic (c). PRO values were collected via the Functional Assessment of Cancer Therapy-Breast (FACT-B) questionnaire. All analyses were stratified by study and treatment arms. Analyses adjusted for known prognostic variables were conducted. Exploratory analysis of the prognostic performance of PROs compared to ECOG PS was undertaken.

RESULTS

The study included data from 2894 patients initiated on contemporary therapies including pertuzumab (n = 765), trastuzumab (n = 1173), trastuzumab emtansine (n = 1225), taxanes (n = 1173), lapatinib (n = 496), and capecitabine (n = 496). On univariable and adjusted analysis, patient-reported physical well-being, functional well-being, and BC subscale were all identified to be associated with OS, PFS, and grade ≥3 adverse events (P < 0.05). Patient-reported physical well-being was the most prognostic PRO for all assessed outcomes. The OS prognostic performance of physical well-being (c = 0.58) was superior to ECOG PS (c = 0.56) (P < 0.05), with multivariable analysis indicating that both provide independent information (P < 0.0001).

CONCLUSIONS

PROs were identified as independent prognostic factors for OS, PFS, and grade ≥3 adverse events in patients with human epidermal growth factor receptor 2 (HER2)-positive advanced BC initiating contemporary treatment options. Further, patient-reported physical well-being was more prognostic of OS than ECOG PS and contained independent information. PROs have value as prognostic and stratification factors for clinical use and research trials of anticancer treatment in HER2-positive ABC.

Incorporating conditional survival into prognostication for gunshot wounds to the head

OBJECTIVE

Several scores estimate the prognosis for gunshot wounds to the head (GSWH) at the point of hospital admission. However, prognosis may change over the course of the hospital stay. This study measures the accuracy of the Baylor score among patients who have already survived the acute phase of hospitalization and generates conditional outcome curves for the duration of hospital stay for patients with GSWH.

METHODS

Patients in whom GSWH with dural penetration occurred between January 2009 and June 2019 were identified from a trauma registry at a level I trauma center in the southeastern US. The Baylor score was calculated using component variables. Conditional overall survival and good functional outcome (Glasgow Outcome Scale score of 4 or 5) curves were generated. The accuracy of the Baylor score in predicting mortality and functional outcome among acute-phase survivors (survival > 48 hours) was assessed using receiver operating characteristic curves and the area under the curve (AUC).

RESULTS

A total of 297 patients were included (mean age 38.0 [SD 15.7] years, 73.4% White, 85.2% male), and 129 patients survived the initial 48 hours of admission. These acute-phase survivors had a decreased mortality rate of 32.6% (n = 42) compared to 68.4% (n = 203) for all patients, and an increased rate of good functional outcome (48.1%; n = 62) compared to the rate for all patients (23.2%; n = 69). Among acute-phase survivors, the Baylor score accurately predicted mortality (AUC = 0.807) and functional outcome (AUC = 0.837). However, the Baylor score generally overestimated true mortality rates and underestimated good functional outcome. Additionally, hospital day 18 represented an inflection point of decreasing probability of good functional outcome.

CONCLUSIONS

During admission for GSWH, surviving beyond the acute phase of 48 hours doubles the rates of survival and good functional outcome. The Baylor score maintains reasonable accuracy in predicting these outcomes for acute-phase survivors, but generally overestimates mortality and underestimates good functional outcome. Future prognostic models should incorporate conditional survival to improve the accuracy of prognostication after the acute phase.

Promising Advances in LINC01116 Related to Cancer

Long non-coding RNAs (lncRNAs) are RNAs with a length of no less than 200 nucleotides that are not translated into proteins. Accumulating evidence indicates that lncRNAs are pivotal regulators of biological processes in several diseases, particularly in several malignant tumors. Long intergenic non-protein coding RNA 1116 (LINC01116) is a lncRNA, whose aberrant expression is correlated with a variety of cancers, including lung cancer, gastric cancer, colorectal cancer, glioma, and osteosarcoma. LINC01116 plays a crucial role in facilitating cell proliferation, invasion, migration, and apoptosis. In addition, numerous studies have recently suggested that LINC01116 has emerged as a novel biomarker for prognosis and therapy in malignant tumors. Consequently, we summarize the clinical significance of LINC01116 associated with biological processes in various tumors and provide a hopeful orientation to guide clinical treatment of various cancers in future studies.

Development and validation of a prediction model of poor performance status and severe symptoms over time in cancer patients (PROVIEW+)

BACKGROUND

Predictive cancer tools focus on survival; none predict severe symptoms.

AIM

To develop and validate a model that predicts the risk for having low performance status and severe symptoms in cancer patients.

DESIGN

Retrospective, population-based, predictive study.

SETTING/PARTICIPANTS

We linked administrative data from cancer patients from 2008 to 2015 in Ontario, Canada. Patients were randomly selected for model derivation (60%) and validation (40%). Using the derivation cohort, we developed a multivariable logistic regression model to predict the risk of an outcome at 6 months following diagnosis and recalculated after each of four annual survivor marks. Model performance was assessed using discrimination and calibration plots. Outcomes included low performance status (i.e. 10-30 on Palliative Performance Scale), severe pain, dyspnea, well-being, and depression (i.e. 7-10 on Edmonton Symptom Assessment System).

RESULTS

We identified 255,494 cancer patients (57% female; median age of 64; common cancers were breast (24%); and lung (13%)). At diagnosis, the predicted risk of having low performance status, severe pain, well-being, dyspnea, and depression in 6-months is 1%, 3%, 6%, 13%, and 4%, respectively for the reference case (i.e. male, lung cancer, stage I, no symptoms); the corresponding discrimination for each outcome model had high AUCs of 0.807, 0.713, 0.709, 0.790, and 0.723, respectively. Generally these covariates increased the outcome risk by >10% across all models: lung disease, dementia, diabetes; radiation treatment; hospital admission; pain; depression; transitional performance status; issues with appetite; or homecare.

CONCLUSIONS

The model accurately predicted changing cancer risk for low performance status and severe symptoms over time.

Machine Learning Applied to Electronic Health Records: Identification of Chemotherapy Patients at High Risk for Preventable Emergency Department Visits and Hospital Admissions

PURPOSE

Acute care use (ACU) is a major driver of oncologic costs and is penalized by a Centers for Medicare & Medicaid Services quality measure, OP-35. Targeted interventions reduce preventable ACU; however, identifying which patients might benefit remains challenging. Prior predictive models have made use of a limited subset of the data in the electronic health record (EHR). We aimed to predict risk of preventable ACU after starting chemotherapy using machine learning (ML) algorithms trained on comprehensive EHR data.

METHODS

Chemotherapy patients treated at an academic institution and affiliated community care sites between January 2013 and July 2019 who met inclusion criteria for OP-35 were identified. Preventable ACU was defined using OP-35 criteria. Structured EHR data generated before chemotherapy treatment were obtained. ML models were trained to predict risk for ACU after starting chemotherapy using 80% of the cohort. The remaining 20% were used to test model performance by the area under the receiver operator curve.

RESULTS

Eight thousand four hundred thirty-nine patients were included, of whom 35% had preventable ACU within 180 days of starting chemotherapy. Our primary model classified patients at risk for preventable ACU with an area under the receiver operator curve of 0.783 (95% CI, 0.761 to 0.806). Performance was better for identifying admissions than emergency department visits. Key variables included prior hospitalizations, cancer stage, race, laboratory values, and a diagnosis of depression. Analyses showed limited benefit from including patient-reported outcome data and indicated inequities in outcomes and risk modeling for Black and Medicaid patients.

CONCLUSION

Dense EHR data can identify patients at risk for ACU using ML with promising accuracy. These models have potential to improve cancer care outcomes, patient experience, and costs by allowing for targeted, preventative interventions.

Patient-Reported Outcomes Predict Progression-Free Survival of Patients with Advanced Breast Cancer Treated with Abemaciclib

BACKGROUND

Abemaciclib is a CDK4/6 inhibitor used to treat hormone receptor-positive, human epidermal growth factor receptor 2-negative advanced breast cancer. The prognostic value of patient-reported outcomes (PROs) has been minimally explored for treatment outcomes with CDK4/6 inhibitors. The performance of PROs compared with Eastern Cooperative Oncology Group performance status (ECOG-PS) is unknown.

MATERIALS AND METHODS

This study pooled data from single-arm trial, MONARCH 1, and randomized trials, MONARCH 2 and 3. In total, 900 patients initiated abemaciclib and 384 comparator therapy. Pretreatment PRO association with progression-free survival (PFS) was modeled using Cox proportional hazards regression. Prediction performance was assessed via the C-statistic (c). PROs were recorded via the European Organisation for Research and Treatment of Cancer QLQ-C30.

RESULTS

Patient-reported physical function, pain, role function, fatigue, and appetite loss were associated with PFS on univariable and adjusted analysis (p < .05). Physical function (c = 0.55) was most predictive, superior to ECOG-PS (c = 0.54), with multivariable analysis indicating both provide independent information (p < .02). In the pooled randomized arms of MONARCH 2 and 3, the PFS treatment benefit (hazard ratio [95% confidence interval]) of abemaciclib (vs. comparators) was 0.75 (0.57-1.0) for low physical function, compared with 0.48 (0.40-0.59) for intermediate/high (p[interaction] = .01).

CONCLUSION

PROs were identified as prognostic factors for PFS in patients initiating abemaciclib, with patient-reported physical function containing independent predictive information beyond ECOG-PS. Low physical function was associated with a decrease in the magnitude of PFS benefit from abemaciclib. PROs should be explored as prognostic, predictive, and stratification factors for clinical use and research trials of CDK4/6 inhibitors.

IMPLICATIONS FOR PRACTICE

For the first time, pretreatment patient-reported outcomes have been shown to be independent prognostic markers for progression-free survival (PFS) in patients diagnosed with hormone receptor-positive, human epidermal growth factor receptor 2-negative (HR+/HER2-) advanced breast cancer treated with abemaciclib. Importantly, patients with low physical function had a smaller PFS benefit from abemaciclib (vs. comparator) than patients with intermediate/high physical function. The present study demonstrates patient-reported outcomes as a simple, effective, inexpensive, and independent prognostic marker for patients with HR+/HER2- advanced breast cancer treated with abemaciclib.