Search

PURPOSE The optimal characteristics among patients with breast cancer to recommend neoadjuvant chemotherapy is an active area of clinical research. We developed and compared several approaches to developing prediction models for pathologic complete response (pCR) among patients with breast cancer in Alberta. METHODS The study included all patients with breast cancer who received neoadjuvant chemotherapy in Alberta between 2012 and 2014 identified from the Alberta Cancer Registry. Patient, tumor, and treatment data were obtained through primary chart review. pCR was defined as no residual invasive tumor at surgical excision in breast or axilla. Two types of prediction models for pCR were built: (1) expert model: variables selected on the basis of oncologists' opinions and (2) data-driven model: variables selected by trained machine. These model types were fit using logistic regression (LR), random forests (RF), and gradient-boosted trees (GBT). We compared the models using area under the receiver operating characteristic curve and integrated calibration index, and internally validated using bootstrap resampling. RESULTS A total of 363 cases were included in the analyses, of which 86 experienced pCR. The RF and GBT fits yielded higher optimism-corrected area under the receiver operating characteristic curves compared with LR for the expert (RF: 0.70; GBT: 0.69; LR: 0.65) and data-driven models (RF: 0.71; GBT: 0.68; LR: 0.64). The LR fit yielded the lowest integrated calibration indices for the expert (LR: 0.037; GBT: 0.05; RF: 0.10) and data-driven models (LR: 0.026; GBT: 0.06; RF: 0.099). CONCLUSION Our models demonstrated predictive ability for pCR using routinely collected clinical and demographic variables. We show that machine learning fit methods can be used to optimize models for pCR prediction. We also show that additional variables beyond clinical expertise do not considerably improve predictive ability and may not be of value on the basis of the burden of data collection.

76 Background: In recent years, treatment intensification beyond androgen deprivation therapy (ADT) with several novel therapies have shown survival benefit in patients with metastatic castration-sensitive prostate cancer (mCSPC). Given the rapidly evolving landscape in mCSPC treatment, there is a need to better understand how treatment strategies fit in real-world clinical practice and are combined/sequenced with other available therapies. Methods: Using electronic medical records and administrative data, a population-based retrospective cohort study was conducted. Patients aged ≥18 years of age who were newly diagnosed with de novo mCSPC and initiated ADT post-diagnosis between 2010 to 2020 in Alberta, Canada, were included. Treatment intensification was defined as the receipt of apalutamide, abiraterone acetate, enzalutamide, or chemotherapy (e.g. docetaxel) within 180 days of ADT initiation. Results: A total of 2,515 de novo mCSPC were identified during study period with 2,098 (83%) patients initiating ADT post-diagnosis. Of those, 525 (25%) received intensification beyond ADT. The percentage of patients who were intensified was 3% in 2010-2013 and gradually increased to 67% in 2020. Between 2014-2017, docetaxel was the most common therapy for intensification, but its use decreased considerably in 2018-2020 with abiraterone acetate, apalutamide and enzalutamide becoming increasingly available in the mCSPC setting. Upon progression, 46% and 22% in the intensified group versus 38% and 13% in the ADT-alone group initiated one and two-lines of subsequent therapies respectively. Abiraterone acetate and enzalutamide were the most common subsequent therapy for both the intensified (32% and 31% respectively) and the ADT-alone (56% and 38% respectively) groups. Docetaxel (24%) was used as subsequent therapy among mCSPC patients who were intensified with oral systemic agents. In multivariable logistic regression analyses of patients diagnosed in 2014-2020, significant predictors of intensification were younger age at diagnosis, lower Charlson comorbidity index, greater number of metastatic sites, shorter time to ADT initiation, referral to a specialists/cancer centres, surgery or radiation prior to ADT, and more recent year of diagnosis (all p

56 Background: The use of biologics in combination with chemotherapy early in the course of treatment in mCRC improves outcomes, but whether this approach is uniformly adopted in clinical practice is unclear. This study aims to describe biomarker testing, treatment patterns, and survival of RAS WT, left-sided mCRC patients in a real-world population. Methods: We performed deterministic linkages with population-based data sources, including the cancer registry, pharmacy data, electronic medical records, and vital statistics to identify all patients who were diagnosed with RAS WT, left-sided mCRC, and treated with first-line (1L) combination systemic therapy from 2014 to 2019 in Alberta, Canada. Patients were followed until December 31, 2020. Results: Of 2,721 patients with left-sided mCRC, 1,375 were referred and received systemic therapy, 977 underwent RAS testing prior to or within 30 days of initiating 1L treatment, and 420 were found to be RAS WT. Among them, 320 were treated with 1L combination systemic therapy: FOLFOX/CAPOX/FOLFIRI (n=204), panitumumab (pani) with doublet chemotherapy (n=64), or bevacizumab (bev) with doublet chemotherapy (n=52). The interval from diagnosis to RAS test results was 38 (IQR 24-61) days. Only 207 (65%) and 125 (39%) of the 320 1L-treated patients reached 2L and 3L therapy, respectively. Median overall survival based on the different 1L regimens are summarized. Conclusions: Many real-world patients with mCRC were not exposed to biologics as part of their front-line treatment. There is a need to enhance RAS testing in 1L as 29% of patients were not tested, despite recommendations from guidelines. Because attrition across lines of therapy is high in mCRC, the opportunity to receive and benefit from biologics in later lines of therapy may be importantly reduced. [Table: see text]

228 Background: Medical cannabis (MC) and prescription opioid medication (POM) use is common among cancer patients. There is conflicting evidence on the association of cannabis with POM as to whether cannabis can help decrease/ cease opioid use. In this population-based study, we examine the association between MC authorization and cessation or reduction in POM use among patients with early stage cancer. Methods: This is a retrospective, population-based study of patients with early stage (stage I-III) cancer diagnosed between January 1, 2014 and December 31, 2018 in the province of Alberta, Canada. Cases were identified from the Alberta Cancer Registry (ACR) and linked to the provincial pharmacy information network (PIN) and the database from the College of Physician and Surgeons of Alberta (CPSA). Patient and treatment characteristic were used to identify a comparable non-MC group with prior POM use via probabilistic modelling. Descriptive statistics were used to describe differences between patients with and without a MC authorization. Modified Poisson regression was used to compare the likelihood of opioid cessation and reduction among groups. Results: We identified 8,801 patients of whom 326 (3.7%) had a MC authorization. Patients with a MC authorization were younger, had higher stage disease, underwent radiation and/or systemic therapy and had a higher total oral morphine equivalent (OME) use at baseline (p < 0.01). Patients with a MC authorization were less likely to cease POM at 9-12 months post MC authorization (RR 0.63, 95% CI 0.57-0.70), and less likely to reduce their POM dose by 25% (RR 0.79, 95% CI 0.74-0.85) and 50%. (RR 0.73, 95% CI 0.67-0.79). Conclusions: Patients with early stage, non-metastatic cancer with a MC authorization have higher rates of baseline POM use and are less likely to cease or reduce their POM use up to 1 year after MC authorization. Further study is required to understand the harms of concomitant MC and POM use and the impact on survivorship care.

e20504 Background: The prognosis of eNSCLC remains uniformly poor. An understanding of current therapies and outcomes can provide insights into how novel therapies can be integrated into our management paradigm. Methods: We conducted a large, retrospective, population-based cohort study of de novo eNSCLC patients (stages IB, IIA, IIB, and IIIA) diagnosed in Alberta, Canada between 2010-2019 using electronic medical records and administrative claims data. The primary objectives were to describe treatment patterns and survival outcomes among eNSCLC patients. In addition, we examined the association between systemic therapy (ST) and overall survival (OS) using multivariable Cox proportional hazards models. Results: A total of 5,126 eNSCLC patients were included. The stage distributions were: 31.0% IB, 13.4% IIA, 17.7% IIB, and 37.9% IIIA. The mean (SD) age was 71.3 (10.3) years and 52.5% were female. 45.3% of patients were referred to a medical oncologist, ranging from 23.7% in stage IB and 58.3% of IIIA. Among stage IB and II patients, 59.2% and 58.1% received surgery, respectively, while 25.7% of stage IIIA patients underwent surgery. 23.6% of patients initiated ST, ranging from 3.5% in stage IB to 38.5% in IIIA. ST use increased over the study period by 9.3% and 19.5% in stage IIB and IIIA disease, respectively. Median follow-up for the cohort was 21.86 months; median OS was 28.18 months (95% CI: 26.56-29.69). Median OS for stage IB, IIA, IIB, and IIIA were 49.01 (95% CI: 45.00-54.15), 36.56 (95% CI: 32.94-42.25), 29.23 (95% CI: 25.32-33.11), and 16.50 (95% CI: 15.39-17.59). Findings from the Cox analyses are tabulated (see Table). For stage IIB and IIIA individuals who received surgery, adjuvant ST was also associated with a decreased likelihood of death [hazard ratios (HR) of 0.77 (95% CI: 0.56-1.07) and 0.69 (95% CI: 0.54-0.89), respectively]. Conclusions: In a Canadian real-world setting, stage IIB and IIIA patients who received adjuvant ST tended to have better survival than patients who did not. However, a considerable proportion of patients are not referred to a medical oncologist to be considered for ST. Improving referral pathways appears to be an essential step to ensure that emerging novel therapies are implemented effectively in the real world so that potential survival gains from new drugs can be realized.[Table: see text]

e21139 Background: While Epidermal Growth Factor Receptor (EGFR) Tyrosine Kinase Inhibitors have been shown to be effective in phase III randomized trials, the value of targeted therapies have been challenging to evaluate at the population level. We examined the impact of population-level EGFR testing and treatment on survival outcomes among metastatic Non-Small Cell Lung Cancer (NSCLC) patients. Methods: Real-world, population-level data were collected from all de novo metastatic non-squamous NSCLC patients in Alberta, Canada from 2004 to 2020. EGFR testing data were collected through Alberta Precision Laboratories using various text mining approaches. Differences in survival rates and overall survival (OS) pre (2004-2012) and post-initiation (post) (2013-2019) testing periods were evaluated using interrupted time series analyses. The impact of testing and subsequent treatment weas evaluated using multivariable Cox Proportional Hazards models. Results: In total, 4,578 metastatic NSCLC patients with a confirmed non-squamous cell carcinoma histology were diagnosed pre- EGFR testing and 4,457 patients were diagnosed post- EGFR testing (2013-2019). Among patients diagnosed in the pre- EGFR testing period, the 6-month, 1-year, and 2-year survival probabilities were 0.39 (95% CI: 0.38-0.41), 0.22 (95% CI: 0.21-0.23), and 0.09 (95% CI: 0.08-0.10), while the survival probabilities for patients diagnosed in the post- EGFR testing period were 0.45 (95% CI: 0.43-0.46), 0.29 (95% CI: 0.27-0.30), and 0.16 (95% CI: 0.15-0.1 7), respectively. After adjusting for baseline patient and clinical characteristics, OS in the post- EGFR period was significantly improved compared to the pre- EGFR period (HR: 0.81; 95% CI: 0.78-0.85). In the post-EGFR period, among patients who were treated with systemic therapy, those tested for an EGFR mutation had significantly greater survival than patients who were not tested HR of 0.81 (95% CI: 0.70-0.95). Conclusions: These results show the considerable impact of population-based molecular testing and subsequent targeted therapies on survival among advanced NSCLC patients. The estimates here can be used in future studies to evaluate the population-level cost-effectiveness of testing and treatment.

e18776 Background: Adults aged ≥70 years represent approximately half of all patients diagnosed with colon cancer (CC), but undertreatment in this population persists. Recent guidelines have aimed to reduce age-related biases in the treatment of CC and emphasized the importance of personalizing management with comprehensive geriatric assessments (CGAs). Therefore, we hypothesized that age-related disparities in the curative-intent treatment of CC would improve over time. Methods: This was a retrospective, population-based cohort study of adults diagnosed with CC between 2010 and 2018 in Alberta, Canada. The study data included patient demographics and clinical characteristics collected through the Alberta Cancer Registry and electronic medical records. Patients were stratified by age: < 70 and ≥70 years. Cox proportional hazard models (CPHM) were generated to evaluate the associations and interaction between age groups and treatment status on disease-specific survival (DSS), after adjusting for important covariates. Multivariable logistic regression was used to identify time trends and predictors of treatment receipt. Results: A total of 10,838 patients were included, of whom 5,176 (48%) were aged ≥70 years and 2,468 (23%) had stage IV CC at initial diagnosis. Older age was associated with greater comorbidity and less advanced disease ( p < 0.001, standardized mean difference > 0.1 for both). The vast majority (87%) of patients in the overall cohort received surgery while 34% received systemic therapy. In multivariable CPHM, older age was associated with lower DSS (HR 1.42, 95%CI 1.31-1.54, p < 0.001) while surgery and systemic therapy were each associated with higher DSS (HR 0.30, 95%CI 0.27-0.33, p < 0.001; HR 0.40, 95%CI 0.37-0.43, p < 0.001; respectively). However, the interaction between age and treatment status was not statistically significant ( p = 0.78 for surgery; p = 0.17 for systemic therapy). Compared to the younger age group, the odds of receiving surgery and systemic therapy were 3 and 5 times lower, respectively, among older patients (OR 0.27, 95%CI 0.18-0.40, p < 0.001; OR 0.18, 95%CI 0.16-0.20, p < 0.001; respectively). In addition to younger age, predictors of surgery receipt included less comorbidity and stage II/III vs I disease, whereas predictors of systemic therapy receipt included male sex, southern residence, higher neighbourhood income, less comorbidity, and stage III vs IV disease ( p < 0.05 for all). There were no statistically significant correlations between year of diagnosis and treatment receipt ( ptrend = 0.07 for surgery; ptrend = 0.26 for systemic therapy). Conclusions: Surgery and systemic therapy continue to improve CC outcomes regardless of age. However, rates of curative-intent treatment for CC were consistently lower in patients aged ≥70 years, with minimal changes over time. Better integration of CGAs into routine care may be needed to reduce persistent age-related treatment disparities.

e15539 Background: There has been a consistent increase in the incidence of early-onset colorectal cancer (eoCRC), under the age of 50, in Canada since the late 1990s. Questions remain surrounding how these trends vary by topography, histology, and stage, and related trends with CRC-specific mortality. Methods: CRC incidence data were obtained from the Canadian Cancer Registry and CRC-specific mortality data from the Canadian Vital Statistics – Death Database for the years 2000 to 2017. Age-specific average annual percent changes (AAPC) in the incidence (by topography and histology) and mortality rates of CRC were estimated using the National Cancer Institute’s Joinpoint Regression Program. To determine age-specific differences (5-year age groups) in CRC diagnosis at late stage (III and IV) for years 2011 to 2017 combined, a logistic regression model adjusting for sex with the 50-54 age group as the referent was conducted. Results: AAPCs and 95% confidence intervals in the rates of incidence (topography and histology) and mortality of eoCRC from 2000 to 2017 in Canada are presented in Table. Different trends in topography were observed across sexes with the largest increases in the distal colon (splenic flexure, descending, and sigmoid) and rectum among males and rectum only among females. Significant increases were observed for non-mucinous adenocarcinomas, while significant decreases were observed for mucinous adenocarcinomas among the 40-49 age group. Compared to the 50-54 age group, only the 45-49 group had a significantly higher odds of developing late-stage colon cancer, while men and adults 25-49 had a higher odds of developing late stage rectal cancer. Despite increases in the incidence of eoCRC there has only been a significant increase in mortality for men aged 20-39. Trends in mortality vary by site, with significant decreases observed for colon cancer-specific mortality among the 40-49 age group and increases in rectal cancer-specific mortality for adults aged 20-49. Conclusions: These results indicate that the largest increases in incidence and mortality for eoCRC have occurred in the rectum and trends have varied by sex. Further research on the etiology and treatment outcomes of eoCRC patients are warranted for this patient population.[Table: see text]

e18713 Background: When randomized trials are not available, observational real-world data can be used to emulate a (hypothetical) target trial. The procedure starts with the specification of the protocol of the target trial, whose components are then explicitly emulated using observational data. This approach prevents biases that are common when using more conventional methods for real-world data. Advanced pancreatic cancer represents an opportunity for trial emulation since two main frontline therapies, FOLFIRINOX and GN, have never been directly compared in a randomized fashion. The choice between the two regimens is largely based on physician discretion and patient preference rather than direct comparison of effectiveness. Methods: We emulated a target trial using linked data from the provincial cancer registry, electronic health records and various administrative databases from Alberta, Canada. Eligible individuals had locally advanced or metastatic pancreatic cancer diagnosed between Jan. 2015- Dec. 2018, no prior treatment, and adequate hematologic and serum creatinine values. They were followed from diagnosis until March 2020, death, or date of last known contact with the healthcare system. We estimated the effect of initiating FOLFIRINOX vs. GN within 8 weeks of diagnosis on overall survival. Cloning, artificial censoring, and inverse probability weighting were used to address unknown treatment assignment at baseline, non-adherence, and confounding. Adjusted Kaplan-Meier survival curves and hazard ratios were estimated. Results: Of 298 eligible individuals, 70 adhered to the FOLFIRINOX strategy and 147 to the GN strategy. The mean age was 65 years, 173 (58%) were male, and 247 (83%) had metastatic disease. The adjusted median survival, 1-year survival, and 2-year survival for FOLFIRNOX was 8.2 months (95% CI: 5.3 to 9.4), 36.9% (22.2 to 55.8), and 14.1% (4.8 to 32.2), respectively; and for GN was 4.8 months (3.3 to 5.3), 22.2% (13.6 to 35.9), and 4.7% (1.7 to 13.0), respectively. The adjusted difference in median survival was 3.4 months (0.6 to 11.1) and the adjusted hazard ratio was 0.79 (0.56 to 1.05). Conclusions: Target trial emulations can help to inform medical decision making in situations where head-to-head randomized trial data are unavailable or unfeasible. Findings from this real-world trial emulation suggest improved overall survival with FOLFIRINOX over GN.

e14049 Background: Patient-level heterogeneity in response to treatment remains a major challenge in cancer immunotherapy. Long-term individual-level modelling of survival, tumour response and safety outcomes jointly can help improve efficacy and durable clinical benefits. Machine-learned Bayesian networks provide a solution to the “black box” problem of other machine-learning approaches. Objectives: To develop a dynamic Bayesian network model for multivariate risk prediction, survival modelling and long-term simulations of immunotherapy patients using a simulated trial dataset. Methods: A simulated randomized clinical trial dataset for second line immunotherapy of patients with renal cell carcinoma was used for analysis. We machine-learned a dynamic Bayesian network from censored data with incorporation of prior clinical knowledge. Classification performance, probability calibration, goodness-of-fit metrics and prognostic variables were calculated following TRIPOD guidelines. Results: The machine-learned graphical model encoded expected relationships between variables with minimal prior information. Visual and numerical goodness-of-fit checks for survival extrapolations showed that the model fitted data well and was appropriate. Probability calibration was < 10% from ideal. Classification performance for overall survival was high ( c-statistic ~0.85) soon after treatment initiation and gradually plateaued over time. Prognostic variables were calculated by treatment arm for overall survival and severe adverse events. Conclusions: Dynamic Bayesian network model was useful for transparently representing joint relationships between all variables in the trial dataset, for multivariate risk prediction and long-term extrapolations that can be used in economic evaluations for health technology assessments in oncology.

e14081 Background: The identification of characteristics that stratify patients according to the likelihood of treatment response is a key goal of precision oncology research. Within randomized trials, effect modification is typically examined through subgroup analyses or the inclusion of an interaction term(s) in a multivariable regression model. To date, systematic reviews of such findings have relied upon qualitative summaries or subgroup-specific meta-analyses. Such approaches are problematic because they do not quantify the magnitude or the degree of uncertainty in the difference in the treatment effects between subgroups. To address this gap, we propose a novel approach that can be used to quantitatively pool subgroup findings from multiple trials. Methods: Our procedure is focused on the estimation of the pooled difference between the subgroup-specific treatment effects within a two-stage meta-analysis. In the first stage, the magnitude and standard error of the difference between the subgroup-specific treatment effects is estimated for each study. When studying relative effect measures, this difference can be estimated on the log-scale, e.g., study-specific difference = log(RRsubgroup A) – log(RRsubgroup B). The standard error of this difference can then be estimated using subgroup-specific 95% CIs. A meta-analysis of the study-specific differences is then conducted. When the differences are estimated on the log-scale, the pooled quantity can be re-expressed as the ratio of the subgroup specific ORs, RRs, or HRs whereby a ratio equal to 1.00 indicates that the subgroup-specific estimates are equal in magnitude. Results: Case-studies and empirical simulations highlighting the application of our approach will be presented. Potential extensions of this methodology will also be discussed. Such extensions include the use of dose-response models to address subgroup categories of a continuous variable and the inclusion of three or more treatments within a network meta-analysis. Conclusions: This method may help to better identify and quantify the degree of heterogeneity of subgroup differences across trials, particularly in settings where the number of patients within each trial is limited.

e17097 Background: PD-1/PD-L1 inhibitors are immune checkpoint inhibitors widely used in the treatment of metastatic renal cell carcinoma (mRCC) and other cancers. There is a lack of understanding regarding which factors predict response to these therapies. To address this gap, a systematic literature review and meta-analyses were conducted. The objectives of this study were to a) conduct a systematic literature review of studies examining factors that modify the clinical efficacy of PD-1 or PD-L1 inhibitors among patients diagnosed with mRCC and b) quantitatively synthesize the magnitude to which each predictive factor modifies the effect of PD-1/PD-L1 inhibitors. Methods: Electronic databases MEDLINE and COCHRANE were searched for studies published in English from 2006 onwards. We included all phase II/III randomized trials that provided subgroup analyses of any baseline characteristics regarding the effect of PD-1/PD-L1 inhibitors, alone or as part of a combination therapy, with respect to overall or progression-free survival among patients with mRCC. We developed a novel quantitative approach to synthesize subgroup findings across trials. The ratio of the subgroup-specific hazard ratios (HRs) from each study were pooled using a random effects meta-analysis whereby ratios of 1.00 would indicate that the subgroup-specific HRs were equal in magnitude. Results: From an initial 662 studies, five trials were judged to be eligible for inclusion. Meta-analyses suggested the treatment effect of PD-1/PD-L1 inhibitors in mRCC patients was modified by age (overall survival: ratio of HRage > 75 vs. HRage < 65= 1.51; 95% CI:1.01-2.26; I2= 0%, p= 0.04), PD-L1 expression (progression-free survival: ratio of HRPD-L1 > 10% vs. HRPD-L1 < 1%= 2.21; 95% CI:1.14-4.27; I2= 2.26%; p= 0.01), and sarcomatoid tumor presence (progression-free survival: HRno sarcomatoid differentiation vs. HRsarcomatoid differentiation= 1.54; 95% CI:1.07-2.21; I2= 0%; p= 0.02). Conclusions: Evidence suggests that older age, lower levels of PD-L1 expression, and absence of sarcomatoid tumour differentiation diminish the clinical efficacy of anti-PD-1/PD-L1 immunotherapies among mRCC patients.

e20523 Background: Comparing the effectiveness of multiple myeloma treatments presents a challenge due to the limited number of head-to-head trials with which to conduct indirect treatment comparisons. This is particularly true when subgroup analysis is of interest. In comparative effectiveness research Simulated Treatment Comparisons (STCs) are becoming increasingly common in the absence of head-to-head trials. STCs use estimates from limited IPD to adjust for covariate imbalance between trials, however the uncertainty from these estimates is generally ignored when estimating relative treatment effects. This study demonstrates the need to account for this uncertainty when conducting STCs for indications such as multiple myeloma. We introduce an STC method that accounts for the uncertainty due to covariate adjustment, and demonstrate its effectiveness via simulation. Methods: We simulated two single arm studies (N = 300 for both), each containing age and overall survival. We assume study 1 has individual patient data available, and study 2 only has aggregate age data and a digitized Kaplan-Meier curve. We compute a covariate adjustment term based on the mean age difference between the studies and the age coefficients from fitting a parametric survival model to the observed study 1 IPD. We then estimate the variance of this adjustment term via bootstrapping and incorporate this uncertainty into a Bayesian STC model which estimates the relative treatment effect for the two study datasets converted to a digitized Kaplan-Meier format. Results: The proportion of 95% credible intervals (CrI) that captured the true treatment effect was 86.8% without error propagation, whereas 92.0% of CrI’s captured the true treatment with error propagation. 94.9% of CrI’s contained the true treatment effect when using survival regression with the complete IPD. Conclusions: Failing to account for uncertainty from the covariate adjustment when conducting simulated treatment comparisons generally leads to underestimating the uncertainty of the relative treatment effect. This method better captures the uncertainty introduced when conducting an STC and has the potential to yield more reliable estimates of the comparative effectiveness of multiple myeloma treatments.

113 Background: The treatment landscape for M0 CRPC has changed following the demonstrated efficacy of new agents in recent randomized control trials (RCT). However, the comparative effectiveness of these novel agents is unknown. This NMA indirectly compared the efficacy and safety of available therapies for M0 CRPC. Methods: A literature search of MEDLINE (Ovid), EBM Reviews, HealthSTAR, PubMed, PubMed Central, CINAHL, and TRIP Database was performed. Studies were screened by two independent reviewers. Hazard ratios (HR) and confidence intervals were extracted for the primary outcome metastasis-free survival (MFS) and the secondary outcomes overall survival (OS) and grade 3 or higher adverse events (AE). Bone MFS was used as a surrogate for MFS when MFS was not reported. Risk of bias was assessed using the Cochrane Collaboration tool. A Bayesian NMA was performed using a fixed-effects model. Results: Four RCT were analyzed (n=5549). Each trial compared either apalutamide (APA), enzalutamide (ENZA), darolutamide (DARO), or denosumab (DENO) plus androgen deprivation therapy (ADT) to placebo plus ADT. Risk of bias was low. For MFS, APA and ENZA had similar efficacy (Table), and Surface Under the Cumulative Ranking Analysis demonstrated a 59% probability that APA was preferred for MFS, followed by ENZA (41%). There was a trend for improved OS for APA, DARO and ENZA, but no meaningful differences between these agents. APA, ENZA, and DARO had a similar risk of AEs and all had a greater risk of AEs compared to DENO. Conclusions: APA and ENZA appear to be the most efficacious treatments for MFS in M0 CRPC, though more data for OS is required. Compared to DARO, APA and ENZA’s demonstrated efficacy is not at the expense of added toxicity.[Table: see text]

e18021 Background: In cancer, accumulation of somatic and genomic changes occurs from early to late stage disease. Recent data suggest that never smokers (NVS) vs. ever smokers (EVS) develop biolog...