IMPORTANCE Data sets linking comprehensive genomic profiling (CGP) to clinical outcomes may accelerate precision medicine.OBJECTIVE To assess whether a database that combines EHR-derived clinical data with CGP can identify and extend associations in non-small cell lung cancer (NSCLC).DESIGN, SETTING, AND PARTICIPANTS Clinical data from EHRs were linked with CGP results for 28 998 patients from 275 US oncology practices. Among 4064 patients with NSCLC, exploratory associations between tumor genomics and patient characteristics with clinical outcomes were conducted, with data obtained between January 1, 2011, and January 1, 2018.EXPOSURES Tumor CGP, including presence of a driver alteration (a pathogenic or likely pathogenic alteration in a gene shown to drive tumor growth); tumor mutation burden (TMB), defined as the number of mutations per megabase; and clinical characteristics gathered from EHRs. MAIN OUTCOMES AND MEASURESOverall survival (OS), time receiving therapy, maximal therapy response (as documented by the treating physician in the EHR), and clinical benefit rate (fraction of patients with stable disease, partial response, or complete response) to therapy. RESULTS Among 4064 patients with NSCLC (median age, 66.0 years; 51.9% female), 3183 (78.3%) had a history of smoking, 3153 (77.6%) had nonsquamous cancer, and 871 (21.4%) had an alteration in EGFR, ALK, or ROS1 (701 [17.2%] with EGFR, 128 [3.1%] with ALK, and 42 [1.0%] with ROS1 alterations). There were 1946 deaths in 7 years. For patients with a driver alteration, improved OS was observed among those treated with (n = 575) vs not treated with (n = 560) targeted therapies (median, 18.6 months [95% CI, 15.2-21.7] vs 11.4 months [95% CI, 9.7-12.5] from advanced diagnosis; P < .001). TMB (in mutations/Mb) was significantly higher among smokers vs nonsmokers (8.7 [IQR,] vs 2.6 [IQR, 1.7-5.2]; P < .001) and significantly lower among patients with vs without an alteration in EGFR (3.5 [IQR, 1.76-6.1] vs 7.8 [IQR, 3.5-13.9]; P < .001), ALK (2.1 [IQR, 0.9-4.0] vs 7.0 [IQR, 3.5-13.0]; P < .001), RET (4.6 [IQR,] vs 7.0 [IQR, 2.6-13.0]; P = .004), or ROS1 (4.0 [IQR, 1.2-9.6] vs 7.0 [IQR, 2.6-13.0]; P = .03). In patients treated with anti-PD-1/PD-L1 therapies (n = 1290, 31.7%), TMB of 20 or more was significantly associated with improved OS from therapy initiation (16.8 months [95% CI, 11.6-24.9] vs 8.5 months [95% CI, 7.6-9.7]; P < .001), longer time receiving therapy (7.8 months [95% CI, 5.5-11.1] vs 3.3 months [95% CI, 2.8-3.7]; P < .001), and increased clinical benefit rate (80.7% vs 56.7%; P < .001) vs TMB less than 20.CONCLUSIONS AND RELEVANCE Among patients with NSCLC included in a longitudinal database of clinical data linked to CGP results from routine care, exploratory analyses replicated previously described associations between clinical and genomic characteristics, between driver mutations and response to targeted therapy, and between TMB and response to immunotherapy. These findings demonstrate the feasibility of creating a clinicogenomic database der...

IntroductionReal-world evidence derived from electronic health records (EHRs) is increasingly recognized as a supplement to evidence generated from traditional clinical trials. In oncology, tumor-based Response Evaluation Criteria in Solid Tumors (RECIST) endpoints are standard clinical trial metrics. The best approach for collecting similar endpoints from EHRs remains unknown. We evaluated the feasibility of a RECIST-based methodology to assess EHR-derived real-world progression (rwP) and explored non-RECIST-based approaches.MethodsIn this retrospective study, cohorts were randomly selected from Flatiron Health’s database of de-identified patient-level EHR data in advanced non-small cell lung cancer. A RECIST-based approach tested for feasibility (N = 26). Three non-RECIST approaches were tested for feasibility, reliability, and validity (N = 200): (1) radiology-anchored, (2) clinician-anchored, and (3) combined. Qualitative and quantitative methods were used.ResultsA RECIST-based approach was not feasible: cancer progression could be ascertained for 23% (6/26 patients). Radiology- and clinician-anchored approaches identified at least one rwP event for 87% (173/200 patients). rwP dates matched 90% of the time. In 72% of patients (124/173), the first clinician-anchored rwP event was accompanied by a downstream event (e.g., treatment change); the association was slightly lower for the radiology-anchored approach (67%; 121/180). Median overall survival (OS) was 17 months [95% confidence interval (CI) 14, 19]. Median real-world progression-free survival (rwPFS) was 5.5 months (95% CI 4.6, 6.3) and 4.9 months (95% CI 4.2, 5.6) for clinician-anchored and radiology-anchored approaches, respectively. Correlations between rwPFS and OS were similar across approaches (Spearman’s rho 0.65–0.66). Abstractors preferred the clinician-anchored approach as it provided more comprehensive context.ConclusionsRECIST cannot adequately assess cancer progression in EHR-derived data because of missing data and lack of clarity in radiology reports. We found a clinician-anchored approach supported by radiology report data to be the optimal, and most practical, method for characterizing tumor-based endpoints from EHR-sourced data.FundingFlatiron Health Inc., which is an independent subsidiary of the Roche group.Electronic supplementary materialThe online version of this article (10.1007/s12325-019-00970-1) contains supplementary material, which is available to authorized users.

Real-world evidence derived from electronic health records (EHRs) is increasingly recognized as a supplement to evidence generated from traditional clinical trials. In oncology, tumor-based Response Evaluation Criteria in Solid Tumors (RECIST) endpoints are collected in clinical trials. The best approach for collecting similar endpoints from EHRs remains unknown.We evaluated the feasibility of a traditional RECIST-based methodology to assess EHR-derived real-world progression (rwP) and explored non-RECIST-based approaches. In this retrospective study, cohorts were randomly selected from Flatiron Health's database of patient-level EHR data in advanced non-small cell lung cancer. A RECIST-based approach was tested for feasibility (N=26). Three non-RECIST abstraction approaches were tested for feasibility, reliability, and validity (N=200): (1) radiology-anchored, (2) clinician-anchored, and (3) combined.RECIST-based cancer progression could be ascertained from the EHRs of 23% of patients (6/26).In 87% of patients (173/200), at least one rwP event was identified using both the radiologyand clinician-anchored approaches. rwP dates matched 90% of the time. In 72% of patients (124/173), the first clinician-anchored rwP event was accompanied by a downstream event (e.g., treatment change); the association was slightly lower for the radiology-anchored approach (67%; 121/180). Median overall survival (OS) was 17 months (95% confidence interval [CI]: 14, 19). Median real-world progression-free survival (rwPFS) was 5.5 (95% CI: 4.6, 6.3) and 4.9 months (95% CI: 4.2, 5.6) for clinician-anchored and radiology-anchored approaches, respectively. Correlations between rwPFS and OS were similar across approaches (Spearman's rho: 0.65-0.66). Abstractors preferred the clinician-anchored approach as it provided more

Introduction Effectiveness metrics for real-word research, analogous to clinical trial ones, are needed. This study aimed to develop a real-world response (rwR) variable applicable to solid tumors and to evaluate its clinical relevance and meaningfulness. Methods This retrospective study used patient cohorts with advanced non-small cell lung cancer from a nationwide, de-identified electronic health record (EHR)-derived database. Disease burden information abstracted manually was classified into response categories anchored to discrete therapy lines (per patient-line). In part 1, we quantified the feasibility and reliability of data capture, and estimated the association between rwR status and real-world progression-free survival (rwPFS) and real-world overall survival (rwOS). In part 2, we investigated the correlation between published clinical trial overall response rates (ORRs) and real-world response rates (rwRRs) from corresponding real-world patient cohorts. Results In part 1, 85.4% of patients ( N = 3248) had  at least one radiographic assessment documented. Median abstraction time per patient-line was 15.0 min (IQR 7.8–28.1). Inter-abstractor agreement on presence/absence of at least one assessment was 0.94 (95% CI 0.92–0.96; n = 503 patient-lines abstracted in duplicate); inter-abstractor agreement on best confirmed response category was 0.82 (95% CI 0.78–0.86; n = 384 with at least one captured assessment). Confirmed responders at a 3-month landmark showed significantly lower risk of death and progression in rwOS and rwPFS analyses across all line settings. In part 2, rwRRs (from 12 rw cohorts) showed a high correlation with trial ORRs (Spearman’s ρ = 0.99). Conclusions We developed a rwR variable generated from clinician assessments documented in EHRs following radiographic evaluations. This variable provides clinically meaningful information and may provide a real-world measure of treatment effectiveness. Supplementary Information The online version contains supplementary material available at 10.1007/s12325-021-01659-0.

2514 Background: Genomic findings have diagnostic, prognostic, and predictive utility in clinical oncology. Population studies have been limited by reliance on trials, registries, or institutional chart review, which are costly and represent narrow populations. Integrating electronic health record (EHR) and genomic data collected as part of routine clinical practice may overcome these hurdles. Methods: Patients in the Flatiron Health Database with non-small cell lung cancer (NSCLC) who underwent comprehensive genomic profiling (CGP) by Foundation Medicine were included. EHR processing included structured data harmonization and abstraction of variables from unstructured documents. EHR and CGP data were de-identified and linked in a HIPAA-compliant process. Data included clinical characteristics, alterations across > 300 genes, tumor mutation burden (TMB), therapies and associated real-world responses, progression, and overall survival (OS). Results: The cohort (n = 1619) had expected clinical (mean age 66; 75% with smoking hx; 80% non-squamous) and genomic (18% EGFR; 4% ALK; 1% ROS1) properties of NSCLC. Presence of a driver mutation (EGFR, ALK, ROS1, MET, BRAF, RET, or ERBB2; n = 576) was associated with younger age, female gender, non-smoking, improved OS (35 vs 19 mo, LR p < 0.0001), and prolonged survival when treated with NCCN-recommended therapy (42 vs 28 mo, LR p = 0.001). CGP identified false negative results in up to 30% of single-biomarker tests for EGFR, ALK, and ROS1. CGP accuracy was supported by clinical outcomes. For example, 5 patients with prior negative ALK-fusion testing began ALK-directed therapy after positive CGP results. All 5 exhibited at least a partial response as recorded in the EHR by treating clinicians. Immunotherapy was used in 22% of patients (n = 353). TMB predicted response to nivolumab, including in PD-L1 negative populations. We recapitulated known associations with smoking, histology, and driver mutations. Conclusions: We present and validate a new paradigm for rapidly generating large, research-grade, longitudinal clinico-genomic databases by linking genomic data with EHR clinical annotation. This method offers a powerful tool for understanding cancer genomics and advancing precision medicine.