Natural History and Characteristics of ERBB2-mutated Hormone Receptor-positive Metastatic Breast Cancer: A Multi-institutional Retrospective Case-control Study from AACR Project GENIE.

PURPOSE: We wanted to determine the prognosis and the phenotypic characteristics of hormone receptor-positive advanced breast cancer tumors harboring an ERBB2 mutation in the absence of a HER2 amplification. EXPERIMENTAL DESIGN: We retrospectively collected information from the American Association of Cancer Research-Genomics Evidence Neoplasia Information Exchange registry database from patients with hormone receptor-positive, HER2-negative, ERBB2-mutated advanced breast cancer. Phenotypic and co-mutational features, as well as response to treatment and outcome were compared with matched control cases ERBB2 wild type. RESULTS: A total of 45 ERBB2-mutant cases were identified for 90 matched controls. The presence of an ERBB2 mutation was not associated with worse outcome determined by overall survival (OS) from first metastatic relapse. No significant differences were observed in phenotypic characteristics apart from higher lobular infiltrating subtype in the ERBB2-mutated group. ERBB2 mutation did not seem to have an impact in response to treatment or time-to-progression (TTP) to endocrine therapy compared with ERBB2 wild type. In the co-mutational analyses, CDH1 mutation was more frequent in the ERBB2-mutated group (FDR < 1). Although not significant, fewer co-occurring ESR1 mutations and more KRAS mutations were identified in the ERBB2-mutated group. CONCLUSIONS: ERBB2-activating mutation was not associated with a worse OS from time of first metastatic relapse, or differences in TTP on treatment as compared with a series of matched controls. Although not significant, differences in coexisting mutations (CDH1, ESR1, and KRAS) were noted between the ERBB2-mutated and the control group.

A Scalable Quality Assurance Process for Curating Oncology Electronic Health Records: The Project GENIE Biopharma Collaborative Approach.

PURPOSE: The American Association for Cancer Research Project Genomics Evidence Neoplasia Information Exchange Biopharma Collaborative is a multi-institution effort to build a pan-cancer repository of genomic and clinical data curated from the electronic health record. For the research community to be confident that data extracted from electronic health record text are reliable, transparency of the approach used to ensure data quality is essential. MATERIALS AND METHODS: Four institutions participating in AACR's Project GENIE created an observational cohort of patients with cancer for whom tumor molecular profiling data, therapeutic exposures, and treatment outcomes are available and will be shared publicly with the research community. A comprehensive approach to quality assurance included assessments of (1) feasibility of the curation model through pressure test cases; (2) accuracy through programmatic queries and comparison with source data; and (3) reproducibility via double curation and code review. RESULTS: Assessments of feasibility resulted in critical modifications to the curation directives. Queries and comparison with source data identified errors that were rectified via data correction and curator retraining. Assessment of intercurator reliability indicated a reliable curation model. CONCLUSION: The transparent quality assurance processes for the GENIE BPC data ensure that the data can be used for analyses that support clinical decision making and advances in precision oncology.

Implications of Selection Bias Due to Delayed Study Entry in Clinical Genomic Studies.

IMPORTANCE: Real-world data sets that combine clinical and genomic data may be subject to left truncation (when potential study participants are not included because they have already passed the milestone of interest at the time of study recruitment). The lapse between diagnosis and molecular testing can present analytic challenges and threaten the validity and interpretation of survival analyses. OBSERVATIONS: Effects of ignoring left truncation when estimating overall survival are illustrated using data from the American Association for Cancer Research (AACR) Project Genomics Evidence Neoplasia Information Exchange Biopharma Collaborative (GENIE BPC), and a straightforward risk-set adjustment approach is described. Ignoring left truncation results in overestimation of overall survival: unadjusted median survival estimates from diagnosis among patients with stage IV non-small cell lung cancer or stage IV colorectal cancer were overestimated by more than 1 year. CONCLUSIONS AND RELEVANCE: Clinicogenomic data are a valuable resource for evaluation of real-world cancer outcomes and should be analyzed using appropriate methods to maximize their potential. Analysts must become adept at application of appropriate statistical methods to ensure valid, meaningful, and generalizable research findings.

Correlation Between Surrogate End Points and Overall Survival in a Multi-institutional Clinicogenomic Cohort of Patients With Non-Small Cell Lung or Colorectal Cancer.

Importance: Contemporary observational cancer research requires associating genomic biomarkers with reproducible end points; overall survival (OS) is a key end point, but interpretation can be challenging when multiple lines of therapy and prolonged survival are common. Progression-free survival (PFS), time to treatment discontinuation (TTD), and time to next treatment (TTNT) are alternative end points, but their utility as surrogates for OS in real-world clinicogenomic data sets has not been well characterized. Objective: To measure correlations between candidate surrogate end points and OS in a multi-institutional clinicogenomic data set. Design, Setting, and Participants: A retrospective cohort study was conducted of patients with non-small cell lung cancer (NSCLC) or colorectal cancer (CRC) whose tumors were genotyped at 4 academic centers from January 1, 2014, to December 31, 2017, and who initiated systemic therapy for advanced disease. Patients were followed up through August 31, 2020 (NSCLC), and October 31, 2020 (CRC). Statistical analyses were conducted on January 5, 2021. Exposures: Candidate surrogate end points included TTD; TTNT; PFS based on imaging reports only; PFS based on medical oncologist ascertainment only; PFS based on either imaging or medical oncologist ascertainment, whichever came first; and PFS defined by a requirement that both imaging and medical oncologist ascertainment have indicated progression. Main Outcomes and Measures: The primary outcome was the correlation between candidate surrogate end points and OS. Results: There were 1161 patients with NSCLC (648 women [55.8%]; mean [SD] age, 63 [11] years) and 1150 with CRC (647 men [56.3%]; mean [SD] age, 54 [12] years) identified for analysis. Progression-free survival based on both imaging and medical oncologist documentation was most correlated with OS (NSCLC: ρ = 0.76; 95% CI, 0.73-0.79; CRC: ρ = 0.73; 95% CI, 0.69-0.75). Time to treatment discontinuation was least associated with OS (NSCLC: ρ = 0.45; 95% CI, 0.40-0.50; CRC: ρ = 0.13; 95% CI, 0.06-0.19). Time to next treatment was modestly associated with OS (NSCLC: ρ = 0.60; 0.55-0.64; CRC: ρ = 0.39; 95% CI, 0.32-0.46). Conclusions and Relevance: This cohort study suggests that PFS based on both a radiologist and a treating oncologist determining that a progression event has occurred was the surrogate end point most highly correlated with OS for analysis of observational clinicogenomic data.

Clinical Inflection Point Detection on the Basis of EHR Data to Identify Clinical Trial-Ready Patients With Cancer.

PURPOSE: To inform precision oncology, methods are needed to use electronic health records (EHRs) to identify patients with cancer who are experiencing clinical inflection points, consistent with worsening prognosis or a high propensity to change treatment, at specific time points. Such patients might benefit from real-time screening for clinical trials. METHODS: Using serial unstructured imaging reports for patients with solid tumors or lymphoma participating in a single-institution precision medicine study, we trained a deep neural network natural language processing (NLP) model to dynamically predict patients' prognoses and propensity to start new palliative-intent systemic therapy within 30 days. Model performance was evaluated using Harrell's c-index (for prognosis) and the area under the receiver operating characteristic curve (AUC; for new treatment and new clinical trial enrollment). Associations between model outputs and manual annotations of cancer progression were also evaluated using the AUC. RESULTS: A deep NLP model was trained and evaluated using 302,688 imaging reports for 16,780 patients. In a held-out test set of 34,770 reports for 1,952 additional patients, the model predicted survival with a c-index of 0.76 and initiation of new treatment with an AUC of 0.77. Model-generated prognostic scores were associated with annotation of cancer progression on the basis of manual EHR review (n = 1,488 reports for 110 patients with lung or colorectal cancer) with an AUC of 0.78, and predictions of new treatment were associated with annotation of cancer progression on the basis of manual EHR review with an AUC of 0.84. CONCLUSION: Training a deep NLP model to identify clinical inflection points among patients with cancer is feasible. This approach could identify patients who may benefit from real-time targeted clinical trial screening interventions at health system scale.

Characteristics and Outcome of AKT1 E17K-Mutant Breast Cancer Defined through AACR Project GENIE, a Clinicogenomic Registry.

AKT inhibitors have promising activity in AKT1 E17K-mutant estrogen receptor (ER)-positive metastatic breast cancer, but the natural history of this rare genomic subtype remains unknown. Utilizing AACR Project GENIE, an international clinicogenomic data-sharing consortium, we conducted a comparative analysis of clinical outcomes of patients with matched AKT1 E17K-mutant (n = 153) and AKT1-wild-type (n = 302) metastatic breast cancer. AKT1-mutant cases had similar adjusted overall survival (OS) compared with AKT1-wild-type controls (median OS, 24.1 vs. 29.9, respectively; P = 0.98). AKT1-mutant cases enjoyed longer durations on mTOR inhibitor therapy, an observation previously unrecognized in pivotal clinical trials due to the rarity of this alteration. Other baseline clinicopathologic features, as well as durations on other classes of therapy, were broadly similar. In summary, we demonstrate the feasibility of using a novel and publicly accessible clincogenomic registry to define outcomes in a rare genomically defined cancer subtype, an approach with broad applicability to precision oncology. SIGNIFICANCE: We delineate the natural history of a rare genomically distinct cancer, AKT1 E17K-mutant ER-positive breast cancer, using a publicly accessible registry of real-world patient data, thereby illustrating the potential to inform drug registration through synthetic control data.See related commentary by Castellanos and Baxi, p. 490.

Assessment of Deep Natural Language Processing in Ascertaining Oncologic Outcomes From Radiology Reports.

IMPORTANCE: A rapid learning health care system for oncology will require scalable methods for extracting clinical end points from electronic health records (EHRs). Outside of clinical trials, end points such as cancer progression and response are not routinely encoded into structured data. OBJECTIVE: To determine whether deep natural language processing can extract relevant cancer outcomes from radiologic reports, a ubiquitous but unstructured EHR data source. DESIGN, SETTING, AND PARTICIPANTS: A retrospective cohort study evaluated 1112 patients who underwent tumor genotyping for a diagnosis of lung cancer and participated in the Dana-Farber Cancer Institute PROFILE study from June 26, 2013, to July 2, 2018. EXPOSURES: Patients were divided into curation and reserve sets. Human abstractors applied a structured framework to radiologic reports for the curation set to ascertain the presence of cancer and changes in cancer status over time (ie, worsening/progressing vs improving/responding). Deep learning models were then trained to capture these outcomes from report text and subsequently evaluated in a 10% held-out test subset of curation patients. Cox proportional hazards regression models compared human and machine curations of disease-free survival, progression-free survival, and time to improvement/response in the curation set, and measured associations between report classification and overall survival in the curation and reserve sets. MAIN OUTCOMES AND MEASURES: The primary outcome was area under the receiver operating characteristic curve (AUC) for deep learning models; secondary outcomes were time to improvement/response, disease-free survival, progression-free survival, and overall survival. RESULTS: A total of 2406 patients were included (mean [SD] age, 66.5 [10.8] years; 1428 female [59.7%]; 2170 [90.2%] white). Radiologic reports (n = 14 230) were manually reviewed for 1112 patients in the curation set. In the test subset (n = 109), deep learning models identified the presence of cancer, improvement/response, and worsening/progression with accurate discrimination (AUC >0.90). Machine and human curation yielded similar measurements of disease-free survival (hazard ratio [HR] for machine vs human curation, 1.18; 95% CI, 0.71-1.95); progression-free survival (HR, 1.11; 95% CI, 0.71-1.71); and time to improvement/response (HR, 1.03; 95% CI, 0.65-1.64). Among 15 000 additional reports for 1294 reserve set patients, algorithm-detected cancer worsening/progression was associated with decreased overall survival (HR for mortality, 4.04; 95% CI, 2.78-5.85), and improvement/response was associated with increased overall survival (HR, 0.41; 95% CI, 0.22-0.77). CONCLUSIONS AND RELEVANCE: Deep natural language processing appears to speed curation of relevant cancer outcomes and facilitate rapid learning from EHR data.

Time to adjuvant chemotherapy for breast cancer in National Comprehensive Cancer Network institutions.

BACKGROUND: High-quality care must be not only appropriate but also timely. We assessed time to initiation of adjuvant chemotherapy for breast cancer as well as factors associated with delay to help identify targets for future efforts to reduce unnecessary delays. METHODS: Using data from the National Comprehensive Cancer Network (NCCN) Outcomes Database, we assessed the time from pathological diagnosis to initiation of chemotherapy (TTC) among 6622 women with stage I to stage III breast cancer diagnosed from 2003 through 2009 and treated with adjuvant chemotherapy in nine NCCN centers. Multivariable models were constructed to examine factors associated with TTC. All statistical tests were two-sided. RESULTS: Mean TTC was 12.0 weeks overall and increased over the study period. A number of factors were associated with a longer TTC. The largest effects were associated with therapeutic factors, including immediate postmastectomy reconstruction (2.7 weeks; P < .001), re-excision (2.1 weeks; P < .001), and use of the 21-gene reverse-transcription polymerase chain reaction assay (2.2 weeks; P < .001). In comparison with white women, a longer TTC was observed among black (1.5 weeks; P < .001) and Hispanic (0.8 weeks; P < .001) women. For black women, the observed disparity was greater among women who transferred their care to the NCCN center after diagnosis (P (interaction) = .008) and among women with Medicare vs commercial insurance (P (interaction) < .001). CONCLUSIONS: Most observed variation in TTC was related to use of appropriate therapeutic interventions. This suggests the importance of targeted efforts to minimize potentially preventable causes of delay, including inefficient transfers in care or prolonged appointment wait times.

Influential factors for post-fellowship career decision-making: an NCCN survey.

Insight into factors important to fellows' decision-making about their career paths is critical to successfully developing program curricula, making capacity projections, and recruiting oncology physicians. This study was performed to determine the factors associated with post-fellowship career decision-making. Program evaluation surveys were administered to oncology fellows who attended the Fellows Recognition Program at the 2009 NCCN Annual Conference. A total of 125 (75%) fellows completed the initial survey. Overall, 73% of fellows reported participating in clinical research and 58% received formal training as part of their fellowship program. Receipt of formal training was correlated with greater program satisfaction (r(s) = 0.20; P = .03), feeling more prepared for a post-fellowship career (r(s) = 0.30; P < .001), and greater interest in clinical research post fellowship (r(s) = 0.32; P < .001). Interest in post-fellowship clinical research (r(s) = 0.49; P < .001) and importance of protected academic time (r(s) = 0.57; P < .001) were strongly correlated with interest in practicing in an academic environment, whereas institutional reputation (r(s) = 0.18; P = .04) and a multidisciplinary practice environment (r(s) = 0.22; P = .02) were moderately associated with interest. Location, salary, multidisciplinary environment, and flexible scheduling were the most important controllable lifestyle (CL) factors. These results suggest that fellowship programs may be able to foster a desire to participate in research and subsequent interest in practicing in an academic institution through providing opportunities for formal training in clinical research skills. However, even in an academic setting, CL factors are important to attracting and retaining faculty.

Comparative outcome of initial therapy for younger patients with mantle cell lymphoma: an analysis from the NCCN NHL Database.

Few randomized trials have compared therapies in mantle cell lymphoma (MCL), and the role of aggressive induction is unclear. The National Comprehensive Cancer Network (NCCN) Non-Hodgkin Lymphoma (NHL) Database, a prospective cohort study collecting clinical, treatment, and outcome data at 7 NCCN centers, provides a unique opportunity to compare the effectiveness of initial therapies in MCL. Patients younger than 65 diagnosed between 2000 and 2008 were included if they received RHCVAD (rituximab fractionated cyclophosphamide, vincristine, adriamycin, and dexamethasone), RCHOP+HDT/ASCR (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone + high-dose therapy/autologous stem cell rescue), RHCVAD+HDT/ASCR, or RCHOP. Clinical parameters were similar for patients treated with RHCVAD (n = 83, 50%), RCHOP+HDT/ASCR (n = 34, 20%), RCHOP (n = 29, 17%), or RHCVAD+HDT/ASCR (n = 21, 13%). Overall, 70 (42%) of the 167 patients progressed and 25 (15%) expired with a median follow-up of 33 months. There was no difference in progression-free survival (PFS) between aggressive regimens (P > .57), which all demonstrated superior PFS compared with RCHOP (P < .004). There was no difference in overall survival (OS) between the RHCVAD and RCHOP+HDT/ASCR (P = .98). RCHOP was inferior to RHCVAD and RCHOP+HDT/ASCR, which had similar PFS and OS. Despite aggressive regimens, the median PFS was 3 to 4 years. Future trials should focus on novel agents rather than comparing current approaches.

Reliability of staging, prognosis, and comorbidity data collection in the National Comprehensive Cancer Network (NCCN) non-Hodgkin lymphoma (NHL) multicenter outcomes database.

BACKGROUND: Clinical trials and outcomes studies often rely on nonphysicians to abstract complex data from medical records, but the reliability of these data are rarely assessed. METHODS: We used standardized charts of patients with non-Hodgkin lymphoma to assess the reliability of key clinical data elements abstracted by 6 clinical research associates (CRAs), 3 project staff, and 3 medical oncologists. We assessed reliability on 5 variables: MD-reported and rater-determined disease stage; International Prognostic Index (IPI; low-low intermediate, intermediate-high, high); Charlson comorbidity index score; and presence of any item from the Charlson index. Intraclass correlation coefficients (ICCs) of 0-0.20 were indicative of "slight", 0.21-0.40 indicated "fair", 0.41-0.60 indicated "moderate", 0.61-0.80 "substantial" and >0.80 "almost perfect" reliability. RESULTS: By outcome, the ICC (95% confidence interval) values for MD-reported stage, rater-determined stage, and IPI were 0.86 (0.67, 0.94), 0.82 (0.59, 0.93), and 0.80 (0.55, 0.92), respectively. In contrast, the ICC (95% confidence interval) of the Charlson score, or presence of any Charlson comorbidity item was 0.47 (0.03, 0.75) and 0.61 (0.23, 0.83), respectively. Reliability varied by rater group; no rater group was consistently more reliable than others. CONCLUSIONS: Trained CRAs abstracted key clinical variables with a very high degree of reliability, and performed at a level similar to study trainers and oncologists. Elements of the Charlson index were less reliable than other data types, possibly because of inherent ambiguity in the index itself.

Documentation of pain in comprehensive cancer centers in the United States: a preliminary analysis.

The National Comprehensive Cancer Network (NCCN), an organization of 19 of the world's leading cancer centers, developed and communicated a cancer pain treatment guideline. NCCN seeks to implement guidelines through performance measurement using a NCCN Oncology Outcomes Database. This is a preliminary report from the NCCN Cancer Pain Management Database Project. The primary objective of this NCCN Cancer Pain Management Database Project study is to evaluate the frequency, methods, and extent of documentation of cancer pain assessment and management at NCCN institutions. A pain data dictionary and related data collection forms were first developed. The records of 209 breast cancer patients with bone metastases were then studied. The frequency of pain mentions, type of pain assessment tool used, pain characteristics, type of clinician documenting pain, location in the medical record, and pain treatment characteristics were noted. The majority of clinical encounters included pain mentions, although considerable variability was found in pain documentation between providers and between inpatient and outpatient settings. Nurses more frequently recorded pain, usually as a numeric pain intensity score. Pain specialists were more likely to record a complete description of pain. A significant minority of patients experienced moderate to severe pain. In a small subgroup of patients with moderate to severe pain, pain treatment was not recorded. The undertreatment of cancer pain has been a focus of investigation and review for the past two decades. Quality improvement efforts to raise the standard of pain management have been underway. The results of this study highlight the need for standardization of pain documentation in comprehensive cancer centers as a prerequisite for the proper assessment of cancer pain and the improvement of clinical outcomes of pain management.