Systematic analysis of off-label and off-guideline cancer therapy usage in a real-world cohort of 165,912 US patients.

Patients with cancer may be given treatments that are not officially approved (off-label) or recommended by guidelines (off-guideline). Here we present a data science framework to systematically characterize off-label and off-guideline usages using real-world data from de-identified electronic health records (EHR). We analyze treatment patterns in 165,912 US patients with 14 common cancer types. We find that 18.6% and 4.4% of patients have received at least one line of off-label and off-guideline cancer drugs, respectively. Patients with worse performance status, in later lines, or treated at academic hospitals are significantly more likely to receive off-label and off-guideline drugs. To quantify how predictable off-guideline usage is, we developed machine learning models to predict which drug a patient is likely to receive based on their clinical characteristics and previous treatments. Finally, we demonstrate that our systematic analyses generate hypotheses about patients' response to treatments.

Applying the estimand and target trial frameworks to external control analyses using observational data: a case study in the solid tumor setting.

Introduction: In causal inference, the correct formulation of the scientific question of interest is a crucial step. The purpose of this study was to apply causal inference principles to external control analysis using observational data and illustrate the process to define the estimand attributes. Methods: This study compared long-term survival outcomes of a pooled set of three previously reported randomized phase 3 trials studying patients with metastatic non-small cell lung cancer receiving front-line chemotherapy and similar patients treated with front-line chemotherapy as part of routine clinical care. Causal inference frameworks were applied to define the estimand aligned with the research question and select the estimator to estimate the estimand of interest. Results: The estimand attributes of the ideal trial were defined using the estimand framework. The target trial framework was used to address specific issues in defining the estimand attributes using observational data from a nationwide electronic health record-derived de-identified database. The two frameworks combined allow to clearly define the estimand and the aligned estimator while accounting for key baseline confounders, index date, and receipt of subsequent therapies. The hazard ratio estimate (point estimate with 95% confidence interval) comparing the randomized clinical trial pooled control arm with the external control was close to 1, which is indicative of similar survival between the two arms. Discussion: The proposed combined framework provides clarity on the causal contrast of interest and the estimator to adopt, and thus facilitates design and interpretation of the analyses.

Identification of non-actionable mutations with prognostic and predictive value in patients with advanced or metastatic non-small cell lung cancer.

INTRODUCTION: Lung cancer is one of the most prevalent cancers and the leading cause of cancer death. Advanced non-small cell lung cancer (aNSCLC) patients frequently harbor mutations that impact their survival outcomes. There are limited data regarding the prognostic and predictive significance of these mutations on survival outcomes in the real-world setting. METHODS: This observational retrospective study analyzed de-identified electronic medical records from the Flatiron Health Clinico-Genomic and FoundationCore® databases to identify patients with aNSCLC who initiated first-line immune checkpoint inhibitors (ICI; alone or in combination) or chemotherapy under routine care between 2016 and 2021. The primary objectives were to assess the prevalence of non-actionable mutations and to determine their association with overall survival (OS). Real-world progression-free survival (rwPFS) and real-world response (rwR) were investigated as secondary exploratory outcomes. RESULTS: Based on an assessment of 185 non-actionable mutations in 2999 patients, the most prevalent mutations were TP53 (70%), KRAS (42%), CDKN2A/B (31%), and STK11 (21%). STK11, KEAP1, and CDKN2A/B mutations were significantly associated with lower rwR, shorter rwPFS and OS. KRAS mutations were clinically associated with shorter rwPFS in CIT-treated patients. Subgroup analysis revealed that fast progressors were significantly more likely to harbor STK11, KEAP1, and CDKN2A/B mutations. Accordingly, long-term survivors (LTS) showed a significantly lower prevalence of these mutations. CONCLUSION: Our results provide evidence on the prognostic value of STK11, KEAP1, and CDKN2A/B mutations in patients with aNSCLC. Further research is required to better understand the implications of these findings on patient management and future trial design and treatment selection.

Overall survival in patients with advanced non-small cell lung cancer with KRAS G12C mutation with or without STK11 and/or KEAP1 mutations in a real-world setting.

BACKGROUND: KRAS mutations occur frequently in advanced non-small cell lung cancer (aNSCLC); the G12C mutation is the most prevalent. Alterations in STK11 or KEAP1 commonly co-occur with KRAS mutations in aNSCLC. Using real-world data, we assessed the effect of KRAS G12C mutation with or without STK11 and/or KEAP1 mutations on overall survival (OS) in patients with aNSCLC receiving cancer immunotherapy (CIT), chemotherapy, or both in first line (1L) and second line (2L). METHODS: Patients diagnosed with aNSCLC between January 2011 and March 2020 in a clinico-genomic database were included. Cox proportional hazards models adjusted for left truncation, baseline demographics and clinical characteristics were used to analyze the effect of STK11 and/or KEAP1 co-mutational status on OS in patients with KRAS wild-type (WT) or G12C mutation. RESULTS: Of 2715 patients with aNSCLC without other actionable driver mutations, 1344 (49.5%) had KRAS WT cancer, and 454 (16.7%) had KRAS G12C-positive cancer. At 1L treatment start, significantly more patients with KRAS G12C-positive cancer were female, smokers, and had non-squamous histology, a higher prevalence of metastasis and programmed death-ligand 1 positivity than those with KRAS WT cancer. Median OS was comparable between patients with KRAS G12C-positive and KRAS WT cancer when receiving chemotherapy or combination CIT and chemotherapy in the 1L or 2L. Median OS was numerically longer in patients with KRAS G12C vs KRAS WT cancer treated with 1L CIT (30.2 vs 10.6 months, respectively) or 2L CIT (11.3 vs 7.6 months, respectively). Co-mutation of STK11 and KEAP1 was associated with significantly shorter OS in patients receiving any type of 1L therapy, regardless of KRAS G12C mutational status. CONCLUSIONS: This real-world study showed that patients with KRAS G12C-positive or KRAS WT cancer have similar OS in the 1L or 2L when treated with chemotherapy or combination CIT and chemotherapy. In contrast to aNSCLC patients with EGFR or ALK driver mutations, patients with KRAS G12C-positive cancer may benefit from CIT monotherapy. Co-mutation of STK11 and KEAP1 was associated with significantly shorter survival, independent of KRAS G12C mutational status, reflecting the poor prognosis and high unmet need in this patient population.

Real-World Comparative Effectiveness of First-Line Alectinib Versus Crizotinib in Patients With Advanced ALK-Positive NSCLC With or Without Baseline Central Nervous System Metastases.

Introduction: Alectinib was found to have superior efficacy to crizotinib in the phase 3 ALEX study and is a preferred initial treatment for patients with advanced ALK-positive NSCLC. To understand the efficacy of alectinib in U.S. clinical practice, we conducted a retrospective real-world comparative effectiveness analysis of first-line alectinib versus crizotinib. Methods: Adults with advanced ALK-positive NSCLC who received first-line alectinib (from December 11, 2015) or crizotinib (from January 1, 2014) were included from a real-world database. Propensity scores were applied to balance baseline characteristics. Real-world data (RWD), including real-world progression-free survival (rwPFS), real-world overall survival, real-world time to new central nervous system (CNS) metastases, and outcomes in patients with or without baseline CNS metastases were analyzed. The ALEX-like RWD cohort (filtered by ALEX laboratory eligibility criteria) was used to compare real-world comparative effectiveness with ALEX. Results: The RWD cohort comprised 364 patients (141 alectinib; 223 crizotinib); rwPFS (weighted hazard ratio [wHR] = 0.46, 95% confidence interval [CI]: 0.33-0.65) and real-world overall survival (wHR = 0.46, 95% CI: 0.31-0.69) were significantly improved with alectinib versus crizotinib. In patients with baseline brain scans, a substantial rwPFS benefit was found regardless of baseline CNS metastases. Real-world time to new CNS metastases was delayed with alectinib versus crizotinib in patients with (wHR = 0.28, 95% CI: 0.16-0.52) and without (wHR = 0.42, 95% CI: 0.24-0.76) baseline CNS metastases. The ALEX-like RWD cohort comprised 325 patients (120 alectinib; 205 crizotinib); alectinib was found to have similar rwPFS benefits with ALEX. Conclusions: Outcomes were significantly improved with first-line alectinib versus crizotinib in patients with advanced ALK-positive NSCLC in the U.S. real-world setting.

Systematic pan-cancer analysis of mutation-treatment interactions using large real-world clinicogenomics data.

Quantifying the effectiveness of different cancer therapies in patients with specific tumor mutations is critical for improving patient outcomes and advancing precision medicine. Here we perform a large-scale computational analysis of 40,903 US patients with cancer who have detailed mutation profiles, treatment sequences and outcomes derived from electronic health records. We systematically identify 458 mutations that predict the survival of patients on specific immunotherapies, chemotherapy agents or targeted therapies across eight common cancer types. We further characterize mutation-mutation interactions that impact the outcomes of targeted therapies. This work demonstrates how computational analysis of large real-world data generates insights, hypotheses and resources to enable precision oncology.

Replication of Overall Survival, Progression-Free Survival, and Overall Response in Chemotherapy Arms of Non-Small Cell Lung Cancer Trials Using Real-World Data.

PURPOSE: The utility of real-world data (RWD) for use as external controls in drug development is informed by studies that replicate trial control arms for different endpoints. The purpose of this study was to replicate control arms from four non-small cell lung cancer (NSCLC) randomized controlled trials (RCT) to analyze overall survival (OS), progression-free survival (PFS), and overall response rate (ORR) using RWD. PATIENTS AND METHODS: This study used RWD from a nationwide de-identified database and a clinico-genomic database to replicate OS, PFS, and ORR endpoints in the chemotherapy control arms of four first-line NSCLC RCTs evaluating atezolizumab [IMpower150-wild-type (WT), IMpower130-WT, IMpower131, and IMpower132]. Additional objectives were to develop a definition of real-world PFS (rwPFS) and to evaluate the real-world response rate (rwRR) endpoint. RESULTS: Baseline demographic and clinical characteristics were balanced after application of propensity score weighting methods. For rwPFS and OS, RWD external controls were generally similar to their RCT control counterparts. Across all four trials, the hazard ratio (HR) point estimates comparing trial controls with external controls were closer to 1.0 for the PFS endpoint than for the OS endpoint. An exploratory assessment of rwRR in RWD revealed a slight but nonsignificant overestimation of RCT ORR, which was unconfounded by baseline characteristics. CONCLUSIONS: RWD can be used to reasonably replicate the OS and PFS of chemotherapy control arms of first-line NSCLC RCTs. Additional studies can provide greater insight into the utility of RWD in drug development.

Machine Learning Prediction of Clinical Trial Operational Efficiency.

Clinical trials are the gatekeepers and bottlenecks of progress in medicine. In recent years, they have become increasingly complex and expensive, driven by a growing number of stakeholders requiring more endpoints, more diverse patient populations, and a stringent regulatory environment. Trial designers have historically relied on investigator expertise and legacy norms established within sponsor companies to improve operational efficiency while achieving study goals. As such, data-driven forecasts of operational metrics can be a useful resource for trial design and planning. We develop a machine learning model to predict clinical trial operational efficiency using a novel dataset from Roche containing over 2,000 clinical trials across 20 years and multiple disease areas. The data includes important operational metrics related to patient recruitment and trial duration, as well as a variety of trial features such as the number of procedures, eligibility criteria, and endpoints. Our results demonstrate that operational efficiency can be predicted robustly using trial features, which can provide useful insights to trial designers on the potential impact of their decisions on patient recruitment success and trial duration.

Evaluating eligibility criteria of oncology trials using real-world data and AI.

There is a growing focus on making clinical trials more inclusive but the design of trial eligibility criteria remains challenging1-3. Here we systematically evaluate the effect of different eligibility criteria on cancer trial populations and outcomes with real-world data using the computational framework of Trial Pathfinder. We apply Trial Pathfinder to emulate completed trials of advanced non-small-cell lung cancer using data from a nationwide database of electronic health records comprising 61,094 patients with advanced non-small-cell lung cancer. Our analyses reveal that many common criteria, including exclusions based on several laboratory values, had a minimal effect on the trial hazard ratios. When we used a data-driven approach to broaden restrictive criteria, the pool of eligible patients more than doubled on average and the hazard ratio of the overall survival decreased by an average of 0.05. This suggests that many patients who were not eligible under the original trial criteria could potentially benefit from the treatments. We further support our findings through analyses of other types of cancer and patient-safety data from diverse clinical trials. Our data-driven methodology for evaluating eligibility criteria can facilitate the design of more-inclusive trials while maintaining safeguards for patient safety.

Prospective phase II trial of levonorgestrel intrauterine device: nonsurgical approach for complex atypical hyperplasia and early-stage endometrial cancer.

BACKGROUND: The incidence of complex atypical hyperplasia and early-stage endometrioid endometrial cancer is increasing, in part owing to the epidemic of obesity, which is a risk factor tightly linked to the development of endometrial hyperplasia and cancer. The standard upfront treatment for complex atypical hyperplasia and early-stage endometrial cancer is hysterectomy. However, nonsurgical treatment of early-stage endometrial neoplasia may be necessary owing to medical comorbidities precluding surgery or desired future fertility. OBJECTIVE: This study aimed to evaluate the efficacy of the levonorgestrel intrauterine device to treat complex atypical hyperplasia and grade 1 endometrioid endometrial carcinoma. STUDY DESIGN: A single-institution, single-arm, phase II study of the levonorgestrel intrauterine device (52 mg levonorgestrel, Mirena) was conducted in patients with complex atypical hyperplasia or grade 1 endometrioid endometrial cancer. The primary endpoint was pathologic response rate at 12 months, including complete or partial response. Quality of life and toxicity were assessed. Molecular analyses for proliferation markers, hormone-regulated genes, and wingless-related integration site pathway activation were performed at baseline and 3 months. RESULTS: A total of 57 patients were treated (21 endometrial cancer, 36 complex atypical hyperplasia). The median age was 48.0 years, and the median body mass index was 45.5 kg/m2. Of the 47 evaluable patients, 12-month response rate was 83% (90% credible interval, 72.7-90.3)-37 were complete responders (8 endometrial cancer; 29 complex atypical hyperplasia), 2 were partial responders (2 endometrial cancer), 3 had stable disease (2 endometrial cancer; 1 complex atypical hyperplasia), and 5 had progressive disease (3 endometrial cancer; 2 complex atypical hyperplasia). After stratification for histology, the response rate was 90.6% for complex atypical hyperplasia and 66.7% for grade 1 endometrioid endometrial cancer. Notably, 4 patients (9.5%) experienced relapse after the initial response. Adverse events were mild, primarily irregular bleeding and cramping. Quality of life was not negatively affected. At 3 months, exogenous progesterone effect was present in 96.9% of responders (31 of 32) vs 25% of nonresponders (2 of 8) (P=.001). Nonresponders had higher baseline proliferation (Ki67) and lower dickkopf homolog 3 gene expression than responders (P=.023 and P=.030). Nonresponders had significantly different changes in secreted frizzled-related protein 1, frizzled class receptor 8, and retinaldehyde dehydrogenase 2 compared with responders. CONCLUSION: The levonorgestrel intrauterine device has a substantial activity in complex atypical hyperplasia and grade 1 endometrioid endometrial cancer, with a modest proportion demonstrating upfront progesterone resistance. Potential biomarkers were identified that may correlate with resistance to therapy; further exploration is warranted.

Treatment of Low-Risk Endometrial Cancer and Complex Atypical Hyperplasia With the Levonorgestrel-Releasing Intrauterine Device.

OBJECTIVE: To assess efficacy of the levonorgestrel-releasing intrauterine device (LNG-IUD) for treatment of complex atypical hyperplasia or low-grade endometrial cancer. METHODS: This retrospective case series included all patients treated with the LNG-IUD for complex atypical hyperplasia or early-grade endometrial cancer from January 2003 to June 2013. Response rates were calculated and the association of response with clinicopathologic factors, including age, body mass index, and uterine size, was determined. RESULTS: Forty-six patients diagnosed with complex atypical hyperplasia or early-grade endometrial cancer were treated with the LNG-IUD. Of 32 evaluable patients at the 6-month time point, 15 had complex atypical hyperplasia (47%), nine had G1 endometrial cancer (28%), and eight had grade 2 endometrial cancer (25%). Overall response rate was 75% (95% CI 57-89) at 6 months; 80% (95% CI 52-96) in complex atypical hyperplasia, 67% (95% CI 30-93) in grade 1 endometrial cancer, and 75% (CI 35-97) in grade 2 endometrial cancer. Of the clinicopathologic features evaluated, there was a trend toward the association of lack of exogenous progesterone effect in the pathology specimen with nonresponse to the IUD (P=.05). Median uterine diameter was 1.3 cm larger in women who did not respond to the IUD (P=.04). CONCLUSION: Levonorgestrel-releasing IUD therapy for the conservative treatment of complex atypical hyperplasia or early-grade endometrial cancer resulted in return to normal histology in a majority of patients.

Factors prognostic of survival in advanced-stage uterine serous carcinoma.

OBJECTIVES: The study objective was to analyze the impact of prognostic factors, including treatment modality, on outcome in patients with advanced-stage uterine serous carcinoma (USC). METHODS: A retrospective review of patients diagnosed with stage III or IV USC between 1993 and 2012 was performed. Summary statistics were used to describe demographic and clinical characteristics. Overall survival (OS) and recurrence free survival (RFS) were estimated by Kaplan-Meier analysis. Cox proportional hazards regression was used to model the association of potential prognostic factors with OS and RFS. RESULTS: The study included 260 patients with median follow-up of 26.6months (range 1-172.8). Median age was 63years (range 30-88) and 52.3% had stage III disease. In all, 60% were treated with surgery followed by chemotherapy, 18.1% received surgery, chemotherapy, and radiotherapy, 11.5% had surgery and radiotherapy, and 10.4% had neoadjuvant chemotherapy. The overall complete response rate was 68.9%, and the cumulative incidence of recurrence was 82.7%. Treatment that included surgery, chemotherapy, and radiation and stage III disease were associated with improved RFS on multivariate analysis. For OS, therapy with surgery, chemotherapy, and radiation, mixed histology, and stage III disease were associated with better OS on multivariate analysis. CONCLUSIONS: Patients with advanced-stage USC have a poor prognosis, regardless of clinical factors or treatment received. However, combination therapy that includes chemotherapy and radiation appears to be associated with improved survival in these women.

PTEN loss is a context-dependent outcome determinant in obese and non-obese endometrioid endometrial cancer patients.

Endometrial cancer incidence is increasing, due in part to a strong association with obesity. Mutations in the phosphatidylinositol 3-kinase (PI3K) pathway, the central relay pathway of insulin signals, occur in the majority of endometrioid adenocarcinomas, the most common form of endometrial cancer. We sought to determine the impact of PI3K pathway alterations on progression free survival in a cohort of endometrioid endometrial cancers. Prognostic utility of PIK3CA, PIK3R1, and PTEN mutations, as well as PTEN protein loss by immunohistochemistry, was explored in the context of patient body mass index. Reverse-phase protein arrays were utilized to assess protein expression based on PTEN status. Among 187 endometrioid endometrial cancers, there were no statistically significant associations between PFS and PIK3CA, PIK3R1, PTEN mutation or loss. When stratified by body mass index, PTEN loss was associated with improved progression free survival (P < 0.006) in obese (body mass index ≥ 30) patients. PTEN loss resulted in distinct protein changes: Canonical PI3K pathway activation was observed only in the non-obese population while decreased expression of ß-CATENIN and phosphorylated FOXO3A was observed in obese patients. These data suggest the impact of PTEN loss on tumor biology and clinical outcomes must be interpreted in the context of body mass index, and provide a potential explanation for discrepant reports on the effect of PTEN status and obesity on prognosis in endometrial cancer. This reveals a clinically important interaction between metabolic state and tumor genetics that may unveil the biologic underpinning of obesity-related cancers and impact ongoing clinical trials with PI3K pathway inhibitors.

Overall survival after pelvic exenteration for gynecologic malignancy.

BACKGROUND: Five-Year survival after pelvic exenteration for gynecologic malignancies has been reported as high as 60%. The objective of this study was to determine overall survival (OS) after pelvic exenteration and evaluate factors impacting outcome. METHODS: A retrospective review of all women who underwent pelvic exenteration at our institution between February 1993 and December 2010 was performed. OS was defined as time from exenteration to date of death or last contact. Survival analysis was performed using the Kaplan Meyer method. Multivariate analysis was performed to determine the impact of clinical and pathologic factors on survival outcomes. RESULTS: One hundred sixty patients with gynecologic malignancy underwent pelvic exenteration. Five-year recurrence free survival (RFS) was 33% (95%CI 0.25-0.40). Factors which negatively impacted RFS included shorter treatment-free interval (p=.050), vulvar primary (p=.032), positive margins (p<.001), lymphovascular space invasion (LVSI, p<.001), positive lymph nodes (p<.001) and perineural invasion (p=0.030). In multivariate analysis, positive margins (p=.040), positive nodes (p<.001) and lymphovascular space invasion (LVSI, p=.003) retained a significant impact on RFS. Five-year OS was 40% (95% CI 0.32-0.48). Factors which negatively impacted OS included vulvar primary (p=.04), positive margins (p<.001), LVSI (p<.001), positive lymph nodes (p<.001) and perineural invasion (p=.008). In multivariate analysis, positive nodes (p=.001) and LVSI (p=.001) retained a significant impact on OS. CONCLUSION: Five-year OS after pelvic exenteration was 40%. Survival outcomes have not significantly improved despite improvements in technique and patient selection. Multiple non-modifiable factors at the time of exenteration are associated with poor survival.