Project Data Sphere Research Articles

Organ-specific tumor dynamics predict survival of patients with metastatic colorectal cancer

BackgroundWe aim to compare the prognostic value of organ-specific dynamics with the sum of the longest diameter (SLD) dynamics in patients with metastatic colorectal cancer (mCRC). MethodsAll datasets are accessible in Project Data Sphere, an open-access platform. The tumor growth inhibition models developed based on organ-level SLD and SLD were used to estimate the organ-specific tumor growth rates (KGs) and SLD KG. The early tumor shrinkage (ETS) from baseline to the first measurement after treatment was also evaluated. The relationship between organ-specific dynamics, SLD dynamics, and survival outcomes (overall survival, OS; progression-free survival, PFS) was quantified using Kaplan-Meier analysis and Cox regression. ResultsThis study included 3687 patients from 6 phase III mCRC trials. The liver emerged as the most frequent metastatic site (2901, 78.7 %), with variable KGs across different organs in individual patients (liver 0.0243 > lung 0.0202 > lymph node 0.0127 > other 0.0118 [week−1]). Notably, the dynamics for different organs did not equally contribute to predicting survival outcomes. In liver metastasis cases, liver KG proved to be a superior prognostic indicator for OS and surpasses the predictive performance of SLD, (C-index, liver KG 0.610 vs SLD KG 0.606). A similar result can be found for PFS. Moreover, liver ETS also outperforms SLD ETS in predicting survival. Cox regression analysis confirmed liver KG is the most significant variable in survival prediction. ConclusionsIn mCRC patients with liver metastasis, liver dynamics is the primary prognostic indicator for both PFS and OS. In future drug development for mCRC, greater emphasis should be directed towards understanding the dynamics of liver metastasis development.

Improved estimation of overall survival and progression-free survival for state transition modeling.

Aim: National Institute for Health and Care Excellence guidance (Technical Support Document 19) highlights a key challenge of state transition models (STMs) being their difficulty in achieving a satisfactory fit to the observed within-trial endpoints. Fitting poorly to data over the trial period can then have implications for long-term extrapolations. A novel estimation approach is defined in which the predicted overall survival (OS) and progression-free survival (PFS) extrapolations from anSTM are optimized to provide closer estimates of the within-trial endpoints. Materials & methods: An STM was fitted to the SQUIRE trial data in non-small-cell lung cancer (obtained from Project Data Sphere). Two methods were used: a standard approach whereby the maximum likelihood was utilized for the individual transitions and the best-fitting parametric model selected based on AIC/BIC, and a novel approach in which parameters were optimized by minimizing the area between the STM-predicted OS and PFS curves and the corresponding OS and PFS Kaplan-Meier curves. Sensitivity analyses were conducted to assess uncertainty. Results: The novel approach resulted in closer estimations to the OS and PFS Kaplan-Meier for all combinations of parametric distributions analyzed compared with the standard approach. Though the uncertainty associated with the novel approach was slightly larger, it provided better estimates to the restricted mean survival time in 10 of the 12 parametric distributions analyzed. Conclusion: A novel approach is defined which provides an alternative STM estimation method enabling improved fits to modeled endpoints, which can easily be extended to more complex model structures.

Impact of body mass index and its change on survival outcomes in patients with early breast cancer: A pooled analysis of individual-level data from BCIRG-001 and BCIRG-005 trials.

The relationships between body mass index (BMI) and survival rates are complex, and have not been thoroughly investigated in breast cancer patients who received adjuvant chemotherapy. We collected data on 2394 patients from two randomized, phase III clinical trials that investigated adjuvant chemotherapy in breast cancer identified in Project Data Sphere. The objective was to examine the effect of baseline BMI, BMI after adjuvant chemotherapy, and BMI change from baseline to post-adjuvant chemotherapy on disease-free survival (DFS) and overall survival (OS). Restricted cubic splines were used to examine potential non-linear associations between continuous BMI value and survival. Stratified analyses involved chemotherapy regimens. Severe obesity (BMI≥40.0kg/m2) at baseline was independently associated with worse DFS (hazard ration [HR]=1.48, 95% confidence interval [CI] 1.02-2.16, P=0.04) and OS (HR=1.79, 95%CI 1.17-2.74, P=0.007) compared with underweight/normal weight (BMI≤24.9kg/m2). A BMI loss >10% was also an independent prognostic factor for adverse OS (HR=2.14, 95%CI 1.17-3.93, P=0.014). Stratified analyses revealed that severe obesity adversely affected DFS (HR=2.38, 95%CI 1.26-4.34, P=0.007) and OS (HR=2.90, 95%CI 1.46-5.76, P=0.002) in the docetaxel-based group, but not in the non-docetaxel-based group. Restricted cubic splines revealed a "J-shaped" association of baseline BMI with risk of recurrence or all-cause death, and this relationship was more pronounced in the docetaxel-based group. In early breast cancer patients treated with adjuvant chemotherapy, baseline severe obesity was significantly linked to worse DFS and OS, and a BMI loss over 10% from baseline to post-adjuvant chemotherapy also negatively affected OS. Moreover, the prognostic role of BMI might differ between docetaxel-based and non-docetaxel-based groups.

Effectiveness of Cisplatin in P16+ Oropharyngeal Cancer According to Relative Risk for Cancer Events: Ancillary Analysis of RTOG 1016

To test the hypothesis that the effectiveness of cisplatin in p16+ oropharyngeal squamous cell carcinoma (OPSCC) increases with patients' relative risk for cancer events. Ancillary analysis of 805 patients enrolled on RTOG 1016 accessed via Project DataSphere. Eligible patients had p16+ OPSCC, AJCC 7th T1-T2 N2a-N3 or T3-T4 N0-N3 M0, ECOG PS 0-1. Patients were randomized to RT with concurrent cisplatin vs. cetuximab. Relative risk for competing events was quantified using the Head and Neck Cancer Intergroup predictive classifier (omega score). Higher scores indicate higher relative risk for cancer events (LRF or distant metastasis) vs. competing mortality. We compared this to favorable, unfavorable/low, and unfavorable/intermediate risk groups using standard criteria: NRG HN005 eligible/low RTOG risk (Ang et al.), HN005 ineligible/low RTOG risk, and intermediate RTOG risk. Omega score cutoffs were selected to match numbers in standard risk strata. HRs for the effect of cisplatin vs. cetuximab on PFS and OS were compared for standard vs. relative risk strata. 1-tailed interaction tests were used to test whether cisplatin effectiveness increased within risk strata. There were 354, 219, and 232 patients in standard favorable, unfavorable/low, and unfavorable/intermediate risk groups. Omega score cutoffs were 0.80 and 0.84 to define low, intermediate, and high relative risk groups. Discordant standard vs. relative risk classifications occurred in 559 patients (69.4%). Increasing omega score was associated with significantly higher relative HR (rHR) for cancer events (3.40, 95% CI: 1.66-6.96) and increasing effectiveness of cisplatin vs. cetuximab (Table), but standard risk grouping was not (rHR 0.80, 95% CI: 0.49-1.32). The effect of cisplatin on PFS significantly increased with higher omega score (interaction -0.30, p = .046), but decreased with increasing standard risk strata (interaction +0.27, p = NS). The effectiveness of cisplatin in p16+ OPSCC increased with higher omega score but not with standard risk group. Relative risk for cancer events should be taken into account when designing deintensification strategies.

Explainable machine learning to inform real-world evidence: A case study on small cell lung cancer survival analysis.

113 Background: In recent years, machine learning algorithms for survival analysis have been developed as an alternative to traditional Cox regression. Despite the good performances, these are viewed as black box models that lack interpretability, which may limit their clinical applications. There has been a growing interest in explaining and interpreting machine learning models, which has led to the emergence of the field of explainable artificial intelligence (XAI). However, few studies have focused on how these methods could be of benefit to survival analysis of real-world healthcare data. Methods: In this study, we explored the use of current XAI techniques to analyze the effects of platinum doublet chemotherapy on small cell lung cancer patients in developed machine learning models.The analyzed data included real-world patients treated at Karolinska University Hospital (n=570) and three phase III randomized clinical trials shared by the Project Data Sphere initiative (n=987). The real-world data covariates included were age, sex, TNM staging, ECOG performance status, lab values, brain metastasis, and concomitant radiotherapy. The aggregated dataset including the clinical trials contained the following variables: age, sex, performance status, brain metastasis, and protocol violations. Eight machine learning models were trained and compared with Cox regression. The performance of the models was evaluated using C-index, and the time variation of Brier Score and C/D AUC.Temporal feature importance and partial dependence were used to explore the overall covariate impact on overall survival (Global XAI). Ceteris-paribus and SurvShap(t) were used to investigate the covariate impact for the single patient prediction (local XAI). The models were firstly trained only on real-world data before aggregation with the clinical trials. Results: Ensemble machine learning provided the best performances. XAI techniques showed the potential to increase explainability of survival predictions in function of time. Global XAI showed the time range of the model reliability, trend inversions regarding treatment decisions and covariate importance along the time. Local XAI allowed to test the impact of covariates between long survival patients and the comparison between real-world and clinical trials. Conclusions: Our results demonstrate the potential of XAI techniques applied to survival machine learning and real-world data, thus providing insights into the mechanisms driving model predictions and demonstrate the utility of this approach in clinical research.

BioDigit MG: A Pilot Study Examining a Digital Technology for Quantitative Assessment and Remote Monitoring of Myasthenia Gravis (P7-8.005)

<h3>Objective:</h3> To develop a standardized, clinically-validated digital solution for rigorous, longitudinal assessment of ocular weakness and dysarthria in myasthenia gravis (MG), which can be applied remotely. <h3>Background:</h3> MG is a chronic autoimmune neuromuscular disease characterized by fluctuating weakness of skeletal muscle, which frequent affects ocular muscles and speech. Current assessments of myasthenic weakness require neurologic expertise, are time-consuming and can still be subjective. Additionally, measured at one point in time, assessments provide a snapshot of a patient’s MG, which may not adequately reflect the spectrum of fluctuating weakness. Ratings of ocular weakness and dysarthria are particularly subjective. These shortcomings of current outcome measures limit performance of clinical trials and provide a limited view of the patient’s clinical status. <h3>Design/Methods:</h3> We developed a tablet-based digital health solution that captures video data collected by patients and MG-specific patient-reported outcome measures. Ptosis and dysarthria are rated asynchronously. The tablet provides text instruction and audio prompts to the patient and data are uploaded to a cloud. Predetermined measures of success have been defined. <h3>Results:</h3> A pilot study with a target sample size of 20 individuals with MG is ongoing to determine the feasibility of using the proposed solution as a health monitoring platform for MG. Two MG experts will rate speech and video data for dysarthria and ptosis based on the Quantitative Myasthenia Gravis (QMG) scale. Association between the extracted speech and facial characteristics to clinical scores of dysarthria and ptosis will be investigated. <h3>Conclusions:</h3> Despite considerable recent advances in treatments for MG, rigorous, longitudinal tools for monitoring of its characteristic fluctuating weakness remain limited and represent an unmet need. The feasibility, validity and acceptability of the proposed digital solution will be determined for 20 patients with MG. If successful, this technology will be further developed and studied in a larger observational study. <b>Disclosure:</b> An immediate family member of Dr. Guidon has received personal compensation for serving as an employee of GE Healthcare. Dr. Guidon has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Alexion. Dr. Guidon has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Momenta. Dr. Guidon has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Argenx. The institution of Dr. Guidon has received research support from MGFA. The institution of Dr. Guidon has received research support from Project Data Sphere. The institution of Dr. Guidon has received research support from MGNet. The institution of Dr. Guidon has received research support from MGNet. Dr. Guidon has received publishing royalties from a publication relating to health care. Dr. Mishra has received personal compensation for serving as an employee of Biosensics. Miss Lindsay has nothing to disclose. Mr. Mulukutla has received personal compensation for serving as an employee of BioSensics LLC. Dr. Vaziri has received personal compensation for serving as an employee of Biosensics.

Application of a Validated Algorithm to Identify and Describe Contemporary Epidemiology of Privately Insured Myasthenia Gravis Patients in the United States (N6.002)

<h3>Objective:</h3> 1) To derive a cohort of patients with myasthenia gravis (MG) from a large United States (US) claims database using an algorithm previously validated in a Medicare population; 2) To describe epidemiology of this patient cohort. <h3>Background:</h3> Epidemiologic studies of MG using reliable case-ascertainment are lacking. A previously validated algorithm identified “definite” MG cases with sensitivity and positive predictive value of 80% among Medicare patients. To our knowledge, it has not been applied to other datasets. Analysis of contemporary care utilization for MG depends on accurate patient identification. <h3>Design/Methods:</h3> We performed a retrospective cohort study using years 2017–20 of Truven Marketscan, an administrative claims database. Adult patients with autoimmune MG were identified by presence of at least 2 MG-related ICD10 codes (G70.0, G70.00 and G70.01) associated with outpatient visits, separated by ≥ 4 weeks; OR ≥ 1 hospitalization and 1 outpatient visit using MG-related ICD codes, separated by ≥ 4 weeks, within 2 years. Exclusion criteria included other neuromuscular junction disorders. Annual incidence was calculated for 2019–20, using a 2 year wash-out period. We derived a non-MG comparison group from a random 1% sample of the total population. <h3>Results:</h3> Among 47,803,384 patients, we identified 12,121 (0.025%) with MG. MG patients were more likely female (56% vs 51%), age 40–64 (58% vs 46%) and with history of thymoma (2% vs 0.01%). Like the non-MG group, at least 11% lived in rural areas, but a lower proportion of MG patients lived in Western states (12% vs 17%). Total of 1,556/21,935,502 continuously enrolled patients had incident diagnosis of MG 2019–20 (0.007%). <h3>Conclusions:</h3> An algorithm validated in Medicare patients for MG identification was applied to a private insurer claims database. Incidence and prevalence of MG were calculated and are comparable to previous estimates. The cohort’s healthcare utilization data will be presented. <b>Disclosure:</b> Miss Hua has nothing to disclose. Dr. Semenov has received personal compensation in the range of $5,000-$9,999 for serving as a Consultant for Regeneron. Dr. Semenov has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Galderma. Dr. Semenov has received personal compensation in the range of $500-$4,999 for serving as an Editor, Associate Editor, or Editorial Advisory Board Member for Journal of the American Academy of Dermatology. The institution of Dr. Semenov has received research support from Dermatology Foundation. An immediate family member of Dr. Guidon has received personal compensation for serving as an employee of GE Healthcare. Dr. Guidon has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Alexion. Dr. Guidon has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Momenta. Dr. Guidon has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Argenx. The institution of Dr. Guidon has received research support from MGFA. The institution of Dr. Guidon has received research support from Project Data Sphere. The institution of Dr. Guidon has received research support from MGNet. The institution of Dr. Guidon has received research support from MGNet. Dr. Guidon has received publishing royalties from a publication relating to health care.

Abstract 6204: Age-related disparity in treatment outcome and adverse events among patients with metastatic colorectal cancer (mCRC)

Abstract Background: The number of newly diagnosed mCRC has been increasing in not only the aging population but also those &amp;lt;50 years of age. However, age-related disparity in the management of patients with mCRC has not been adequately evaluated. In this study, we aimed to compare and contrast the treatment outcome and adverse events among different age groups of patients with mCRC. Methods: We used individual patient data of a clinical trial (NCT00364013) from Project Data Sphere with patients who received first-line FOLFOX or FOLFOX + panitumumab for mCRC. Patients in each treatment arm were grouped into &amp;lt;50 years, 50-65 years, and &amp;gt;65 years of age groups. Continuous and categorical variables were compared with t test and Fisher’s exact test, respectively. Survival was summarized using Kaplan-Meier method and compared using log-rank test. Cox proportional hazards model was used to determine the association of adverse events with survival. Results: Among 935 patients included, 112 (12.0%) were in the &amp;lt;50 group; 456 (48.8%) were in the 50-65 group; 367 (39.2%) were in the &amp;gt;65 group. All three groups had similar baseline characteristics. Median overall survival (OS) of both &amp;lt;50 (18.4, 95%CI 13.1-23.1 months) and &amp;gt;65 (18.0, 95%CI 16.3-22.2 months) groups were significantly shorter (p=0.008) than median OS of the 50-65 group (22.7, 95%CI 20.0-26.5 months) in the FOLFOX arm. Similar findings were present in the FOLFOX + panitumumab arm (p=0.049). Both &amp;lt;50 and &amp;gt;65 groups had a trend (p=0.09) toward shorter progression-free survival (PFS) compared to the 50-65 group. As to adverse events, in the FOLFOX arm, all grade nausea/vomiting of the &amp;lt;50 group was more common than that in the &amp;gt;50 group (70.3% vs 55.8%, p=0.05). Compared to the &amp;gt;50 group, the &amp;lt;50 group had earlier onset of neutropenia at 7.8 weeks (p=0.003) and chest pain at 5.4 weeks (p=0.008). In the FOLFOX + panitumumab arm, all grade fatigue (58.3% vs 35.7%, p=0.003), neuropathy (52.1% vs 34.7%, p=0.03) and mucositis (43.8% vs 29.5%, p=0.049) of the &amp;lt;50 group were more common than those in the &amp;gt;50 group. The duration of various adverse events were similar among the age groups. In the FOLFOX arm, severe (grade≥3) liver toxicity was associated with worse PFS (HR 3.6, 95%CI 1.3-9.8, p=0.01) and OS (HR 7.3, 95%CI 2.7-20.1, p&amp;lt;0.001). In contrast, severe neuropathy was associated with better PFS (HR 0.6, 95%CI 0.4-0.9, p=0.009) and OS (HR 0.5, 95% 0.3-0.8, p=0.001). In the FOLFOX + panitumumab arm, severe anemia was associated with worse PFS (HR 1.7, 95%CI 1.02-2.8, p=0.04) and OS (HR 2.8, 95%CI 1.7-4.7, p&amp;lt;0.001). In contrast, severe rash was associated with better PFS (HR 0.5, 95%CI 0.3-0.8, p=0.002) and OS (HR 0.3, 95% 0.2-0.5, p&amp;lt;0.001). Conclusions: Patients with early-onset mCRC had not only worse OS in both treatment arms but also higher incidence of adverse events. Our management approach should adapt to the unique features of this patient population. Citation Format: Lingbin Meng, Ram Thapa, Hao Xie. Age-related disparity in treatment outcome and adverse events among patients with metastatic colorectal cancer (mCRC) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 6204.

Abstract 5255: Tumor heterogeneity in response to therapy and its association with outcome in patients with metastatic colorectal cancer (mCRC)

Abstract Background: Clinical benefit of therapies matched with genetic profiling remains limited in patients with mCRC, which is partially attributed to tumor heterogeneity. Compared to molecular tumor heterogeneity using modern multi-omics approach, tumor heterogeneity in response to therapy has not been adequately studied. We aimed to characterize such tumor heterogeneity and evaluate its association with outcome in patients with mCRC. Methods: We used individual patient data from a clinical trial (NCT00364013) in Project Data Sphere who received first-line FOLFOX for mCRC and had ≥2 baseline metastatic target lesions. We used standard deviations of following tumor features in an individual patient to represent tumor heterogeneity: percentage change in size of a target lesion from baseline to first radiographic evaluation (H1), average speed of such percentage change in size as defined in H1 (H2), percentage change in size of a target lesion from baseline to the time of its best response to therapy (H3), and average speed of such percentage change in size as defined in H3 (H4). Cox proportional hazards models were used to evaluate association of tumor heterogeneity with survival. Results: Mean age of 328 patients included was 61 years. Median follow-up time was 18.2 months. 85% of them had liver metastasis; 15% had lung metastasis. All tumor heterogeneity variables in response to therapy (H1-4) were higher (p&amp;lt;0.001) in patients with disease progression compared to those without. In univariate analyses, higher tumor heterogeneity H1 (HR 7.2, 95% CI 2.3-22.6), H2 (HR 3.1, 95% CI 1.7-5.7), H3 (HR 6.4, 95% CI 2.1-19.3), and H4 (HR 4.1, 95% CI 2.4-7.1) were associated with shorter progression-free survival (PFS) (p&amp;lt;0.001). H4 high group (median as cutoff) had a shorter median PFS of 7.3 months compared to H4 low group (9.1 months, p=0.005). In multivariable analysis, higher H1 (HR 8.0, 95% CI 2.2-28.9, p=0.002), H2 (HR 3.9, 95% CI 2.0-7.6, p&amp;lt;0.001), H3 (HR 6.4, 95% CI 1.9-21.9, p=0.003), or H4 (HR 5.1, 95% CI 2.9-9.1, p&amp;lt;0.001) was each associated with a shorter PFS, after adjusted for baseline number of metastatic sites, number of target lesions, tumor size, age, albumin, ECOG, and lymph node metastasis. Similar findings were found when ratio of maximal change to baseline sum of tumor size due to therapy was adjusted for and when H1-4 were calculated from liver lesions only. Higher H1 (HR 3.6, 95% CI 1.2-11.2, p=0.03), H2 (HR 1.9, 95% CI 1.1-3.4, p=0.02), H3 (HR 5.2, 95% CI 1.8-14.8, p=0.002), or H4 (HR 2.5, 95% CI 1.5-4.0, p&amp;lt;0.001) was also each associated with a shorter overall survival in multivariable analysis. Conclusion: Higher tumor heterogeneity in response to chemotherapy within an individual patient was independently associated with early disease progression and shorter survival. It could provide prognostic value complementary to existing tumor heterogeneity variables in patients with mCRC. Citation Format: Junjia Liu, Hao Xie. Tumor heterogeneity in response to therapy and its association with outcome in patients with metastatic colorectal cancer (mCRC) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5255.

Abstract 3789: Mapping lesion specific response and relapse patterns in metastatic colorectal cancer patients

Abstract Considerable lesion-specific response heterogeneity exists in metastatic colorectal cancer patients, largely due to organ-specific ecological environments and evolutionary pressures. Metastatic lesions with poor response to therapy often become tumor sanctuary sites, leading to systemic resistance and tumor relapse. To map the lesion-specific response and relapse patterns, we investigated the longitudinal dynamics of individual lesions in metastatic colorectal cancer patients. Tumor longitudinal data in 4,308 colorectal cancer patients with 40,612 individual lesions were collected from eight Phase III trials in Project Data Sphere. First, tumor response dynamics (regression after treatment and progression upon resistance) were characterized using an empirical mathematical model. Next, tumor response time (when the lesion size decreases ≥20% from baseline) and relapse time (when the lesion size increases ≥30% from tumor nadir) were estimated for each individual lesion in patients being treated with bevacizumab, panitumumab, and/or chemotherapy. Random effect cox proportional models were applied to predict lesion-specific response and relapse probabilities and temporal sequence. We then took machine learning algorithm k-means to cluster patients based on their lesion relapse sequence. We found the response probabilities across organs are: Liver &amp;gt; Distal Lymph Nodes (LN) &amp;gt; Abdomen &amp;gt; Spleen &amp;gt; Lung &amp;gt; Regional LN &amp;gt; Adrenal &amp;gt; Muscle/Soft Tissue &amp;gt; Bone &amp;gt; Brain/CNS. Lesion relapse temporal sequence are: Brain/CNS &amp;gt; Liver &amp;gt; Adrenal &amp;gt; Muscle/Soft tissue &amp;gt; Abdomen &amp;gt; Bone &amp;gt; Spleen &amp;gt; Lung &amp;gt; Distal LN &amp;gt; Regional LN. Of note, lesions in the bone, brain, adrenal, and muscle/soft tissues often had low responses and high relapse probabilities, implying the greatest potential as tumor sanctuary sites. Liver, the most common metastatic organ in colorectal cancer, showed highest response rate but high relapse probabilities. Interestingly, the organ-specific response rate and relapse probabilities are respectively in line with drug distribution profiles and organ-specific immune landscape. Organ-specific relapse sequence in each patient is significantly correlated with patient long-term survival (p&amp;lt;0.0001). Patients with relapse lesions occurring in multiple organs had worst survival. Patients whose liver lesions relapsed first had worse survival than those who first relapsed in other organs. In conclusion, our study provides insights into the lesion-specific response and relapse heterogeneity in metastatic colorectal cancer, and yields substantial implications for designing lesion-specific therapeutics. Citation Format: Jiawei Zhou, Quefeng Li, Amber Cipriani, Yanguang Cao. Mapping lesion specific response and relapse patterns in metastatic colorectal cancer patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 3789.

Disparity of treatment-related adverse events and outcome in patients with early-onset metastatic colorectal cancer (mCRC).

6560 Background: While the incidence of newly diagnosed early-onset mCRC has been increasing, disparity of treatment-related adverse events (AE) and outcomes of this patient population has been inadequately studied with inconclusive findings. We aimed to evaluate such age-related disparity and explore potential underlying causes. Methods: We used individual patient data from 3 clinical trials in Project Data Sphere where 756 and 467 patients with mCRC received first-line FOLFOX in study 1 (NCT 00305188, NCT00272051) and study 2 (NCT00364013), respectively. Clinical NGS data of 763 patients with mCRC from prospectively maintained Moffitt Clinical Genomics Database were used to assess genomic alterations. Patients were categorized into 3 age groups: &lt;50, 50-65, and &gt;65 years. Continuous and categorical variables were compared with t test and χ2 test, respectively. Kaplan-Meier method and log-rank test were used for survival analysis. Benjamini-Hochberg procedure was used to adjust for multiple comparisons. Results: Among 1986 patients included, 341 (17.2%) in the &lt;50 group, 912 (45.9%) in the 50-65 group, 733 (36.9%) in the &gt;65 group had similar baseline characteristics. Outcomes: Patients in the &lt;50 group had shorter median OS compared to the 50-65 (15.5 vs 20.5 month, p=0.003) and &gt;65 groups (15.5 vs 20.8 months, p=0.004) and shorter median PFS compared to the 50-65 (8.1 vs 9.4 month, p=0.039) and &gt;65 groups (8.1 vs 8.6 months, p=0.07) in study 1. Findings were confirmed in study 2. Toxicity: Compared to other age groups, the &lt;50 group had a higher incidence of severe (grade≥3) abdominal pain (8.4% vs 3.4%, p=0.01), severe anemia (6.1% vs 1.2%, p=0.001), nausea/vomiting (69.3% vs 58.8%, p=0.02), but lower incidence of severe diarrhea (6.1% vs 10.8%, p=0.02), severe neutropenia (25.7% vs 44.2%, p&lt;0.001), and severe fatigue (4.5% vs 7.3%, p=0.02). The &lt;50 group had earlier onset of nausea/vomiting (1.0 vs 2.3 weeks, p=0.01), mucositis (3.6 vs 5.4 weeks, p=0.05), and neutropenia (8.0 vs 9.0 weeks, p=0.04), and shorter duration of mucositis (0.6 vs 0.9 weeks, p=0.006). Outcomes + toxicity: In the &lt;50 group, abdominal pain was associated with worse OS (HR 1.6, p=0.01). In contrast, fatigue (HR 0.66, p=0.03) and mucositis (HR 0.67, p=0.046) were associated with better OS. Neuropathy was associated with better PFS (HR 0.61, p=0.009). Genomics: Our NGS data showed that the &lt;50 group had more ERBB2 amplification (10.5% vs 4.0%, p=0.04) and a trend of more TP53 mutations (80.2% vs 72.0%, p=0.07). These findings were confirmed in an independent mCRC cohort (n=471, MSK, Gastroenterology 2020). Conclusions: Patients with early-onset mCRC had worse outcome and unique treatment-related AE patterns. Distinct genomic profiles could explain some of the disparities. Our findings might improve a personalized management approach in patient selection for chemotherapy, counseling and AE monitoring.

Is recent FDA guidance on HRQoL clinical significance a barrier to clinician/patient treatment risk/benefit discussions: An example using EORTC scale endpoints in a randomized, phase 3 trial.

e21066 Background: Recent FDA guidance on clinical significance thresholds (CSTs) may adversely affect the evaluation of EORTC endpoints, thereby altering clinician/patient discussions about treatment (Tx) risk/benefit for emerging Tx. Specifically, FDA now rejects historic CSTs for EORTC measures (e.g., Osoba, 1998 10-point change). FDA’s guidance implies that EORTC function/symptom CSTs should be ±33.331; in addition, Tx arms must separate at the CST. Few, if any, EORTC domains are likely to achieve CSTs 3 times those of Osoba (1998). Holding health-related quality of life (HRQoL) endpoints to this standard may limit detection of clinically meaningful patient-centric Tx benefits &amp; clinician/patients dialogue about HRQoL for emerging cancer Tx. The NCI-conducted Sunitinib trial data (NCT00693992), obtained through Project DataSphere, was used to generate revised deterioration thresholds for EORTC scales under an anchor-based meaningful change approach. This this Phase 3 trial met it’s primary PFS endpoint. Evidence from this research highlights the potential implications of the FDAs position on COA-based CSTs. Methods: The Sunitinib trial was a randomized, phase 3 study comparing the efficacy &amp; safety of Sunitinib versus placebo as a maintenance therapy in non-progressing patients following chemotherapy in advanced non-small cell lung cancer. Symptoms &amp; HRQoL were measured by EORTC LC13. The analysis population consisted of 107 patients in the HRQoL substudy completing the QLQ-C30 Global Health/QoL scale (GHS). The GHS was used to categorize patients into: no change (0 pt. change), deterioration (≤-1 pt. change) &amp; improvement (≥1 pt. change). Deterioration thresholds were estimated as the mean LC13 scale change-scores &amp; visualized with empirical cumulative distribution functions (eCDFs). Results: See Table. Within this population, interpretable mean deterioration CSTs were found Dyspnea, Diarrhea, and Nausea/Vomiting. At these CSTs, Tx arms separated in eCDFs except for Dyspnea. However, while Dyspnea &amp; Diarrhea CSTs exceeded Osoba’s CSTs, none met the FDA criterion of + 33.33 points change. Conclusions: The LC13 scores had interpretable CSTs; evidence suggests FDA’s CST criterion may be unachievable for EORTC scales &amp; may thereby affect HRQoL endpoints. This research was limited by the use of the GHS as an anchor variable rather than a PGIS. The main risk is that FDA’s CST criterion may limit the characterization of truly meaningful Tx benefits for patients &amp; clinician communication with patients about HRQoL for emerging Tx. Clinical trial information: NCT00693992. [Table: see text]

Impact of the Economic Status of the Patient's Country of Residence on the Outcome of Oncology Clinical Trials.

BackgroundProlongation of overall survival (OS) is commonly evaluated as a primary endpoint in confirmative oncology clinical trials; however, it is potentially affected by subsequent treatments carried out in practice. To design and implement multi-regional clinical trials properly, we compared survival outcomes between Organisation for Economic Co-operation and Development (OECD) and non-OECD countries.Materials and MethodsIndividual patient data from industry-sponsored multi-regional phase III oncology trials were obtained from the Project Data Sphere. Patients of each arm were divided into several subgroups based on race and country where patients were enrolled. We defined the member countries of the OECD. Cox regression analysis was conducted to estimate the hazard ratio (HR) for progression-free survival (PFS) and OS between the different subgroups in each trial, followed by a meta-analysis to estimate the summary HR and its confidence interval with a random-effect model.ResultsEleven arms from 10 clinical trials were eligible for the analysis. No statistically significant difference was observed in PFS and OS between Caucasian and Asian. A prolongation of OS was observed in patients enrolled in the OECD group compared with non-OECD group, while no statistically significant difference was observed in PFS.ConclusionThe economic status and healthcare environment of countries where patients reside have an impact on the outcome of OS. Clinical trial sponsors are recommended to consider carefully how to properly design oncology clinical trials including the selection of countries and data management of subsequent treatments.

Prognostic significance of baseline neutrophil-lymphocyte ratio in patients with non-small-cell lung cancer: a pooled analysis of open phase III clinical trial data.

Aim: This study aimed to assess survival and hematological prognostic indicators of patients with non-small-cell lung cancer (NSCLC). Material & methods: Through the Project Data Sphere portal, two phase III clinical trial datasets were downloaded to analyze survival outcomes and related risk factors. Results: The median progression-free survival and overall survival of 756 patients with stage III-IV NSCLC were 6.2 and 14.2months, respectively. In multivariate Cox analysis, high baseline neutrophil-lymphocyte ratio (NLR; ≥3.8) was associated with worse progression-free survival (hazard ratio: 1.37; p=0.0004) and overall survival (hazard ratio: 1.65; p<0.0001). In addition, it exerted an unfavorable impact on survival across multiple subgroups. Conclusions: NLR, a powerful inflammatory and immunologic indicator, is an independent prognostic indicator in patients with advanced NSCLC.

Carboplatin versus cisplatin in combination with etoposide in the first-line treatment of small cell lung cancer: a pooled analysis

BackgroundExtensive-stage small cell lung cancer (ES-SCLC) is an aggressive disease with poor survival, and platinum-etoposide chemotherapy is indicated as the mainstay of treatment. In this study, we compared the efficacy and safety between the cisplatin plus etoposide (EP) and carboplatin plus etoposide (EC) regimens.MethodsA total of 1305 patients with previously untreated ES-SCLC were included in this study. Data from five trials were collected from the public database Project Data Sphere. Survival analysis and adverse events (AEs) analysis were conducted.ResultsOf the 1305 patients, 800 received the EC regimen whereas 505 received the EP regimen as their front-line treatment. Overall, the median progression-free survival (PFS) and the median overall survival (OS) were 172 and 289 days, respectively. The EP and EC treatment groups did not have significantly different PFS or OS. After adjusting for age, sex, body mass index (BMI) and Eastern Cooperative Oncology Group (ECOG) performance status (PS), the EP regimen was independently associated with better PFS (hazard ratio [HR] = 0.76, 95% CI = 0.63–0.92, p = 0.0041) and OS (HR = 0.79, 95% CI = 0.64–0.97, p = 0.0220) among patients who were overweight and obese (BMI ≥ 25 kg/m2). In the safety analysis, patients who received the EC treatment experienced significantly more grade ≥ 3 AEs (n = 599, 74.9%) than those who received the EP treatment (n = 337, 66.7%; p = 0.002). Furthermore, the EC regimen was associated with a higher risk of grade 3–4 neutropaenia (p = 0.001), thrombocytopaenia (p < 0.001) and hyponatraemia (p = 0.036), whereas the EP regimen was associated with a higher risk of grade 3–4 vomiting (p = 0.021).ConclusionsIn summary, this study presented the efficacy and safety of the EC and EP regimens in patients with ES-SCLC in the first-line setting. Patients who are overweight and obese benefit more from the EP regimen than EC regimen. Approaches to define the optimal chemotherapy regimen in different BMI subgroups are needed.

Evaluation of Sex-Based Differences in Patient Characteristics, Tumor-Related Factors, and Clinical Outcomes in Patients With Newly Diagnosed IDH-Wildtype Glioblastoma Receiving Chemoradiation

While glioblastoma (GBM) is known to have increased incidence in males, recent work has identified possible sex-specific molecular differences that could be associated with improved clinical outcomes for female GBM patients. We evaluated sex-based differences in patient and tumor characteristics, and clinical outcomes in newly diagnosed GBM patients.We reviewed 665 newly diagnosed adult IDH-wild type GBM patients with Karnofsky Performance Status (KPS) ≥60 who were treated at our institution from January 1, 2010 to May 30, 2019. Patient, tumor and treatment details were obtained from the electronic medical record. An unpaired t-test, Fisher's exact test and univariable/multivariable Cox proportional hazards modeling were used to evaluate associations, including progression free survival (PFS) and overall survival (OS).In our GBM cohort, 384 (57.7%) were male and 281 (42.3%) were female. There were no significant differences in age (60.6 years in males vs. 60.0 years in females, P = 0.16), KPS (KPS ≥90 in 46.1% in males vs. 43.4% in females, P = 0.53), extent of resection (gross total resection in 44.8% in males vs. 47.7% in females, P = 0.48), or prevalence of MGMT methylation (37.8% in males vs. 43.4% in females, P = 0.11). There were also no differences in receipt of temozolomide (95.1% of males vs. 95.7% of females, P = 0.71), radiation dose (≥59.4 Gy in 85.2% in males vs. 87.5% in females, P = 0.43), or clinical trial enrollment (24.7% males vs. 21.0% females, P = 0.27). Median OS was 22.5 months for females and 19.3 months for males (hazard ratio [HR] 0.81, 95% CI 1.03-1.48, P = 0.02). On multivariable analysis adjusted for age (adjusted hazard ratio [AHR] 1.04, 95% CI [1.03-1.05], P < 0.001), KPS ≥90 (AHR 0.73, 95% CI [0.60-0.89], P = 0.002), extent of resection (AHR 1.12, 95% CI 0.92-1.37, P = 0.26), and MGMT methylation status (AHR 0.38, 95% CI [0.31-0.47], P < 0.001), female sex (AHR 0.78, 95% CI [0.64-0.95], P = 0.015) was associated with improved OS. Superior OS in females was observed in MGMT unmethylated patients (HR 0.69, 95% CI [0.54-0.90], P = 0.005) but not in MGMT methylated patients (HR 0.85, 95% CI [0.64-1.14], P = 0.28). While age (AHR 1.02, 95% CI [1.02-1.03], P < 0.001), KPS ≥90 (AHR 0.79, 95% CI [0.66-0.95], P = 0.01), and MGMT methylation (AHR 0.40, 95% CI [0.33-0.49], P < 0.001) were associated with PFS, biologic sex was not associated with PFS (AHR 0.94, 95% CI [0.78-1.12], P = 0.49).Female sex was associated with improved survival after adjustment of known clinical covariates in newly diagnosed GBM patients. Further study is necessary to understand biologic mechanism for sex-based differences in clinical outcome given implications for therapy development and clinical trial design.D.D. Shi: None. M. Lim-Fat: None. G.C. Youssef: None. S. Whorral: None. M. Allen: None. D.N. Cagney: None. S. Tanguturi: None. D. Haas-Kogan: Research Grant; Novartis; Cellworks. A.A. Aizer: Research Grant; Varian. Consultant; Novartis. Advisory Board; Seattle Genetics. J. McFaline-Figueroa: None. U.N. Chukwueke: None. E.Q. Lee: None. D.A. Reardon: Research Grant; Celldex, Inovio, Midatech. Consultant; Amgen. Speaker's Bureau; Merck/Schering, Roche/Genentech. Advisory Board; AbbVie, Bristol Myers Squibb, Cavion, Celldex, EMD Serono, Juno Pharmaceuticals, Momenta Pharmaceuticals, Novartis, Novocure, Oxigene, Regeneron, Roche/Genentech, Stemline. L. Nayak: None. W.L. Bi: Research Grant; The Leukemia and Lymphoma Society, NIH/NCI.R. Beroukhim: Independent Contractor; Scorpion Therapeutics. Stock; Scorpion Therapeutics. K.L. Ligon: None. B.M. Alexander: Research Grant; Puma Biotechnology, Eli Lily, Celgene. Stock; Hoffman-La Roche. P.Y. Wen: None. R. Rahman: Research Grant; Project Data Sphere, LLC.

Characteristics of available studies and dissemination of research using major clinical data sharing platforms.

Over the past decade, numerous data sharing platforms have been launched, providing access to de-identified individual patient-level data and supporting documentation. We evaluated the characteristics of prominent clinical data sharing platforms, including types of studies listed as available for request, data requests received, and rates of dissemination of research findings from data requests. We reviewed publicly available information listed on the websites of six prominent clinical data sharing platforms: Biological Specimen and Data Repository Information Coordinating Center, ClinicalStudyDataRequest.com, Project Data Sphere, Supporting Open Access to Researchers-Bristol Myers Squibb, Vivli, and the Yale Open Data Access Project. We recorded key platform characteristics, including listed studies and available supporting documentation, information on the number and status of data requests, and rates of dissemination of research findings from data requests (i.e. publications in a peer-reviewed journals, preprints, conference abstracts, or results reported on the platform's website). The number of clinical studies listed as available for request varied among five data sharing platforms: Biological Specimen and Data Repository Information Coordinating Center (n = 219), ClinicalStudyDataRequest.com (n = 2,897), Project Data Sphere (n = 154), Vivli (n = 5426), and the Yale Open Data Access Project (n = 395); Supporting Open Access to Researchers did not provide a list of Bristol Myers Squibb studies available for request. Individual patient-level data were nearly always reported as being available for request, as opposed to only Clinical Study Reports (Biological Specimen and Data Repository Information Coordinating Center = 211/219 (96.3%); ClinicalStudyDataRequest.com = 2884/2897 (99.6%); Project Data Sphere = 154/154 (100.0%); and the Yale Open Data Access Project = 355/395 (89.9%)); Vivli did not provide downloadable study metadata. Of 1201 data requests listed on ClinicalStudyDataRequest.com, Supporting Open Access to Researchers-Bristol Myers Squibb, Vivli, and the Yale Open Data Access Project platforms, 586 requests (48.8%) were approved (i.e. data access granted). The majority were for secondary analyses and/or developing/validating methods (ClinicalStudyDataRequest.com = 262/313 (83.7%); Supporting Open Access to Researchers-Bristol Myers Squibb = 22/30 (73.3%); Vivli = 63/84 (75.0%); the Yale Open Data Access Project = 111/159 (69.8%)); four were for re-analyses or corroborations of previous research findings (ClinicalStudyDataRequest.com = 3/313 (1.0%) and the Yale Open Data Access Project = 1/159 (0.6%)). Ninety-five (16.1%) approved data requests had results disseminated via peer-reviewed publications (ClinicalStudyDataRequest.com = 61/313 (19.5%); Supporting Open Access to Researchers-Bristol Myers Squibb = 3/30 (10.0%); Vivli = 4/84 (4.8%); the Yale Open Data Access Project = 27/159 (17.0%)). Forty-two (6.8%) additional requests reported results through preprints, conference abstracts, or on the platform's website (ClinicalStudyDataRequest.com = 12/313 (3.8%); Supporting Open Access to Researchers-Bristol Myers Squibb = 3/30 (10.0%); Vivli = 2/84 (2.4%); Yale Open Data Access Project = 25/159 (15.7%)). Across six prominent clinical data sharing platforms, information on studies and request metrics varied in availability and format. Most data requests focused on secondary analyses and approximately one-quarter of all approved requests publicly disseminated their results. To further promote the use of shared clinical data, platforms should increase transparency, consistently clarify the availability of the listed studies and supporting documentation, and ensure that research findings from data requests are disseminated.

Bayesian Modeling for the Detection of Adverse Events Underreporting in Clinical Trials.

IntroductionSafety underreporting is a recurrent issue in clinical trials that can impact patient safety and data integrity. Clinical quality assurance (QA) practices used to detect underreporting rely on on-site audits; however, adverse events (AEs) underreporting remains a recurrent issue. In a recent project, we developed a predictive model that enables oversight of AE reporting for clinical quality program leads (QPLs). However, there were limitations to using solely a machine learning model.ObjectiveOur primary objective was to propose a robust method to compute the probability of AE underreporting that could complement our machine learning model. Our model was developed to enhance patients’ safety while reducing the need for on-site and manual QA activities in clinical trials.MethodsWe used a Bayesian hierarchical model to estimate the site reporting rates and assess the risk of underreporting. We designed the model with Project Data Sphere clinical trial data that are public and anonymized.ResultsWe built a model that infers the site reporting behavior from patient-level observations and compares them across a study to enable a robust detection of outliers between clinical sites.ConclusionThe new model will be integrated into the current dashboard designed for clinical QPLs. This approach reduces the need for on-site audits, shifting focus from source data verification to pre-identified, higher risk areas. It will enhance further QA activities for safety reporting from clinical trials and generate quality evidence during pre-approval inspections.Supplementary InformationThe online version contains supplementary material available at 10.1007/s40264-021-01094-8.

Right-censoring bias in trials shared at the Project Data Sphere platform.

e13543 Background: Independent of the bias inherent to the design and execution of clinical trials, bias may be the result of patient censoring. A bias index (BI) was developed to detect right-censoring bias and tested in datasets availabe at Project Data Sphere, a data sharing research platform maintained by the CEO Roundtable on Cancer, Inc.,* a nonprofit corporation to improve outcomes for cancer patients by openly sharing deidentified data. Methods: Project Data Sphere platform was searched for clinical comparative trials with available experimental and comparator survival datasets: overall survival (OS) and event-free survival (EFS: disease-free survival, DFS, or progresssion-free survival, PFS). The R language and the integrated development environment Rstudio were used to import and manage the datasets. BI was defined in the events time domain as the adjusted proportion of censor times below the mean event time. Comparison of BI in different datasets were made with the two-sided Wilcoson unpaired test. A weighted regression model was applied to estimate the influence of bias on survival results as measured by the hazard ratio (HR). Results: Out of 184 trials, 19 trials offered both comparator and experimental arms, 3 of them not based on survival analysis and 4 of them with 2 substudies, providing 72 datasets based on OS and/or EFS, for a total of 16532 patients (90.8% of the 18198 patients in published trials). BI over the theshold was found in 24% of EFS datasets (versus 0% in OS datasets, Wilcoxon p = 0.0007), especially in PFS (35% vs 0% in DFS datasets, p = 0.00004). Nearly two thirds of the variance in the HR of EFS datasets was explained by the HR of OS datasets (adj.R2 = 0.638, p = 1.5e-5), approaching to what was found in the corresponding publications (adj.R2 = 0.751, p = 7.81e-5). Though the trials sample is small, introducing the BI of control and experimental datasets in the model decreases the residual standard error (3.831 vs 3.958) and increases the correlation (adj.R2 = 0.99, p &lt; 2.2e-16), resulting in the model: HR(EFR) = 0.985 HR(OS) + 0.36 BI(exper) – 0.42 BI(control). Conclusions: This study is a proof of concept that right-censoring bias may be detected and estimated in clinical trials, especially in PFS datasets, and opens the possibility for correcting biased estimations in survival and increasing the precision in the prediction of OS from preliminary EFS. (*) This abstract is based on research using information obtained from ProjectDataSphere.org, which is maintained by Project Data Sphere LLC. Neither Project Data Sphere nor the owners of any information from the web site have contributed to, approved or are in any way responsible for the contents of this abstract.

Predicting toxicity-related docetaxel discontinuation and overall survival in metastatic castration-resistant prostate cancer: a pooled analysis of open phase 3 clinical trial data.

Docetaxel is widely used in metastatic castration-resistant prostate cancer (mCRPC), however its optimal use remains unclear in the current treatment landscape. Biomarkers to predict Docetaxel toxicity may help optimize treatment selection. We aimed to create a predictive model for toxicity-related Docetaxel discontinuation (TRDD). Through Project Data Sphere, we accessed individual patient data from the control arms of three frontline mCRPC trials: ASCENT2, VENICE, and MAINSAIL. The inclusion criteria for these trials were all similar and included patients with chemotherapy-naïve mCRPC. The primary outcome was occurrence of TRDD. A competing risks regression (CRR) was used to predict TRDD, after accounting for the occurrence of competing events (death or progression). The output of the model was used as the dependent variable on a classification and regression tree (CART) to identify risk groups for TRDD. Overall, 1568 patients were considered. Pooled CI of TRDD was 19% after accounting for competing events (death: 474; progression: 59) within 12 months of starting treatment. To build a risk calculator we relied on a CRR that ultimately included age, ECOG performance status, AST, bilirubin, use of analgesics, and presence of diabetes and chronic kidney disease. The CART analysis identified three risk groups that were named: low (model-derived TRDD risk ≤24%), intermediate (25-64%), and high (≥65%) risk group. In each risk group, probability of TRDD during treatment was 14%, 58%, and 79%, and median OS was 24 months, 20 months, and 13 months, respectively (p < 0.001). Treatment selection in mCRPC remains a challenge. Our model can help clinicians balance Docetaxel toxicity and efficacy. The three risk categories that we identified correlated with OS and this is particularly useful for an optimal shared decision-making process.