Towards decision-making using individualized risk estimates for personalized medicine: A systematic review of genomic classifiers of solid tumors

Decision impact studies, evidence of clinical utility for genomic assays in cancer: A scoping review

BACKGROUND

Decision impact studies have become increasingly prevalent in cancer prognostic research in recent years. These studies aim to evaluate the impact of a genomic test on decision-making and appear to be a new form of evidence of clinical utility. The objectives of this review were to identify and characterize decision impact studies in genomic medicine in cancer care and categorize the types of clinical utility outcomes reported.

METHODS

We conducted a search of four databases, Medline, Embase, Scopus and Web of Science, from inception to June 2022. Empirical studies that reported a "decision impact" assessment of a genomic assay on treatment decisions or recommendations for cancer patients were included. We followed scoping review methodology and adapted the Fryback and Thornbury Model to collect and analyze data on clinical utility. The database searches identified 1803 unique articles for title/abstract screening; 269 articles moved to full-text review.

RESULTS

87 studies met inclusion criteria. All studies were published in the last 12 years with the majority for breast cancer (72%); followed by other cancers (28%) (lung, prostate, colon). Studies reported on the impact of 19 different proprietary (18) and generic (1) assays. Across all four levels of clinical utility, outcomes were reported for 22 discrete measures, including the impact on provider/team decision-making (100%), provider confidence (31%); change in treatment received (46%); patient psychological impacts (17%); and costing or savings impacts (21%). Based on the data synthesis, we created a comprehensive table of outcomes reported for clinical utility.

CONCLUSIONS

This scoping review is a first step in understanding the evolution and uses of decision impact studies and their influence on the integration of emerging genomic technologies in cancer care. The results imply that DIS are positioned to provide evidence of clinical utility and impact clinical practice and reimbursement decision-making in cancer care. Systematic review registration: Open Science Framework osf.io/hm3jr.

A Machine Learning Platform to Optimize the Translation of Personalized Network Models to the Clinic

PURPOSE

Dynamic network models predict clinical prognosis and inform therapeutic intervention by elucidating disease-driven aberrations at the systems level. However, the personalization of model predictions requires the profiling of multiple model inputs, which hampers clinical translation.

PATIENTS AND METHODS

We applied APOPTO-CELL, a prognostic model of apoptosis signaling, to showcase the establishment of computational platforms that require a reduced set of inputs. We designed two distinct and complementary pipelines: a probabilistic approach to exploit a consistent subpanel of inputs across the whole cohort (Ensemble) and a machine learning approach to identify a reduced protein set tailored for individual patients (Tree). Development was performed on a virtual cohort of 3,200,000 patients, with inputs estimated from clinically relevant protein profiles. Validation was carried out in an in-house stage III colorectal cancer cohort, with inputs profiled in surgical resections by reverse phase protein array (n = 120) and/or immunohistochemistry (n = 117).

RESULTS

Ensemble and Tree reproduced APOPTO-CELL predictions in the virtual patient cohort with 92% and 99% accuracy while decreasing the number of inputs to a consistent subset of three proteins (40% reduction) or a personalized subset of 2.7 proteins on average (46% reduction), respectively. Ensemble and Tree retained prognostic utility in the in-house colorectal cancer cohort. The association between the Ensemble accuracy and prognostic value (Spearman ρ = 0.43; P = .02) provided a rationale to optimize the input composition for specific clinical settings. Comparison between profiling by reverse phase protein array (gold standard) and immunohistochemistry (clinical routine) revealed that the latter is a suitable technology to quantify model inputs.

CONCLUSION

This study provides a generalizable framework to optimize the development of network-based prognostic assays and, ultimately, to facilitate their integration in the routine clinical workflow.

Big Data Approaches for Modeling Response and Resistance to Cancer Drugs

Despite significant progress in cancer research, current standard-of-care drugs fail to cure many types of cancers. Hence, there is an urgent need to identify better predictive biomarkers and treatment regimes. Conventionally, insights from hypothesis-driven studies are the primary force for cancer biology and therapeutic discoveries. Recently, the rapid growth of big data resources, catalyzed by breakthroughs in high-throughput technologies, has resulted in a paradigm shift in cancer therapeutic research. The combination of computational methods and genomics data has led to several successful clinical applications. In this review, we focus on recent advances in data-driven methods to model anticancer drug efficacy, and we present the challenges and opportunities for data science in cancer therapeutic research.

Genome-Scale Signatures of Gene Interaction from Compound Screens Predict Clinical Efficacy of Targeted Cancer Therapies

Identifying reliable drug response biomarkers is a significant challenge in cancer research. We present computational analysis of resistance (CARE), a computational method focused on targeted therapies, to infer genome-wide transcriptomic signatures of drug efficacy from cell line compound screens. CARE outputs genome-scale scores to measure how the drug target gene interacts with other genes to affect the inhibitor efficacy in the compound screens. Such statistical interactions between drug targets and other genes were not considered in previous studies but are critical in identifying predictive biomarkers. When evaluated using transcriptome data from clinical studies, CARE can predict the therapy outcome better than signatures from other computational methods and genomics experiments. Moreover, the CARE signatures for the PLX4720 BRAF inhibitor are associated with an anti-programmed death 1 clinical response, suggesting a common efficacy signature between a targeted therapy and immunotherapy. When searching for genes related to lapatinib resistance, CARE identified PRKD3 as the top candidate. PRKD3 inhibition, by both small interfering RNA and compounds, significantly sensitized breast cancer cells to lapatinib. Thus, CARE should enable large-scale inference of response biomarkers and drug combinations for targeted therapies using compound screen data.

Revisiting Dukes' paradigm; some node positive colon cancer patients have better prognosis than some node negative patients

BACKGROUND

The current study tried to evaluate the prognostic value of a modified staging system compared to the American Joint Committee on Cancer (AJCC) staging system for patients with colon cancer.

PATIENTS AND METHODS

Surveillance, epidemiology and end results (SEER) database (2004-2014) was queried through SEER*Stat program and AJCC 7th stages were constructed. Through recursive partitioning analysis and subsequent decision tree formation, suggested new stages were formulated based on T and N descriptors. Overall survival analyses were performed through Kaplan-Meier analysis. The cancer-specific Cox regression hazard (adjusted for age, gender, sub-site, grade, race and surgery) was calculated and pair wise comparisons of hazard ratios were conducted.

RESULTS

A total of 159,683 non-metastatic patients with colon cancer were recruited in the analysis. Pair wise hazard ratio comparisons among different AJCC 7th stages were conducted and all comparisons were significant (P < 0.0001). However, it should be noted that the adjusted risk of death among stage IIC patients was higher than stage IIIA and IIIB. Pair wise hazard ratio comparisons among different modified system stages were also conducted and all comparisons were significant (P < 0.0001). The outcomes of survival analysis were the same regardless of the number of examined lymph nodes (< 12 vs. ≥ 12). Concordance index (using death from colon cancer as the dependent variable) for AJCC 6th staging system was 0.704 (SE 0.002; 95% CI 0.701-0.708); for AJCC 7th staging system was 0.708 (SE 0.002; 95% CI 0.704-0.711); for Dukes staging system was 0.670 (SE 0.002; 95% CI 0.666-0.674); and for modified staging system was 0.712 (SE 0.002; 95% CI 0.709-0.716).

CONCLUSION

The proposed modified staging system provided an improved prognostication for colon cancer patients (particularly for stage II/III disease) compared to AJCC staging system. Further external validation of the proposed staging system is needed before adoption into routine practice.