Merging Genomics Public Datasets with Clinical Cancer Registry Data - Lessons Learned.

Publicly available datasets - for example via cBioPortal for Cancer Genomics - could be a valuable source for benchmarks and comparisons with local patient records. However, such an approach is only valid if patient cohorts are comparable to each other and if the documentation is complete and sufficient. In this paper, records from exocrine pancreatic cancer patients documented in a local cancer registry are compared with two public datasets to calculate overall survival. Several data preprocessing steps were necessary to ensure comparability of the different datasets and a common database schema was created. Our assumption that the public datasets could be used to augment the data of the local cancer registry could not be validated, since the analysis on overall survival showed a significant difference. We discuss several reasons and explanations for this finding. So far, comparing different datasets with each other and drawing medical conclusions on such comparisons should be conducted with great caution.

Larotrectinib (LOXO-101).

One of the most challenging issues in oncology research and treatment is identifying oncogenic drivers within an individual patient's tumor which can be directly targeted by a clinically available therapeutic drug. In this context, gene fusions as one important example of genetic aberrations leading to carcinogenesis follow the widely accepted concept that cell growth and proliferation are driven by the accomplished fusion (usually involving former proto-oncogenes) and may therefore be successfully inhibited by substances directed against the fusion. This concept has already been established with oncogenic gene fusions like BCR-ABL in chronic myelogenous leukemia (CML) or anaplastic lymphoma kinase (ALK) in lung cancer, including special tyrosine kinase inhibitors (TKIs) which are able to block the activation of the depending downstream proliferation pathways and, consequently, tumor growth. During the last decade, the NTRK1, 2, and 3 genes, encoding the TRKA, B, and C proteins, have attracted increasing attention as another significant and targetable gene fusion in a variety of cancers. Several TRK inhibitors have been developed, and one of them, Larotrectinib (formerly known as LOXO-101), represents an orally available, selective inhibitor of the TRK receptor family that has already shown substantial clinical benefit in both pediatric and adult patients harboring an NTRK gene fusion over the last few years.

Olaparib.

Olaparib (Lynparza [AstraZeneca, Cambridge, UK], formerly referred to as AZD2281 or KU0059436) is an oral poly(ADP-ribose) polymerase (PARP) inhibitor. It is rationally designed to act as a competitive inhibitor of NAD+ at the catalytic site of PARP1 and PARP2, both members of the PARP family of enzymes that are central to the repair of DNA single-strand breaks (SSBs) mediated via the base excision repair (BER) pathway. Inhibition of the BER pathway by olaparib leads to the accumulation of unrepaired SSBs, which leads to the formation of deleterious double-strand breaks (DSBs). In cells with an intact homologous recombination (HR) pathway, these DSBs can be repaired effectively. However, in tumors with homologous recombination repair deficiencies, olaparib causes synthetic lethality through the combination of two molecular events that are otherwise nonlethal when occurring in isolation. Olaparib is already approved for the treatment of patients with recurrent ovarian cancer and a BRCA mutation, and it has been shown to provide clinically meaningful benefits among such patients. It has also shown promising activity in patients with metastatic breast or prostate cancer and a germline BRCA mutation. Besides its usage as a single agent, olaparib can also act either as a chemo- and/or radiosensitizer, due to its ability to potentiate the cytotoxic effects of these therapeutic agents. However, a clear patient benefit for the latter application has not been demonstrated yet.

Clinical Cancer Registries - Are They Up for Health Services Research?

Clinical cancer registries are a valuable data source for health services research (HSR). HSR is in need of high quality routine care data for its evaluations. However, the secondary use of routine data - such as documented cancer cases in a disease registry - poses new challenges in terms of data quality, IT-management, documentation processes and data privacy. In the clinical cancer registry Heilbronn-Franken, real-world data from the Giessen Tumor Documentation System (GTDS) was utilized for analyses of patients' disease processes and guideline adherence in follow-up care. A process was developed to map disease state definitions to fields of the GTDS database and extract patients' disease progress information. Thus, the disease process of sub-cohorts could be compared to each other, e.g., comparison of disease free survival of HER2 (human epidermal growth factor receptor 2)-positive and -negative women who were treated with Trastuzumab, a targeted therapy applied in breast cancer. In principle, such comparisons are feasible and of great value for HSR as they depict a routine care setting of a diverse patient cohort. Yet, local documentation practice, missing flow of information from external health care providers or small sub-cohorts impede the analyses of clinical cancer registries data bases and usage for HSR.

Transition probabilities of HER2-positive and HER2-negative breast cancer patients treated with Trastuzumab obtained from a clinical cancer registry dataset.

Records of female breast cancer patients were selected from a clinical cancer registry and separated into three cohorts according to HER2-status (human epidermal growth factor receptor 2) and treatment with or without Trastuzumab (a humanized monoclonal antibody). Propensity score matching was used to balance the cohorts. Afterwards, documented information about disease events (recurrence of cancer, metastases, remission of local/regional recurrences, remission of metastases and death) found in the dataset was leveraged to calculate the annual transition probabilities for every cohort.

A method for using real world data in breast cancer modeling.

OBJECTIVES: Today, hospitals and other health care-related institutions are accumulating a growing bulk of real world clinical data. Such data offer new possibilities for the generation of disease models for the health economic evaluation. In this article, we propose a new approach to leverage cancer registry data for the development of Markov models. Records of breast cancer patients from a clinical cancer registry were used to construct a real world data driven disease model. METHODS: We describe a model generation process which maps database structures to disease state definitions based on medical expert knowledge. Software was programmed in Java to automatically derive a model structure and transition probabilities. We illustrate our method with the reconstruction of a published breast cancer reference model derived primarily from clinical study data. In doing so, we exported longitudinal patient data from a clinical cancer registry covering eight years. The patient cohort (n=892) comprised HER2-positive and HER2-negative women treated with or without Trastuzumab. RESULTS: The models generated with this method for the respective patient cohorts were comparable to the reference model in their structure and treatment effects. However, our computed disease models reflect a more detailed picture of the transition probabilities, especially for disease free survival and recurrence. CONCLUSIONS: Our work presents an approach to extract Markov models semi-automatically using real world data from a clinical cancer registry. Health care decision makers may benefit from more realistic disease models to improve health care-related planning and actions based on their own data.

Prediction of 5-Year Survival with Data Mining Algorithms.

Survival time prediction at the time of diagnosis is of great importance to make decisions about treatment and long-term follow-up care. However, predicting the outcome of cancer on the basis of clinical information is a challenging task. We now examined the ability of ten different data mining algorithms (Perceptron, Rule Induction, Support Vector Machine, Linear Regression, Naïve Bayes, Decision Tree, k-nearest Neighbor, Logistic Regression, Neural Network, Random Forest) to predict the dichotomous attribute "5-year-survival" based on seven attributes (sex, UICC-stage, etc.) which are available at the time of diagnosis. For this study we made use of the nationwide German research data set on colon cancer provided by the Robert Koch Institute. To assess the results a comparison between data mining algorithms and physicians' opinions was performed. Therefore, physicians guessed the survival time by leveraging the same seven attributes. The average accuracy of the physicians' opinion was 59%, the average accuracy of the machine learning algorithms was 67.7%.

Personalized treatment for colorectal cancer: novel developments and putative therapeutic strategies.

BACKGROUND: Colorectal cancer (CRC) is the third most common cancer diagnosed worldwide and continues to be a major healthcare concern. Molecular heterogeneity of CRC is believed to be one of the main factors responsible for the considerable variability in treatment response. With the recent development of powerful genomic technologies, novel insights in tumor biology of CRC have now been provided, facilitating the recognition of new molecular subtypes with prognostic and predictive implications. PURPOSE: The purpose of this review article is to summarize current knowledge about genomic, epigenomic, and proteomic characteristics of CRC, as well as their implications for biomarker identification and individualized targeted therapy. CONCLUSION: Supplementing the findings from several previous studies, the Cancer Genome Atlas (TCGA) project recently finalized the systematic characterization of CRC resulting in the first tumor dataset with complete molecular measurements at DNA, RNA, and protein levels. The challenge now is to translate these findings into a robust and reproducible CRC classification system linking molecular features of the tumor to precision medicine.

Analysis of Xq27-28 linkage in the international consortium for prostate cancer genetics (ICPCG) families.

BACKGROUND: Genetic variants are likely to contribute to a portion of prostate cancer risk. Full elucidation of the genetic etiology of prostate cancer is difficult because of incomplete penetrance and genetic and phenotypic heterogeneity. Current evidence suggests that genetic linkage to prostate cancer has been found on several chromosomes including the X; however, identification of causative genes has been elusive. METHODS: Parametric and non-parametric linkage analyses were performed using 26 microsatellite markers in each of 11 groups of multiple-case prostate cancer families from the International Consortium for Prostate Cancer Genetics (ICPCG). Meta-analyses of the resultant family-specific linkage statistics across the entire 1,323 families and in several predefined subsets were then performed. RESULTS: Meta-analyses of linkage statistics resulted in a maximum parametric heterogeneity lod score (HLOD) of 1.28, and an allele-sharing lod score (LOD) of 2.0 in favor of linkage to Xq27-q28 at 138 cM. In subset analyses, families with average age at onset less than 65 years exhibited a maximum HLOD of 1.8 (at 138 cM) versus a maximum regional HLOD of only 0.32 in families with average age at onset of 65 years or older. Surprisingly, the subset of families with only 2-3 affected men and some evidence of male-to-male transmission of prostate cancer gave the strongest evidence of linkage to the region (HLOD = 3.24, 134 cM). For this subset, the HLOD was slightly increased (HLOD = 3.47 at 134 cM) when families used in the original published report of linkage to Xq27-28 were excluded. CONCLUSIONS: Although there was not strong support for linkage to the Xq27-28 region in the complete set of families, the subset of families with earlier age at onset exhibited more evidence of linkage than families with later onset of disease. A subset of families with 2-3 affected individuals and with some evidence of male to male disease transmission showed stronger linkage signals. Our results suggest that the genetic basis for prostate cancer in our families is much more complex than a single susceptibility locus on the X chromosome, and that future explorations of the Xq27-28 region should focus on the subset of families identified here with the strongest evidence of linkage to this region.

Genomewide scan in families with schizophrenia from the founder population of Afrikaners reveals evidence for linkage and uniparental disomy on chromosome 1.

We report on our initial genetic linkage studies of schizophrenia in the genetically isolated population of the Afrikaners from South Africa. A 10-cM genomewide scan was performed on 143 small families, 34 of which were informative for linkage. Using both nonparametric and parametric linkage analyses, we obtained evidence for a small number of disease loci on chromosomes 1, 9, and 13. These results suggest that few genes of substantial effect exist for schizophrenia in the Afrikaner population, consistent with our previous genealogical tracing studies. The locus on chromosome 1 reached genomewide significance levels (nonparametric LOD score of 3.30 at marker D1S1612, corresponding to an empirical P value of.012) and represents a novel susceptibility locus for schizophrenia. In addition to providing evidence for linkage for chromosome 1, we also identified a proband with a uniparental disomy (UPD) of the entire chromosome 1. This is the first time a UPD has been described in a patient with schizophrenia, lending further support to involvement of chromosome 1 in schizophrenia susceptibility in the Afrikaners.

Confirmation of the prostate cancer susceptibility locus HPCX in a set of 104 German prostate cancer families.

BACKGROUND: Several prostate cancer (PCa) susceptibility loci have been reported, but attempts to confirm them in independent data sets have produced inconsistent results. It is not yet clear, how much of this variation is due to differences between different populations. HPCX was originally identified in a combined data set of PCa families from the USA and Scandinavia. Considerable differences in the frequency of linked families were observed in this heterogeneous family sample as well as in following studies. METHODS: In order to estimate the significance of HPCX in the German population, DNA samples from 104 PCa families were genotyped at six polymorphic markers spanning a region of approximately 14 cM on Xq27-28, which includes the proposed HPCX candidate locus. RESULTS: In the entire data set, a maximum NPL Z score of 1.20 (P = 0.11) at marker DXS984 was observed. Statistically significant evidence for linkage was obtained in the subset of 63 families with early-onset disease (i.e., < or = 65 years) with a maximum NPL Z score of 2.32 (P = 0.009) at the same location. CONCLUSION: Our results confirm the existence of a prostate cancer susceptibility gene on Xq27-28 also in the German population.