Process mining is a relatively new method that connects data science and process modelling. In the past years a series of applications with health care production data have been presented in process ...discovery, conformance check and system enhancement. In this paper we apply process mining on clinical oncological data with the purpose of studying survival outcomes and chemotherapy treatment decision in a real-world cohort of small cell lung cancer patients treated at Karolinska University Hospital (Stockholm, Sweden). The results highlighted the potential role of process mining in oncology to study prognosis and survival outcomes with longitudinal models directly extracted from clinical data derived from healthcare.
•Pooled OS/safety data from the pivotal phase II studies of alectinib in ALK+ NSCLC.•Alectinib demonstrated a final pooled median OS of 29.1 months (95% CI 21.3–39.0).•No new or unexpected safety ...findings were observed.•Alectinib shows robust efficacy and manageable safety in advanced ALK+ NSCLC.
A pooled analysis of two open-label phase II studies of alectinib (NP28673 NCT01801111 and NP28761 NCT01871805) demonstrated clinical activity in patients with advanced, anaplastic lymphoma kinase-positive (ALK+) non-small-cell lung cancer (NSCLC) previously treated with crizotinib. Longer-term and final pooled analyses of overall survival (OS) and safety data from the two studies are presented here.
The pooled population totaled 225 patients (NP28673: n = 138, NP28761: n = 87) who received 600 mg oral alectinib twice daily until disease progression, death, or withdrawal. OS was defined as the time from date of first treatment to date of death, regardless of cause. OS was estimated using Kaplan–Meier methodology, with 95% confidence intervals (CIs) determined using the Brookmeyer–Crowley method. Safety was assessed through adverse event (AE) reporting.
Baseline characteristics were generally comparable between the studies. At final data cutoff (October 27, 2017 NP28673, October 12, 2017 NP28761; median pooled follow-up time, ∼21 months), 53.3% of patients had died, 39.1% were alive and in follow-up, and 7.6% had withdrawn consent or were lost to follow-up. Alectinib demonstrated a median OS of 29.1 months (95% CI 21.3–39.0). No new or unexpected safety findings were observed. The most common all-grade AEs included constipation (39.1%), fatigue (35.1%), peripheral edema (28.4%), myalgia (26.2%), and nausea (24.0%).
Updated results from this pooled analysis further demonstrate that alectinib has robust clinical activity and a manageable safety profile in patients with advanced, ALK+ NSCLC pretreated with crizotinib.
Gemcitabine/carboplatin‐induced myelosuppressive adverse drug reactions (ADRs) are clinical problems leading to patient suffering and dose alterations. There is a need for personalised medicine to ...improve treatment effects and patients' well‐being. We tested four genetic variants, rs11141915, rs1901440, rs12046844 and rs11719165, previously suggested as potential biomarkers for gemcitabine‐induced leukopenia/neutropenia in Japanese patients, in 213 Swedish gemcitabine/carboplatin‐treated non‐small cell lung cancer (NSCLC) patients. DNA was genotyped using TaqMan probes and real‐time PCR. The relationships between the risk alleles and low toxicity (non‐ADR: Common Terminology Criteria for Adverse Events CTCAE grades 0) or high toxicity (ADR: CTCAE grades 3–4) of platelets, leukocytes and neutrophils were evaluated using Fisher's exact test. The risk alleles did not correlate with myelosuppression, and the strongest borderline significance (not withstanding adjustment for multiple testing) was for rs1901440 (neutropenia, p = 0.043) and rs11719165 (leukopenia, p = 0.049) where the risk alleles trended towards lower toxicity, contrasting with previous study findings. Risk alleles and higher risk scores were more common among our patients. We conclude that the genetic variants do not apply to Swedish patients treated with gemcitabine/carboplatin. However, they can still be important in other populations and cohorts, especially in a gemcitabine monotherapy setting, where the causal genetic variation might influence myelosuppressive ADRs.
•The severity of gemcitabine/carboplatin-induced hematological toxicities varies extensively.•Severe hematological toxicities can lead to the need for postponed treatment, reduced doses and treatment ...discontinuation.•Genetic variability may explain and predict the variation in severe leukopenia and neutropenia.
Gemcitabine/carboplatin treatment is known to cause severe adverse drug reactions which can lead to the need for reduction or cessation of chemotherapy. It would be beneficial to identify patients at risk of severe hematological toxicity in advance before treatment start. This study aims to identify genetic markers for gemcitabine/carboplatin-induced leukopenia and neutropenia in non-small cell lung cancer patients.
Whole-exome sequencing was performed on 215 patients. Association analysis was performed on single-nucleotide variants (SNVs) and genes, and the validation was based on an independent genome-wide association study (GWAS). Based on the association and validation analyses the genetic variants were then selected for and used in weighted genetic risk score (wGRS) prediction models for leukopenia and neutropenia.
Association analysis identified 50 and 111 SNVs, and 12 and 20 genes, for leukopenia and neutropenia, respectively. Of these SNVS 20 and 19 were partially validated for leukopenia and neutropenia, respectively. The genes SVIL (p = 2.48E-06) and EFCAB2 (p = 4.63E-06) were significantly associated with leukopenia contain the partially validated SNVs rs3740003, rs10160013, rs1547169, rs10927386 and rs10927387. The wGRS prediction models showed significantly different risk scores for high and low toxicity patients.
We have identified and partially validated genetic biomarkers in SNVs and genes correlated to gemcitabine/carboplatin-induced leukopenia and neutropenia and created wGRS models for predicting the risk of chemotherapy-induced hematological toxicity. These results provide a strong foundation for further studies of chemotherapy-induced toxicity.
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.
Chemotherapies are associated with significant interindividual variability in therapeutic effect and adverse drug reactions. In lung cancer, the use of gemcitabine and carboplatin induces grade 3 or ...4 myelosuppression in about a quarter of the patients, while an equal fraction of patients is basically unaffected in terms of myelosuppressive side effects. We therefore set out to identify genetic markers for gemcitabine/carboplatin-induced myelosuppression.
We exome sequenced 32 patients that suffered extremely high neutropenia and thrombocytopenia (grade 3 or 4 after first chemotherapy cycle) or were virtually unaffected (grade 0 or 1). The genetic differences/polymorphism between the groups were compared using six different bioinformatics strategies: (i) whole-exome nonsynonymous single-nucleotide variants association analysis, (ii) deviation from Hardy-Weinberg equilibrium, (iii) analysis of genes selected by a priori biologic knowledge, (iv) analysis of genes selected from gene expression meta-analysis of toxicity datasets, (v) Ingenuity Pathway Analysis, and (vi) FunCoup network enrichment analysis.
A total of 53 genetic variants that differed among these groups were validated in an additional 291 patients and were correlated to the patients' myelosuppression. In the validation, we identified rs1453542 in OR4D6 (P = 0.0008; OR, 5.2; 95% CI, 1.8-18) as a marker for gemcitabine/carboplatin-induced neutropenia and rs5925720 in DDX53 (P = 0.0015; OR, 0.36; 95% CI, 0.17-0.71) as a marker for thrombocytopenia. Patients homozygous for the minor allele of rs1453542 had a higher risk of neutropenia, and for rs5925720 the minor allele was associated with a lower risk for thrombocytopenia.
We have identified two new genetic markers with the potential to predict myelosuppression induced by gemcitabine/carboplatin chemotherapy.
Background: Whole-brain radiotherapy (WBRT) has been the standard of care for multiple NSCLC brain metastases but due to its toxicity and lack of survival benefit, its use in the palliative setting ...is being questioned.
Patient and methods: This was a single institution cohort study including brain metastasized lung cancer patients who received WBRT at Karolinska University Hospital. Information about Recursive Partitioning Analysis (RPA) and Graded Prognostic Assessment (GPA) scores, demographics, histopathological results and received oncological therapy were collected. Predictors of overall survival (OS) from the time of received WBRT were identified by Cox regression analyses. OS between GPA and RPA classes were compared by pairwise log rank test. A subgroup OS analysis was performed stratified by RPA class.
Results: The cohort consisted of 280 patients. RPA 1 and 2 classes had better OS compared to class 3, patients with GPA <1.5 points had better OS compared to GPA≥ 1.5 points and age >70 years was associated with worse OS (p< .0001 for all comparisons). In RPA class 2 subgroup analysis GPA ≥1.5 points, age ≤70 years and CNS surgery before salvage WBRT were independent positive prognostic factors.
Conclusions: RPA class 3 patients should not receive WBRT, whereas RPA class 1 patients should receive WBRT if clinically indicated. RPA class 2 patients with age ≤70 years and GPA ≥1.5 points should be treated as RPA 1. WBRT should be omitted in RPA 2 patients with age >70. In RPA 2 patients with age ≤70 years and GPA <1.5 points WBRT could be a reasonable option.