Gradient Boosting Machine based prediction of chemotherapy response and role of p53 mutational and smoking status for progression free survival in metastatic colorectal cancer

Gradient Boosting Machine based prediction of chemotherapy response and role of p53 mutational and smoking status for progression free survival in metastatic colorectal cancer | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Gradient Boosting Machine based prediction of chemotherapy response and role of p53 mutational and smoking status for progression free survival in metastatic colorectal cancer Oğuzhan Yıldız, Ali Fuat Gürbüz, Melek Karakurt Eryılmaz, Murat Araz, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4265594/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background: Identifying predictors of response or progression after first-line chemotherapy for stage 4 colorectal cancer remains a challenge. This study aims to evaluate the correlation between patient outcomes and the p53 mutational status and smoking status of tumors using various machine learning methods. Material and methods: We consecutively recruited all patients diagnosed with metastatic colorectal cancer at an academic center within a specified time period. Response to first-line chemotherapy and associated factors were assessed using various machine learning models. The most accurate model was further optimized. Additionally, common clinical features, MMR, p53, and RAS status were tested for correlation with the outcome. Feature importance and calibration plots were generated, and univariate and multivariate Cox models were utilized to analyze associates of progression-free survival (PFS). Results: A total of 101 newly diagnosed metastatic colorectal cancer patients initiating first-line chemotherapy were included. The median age was 62, and 69% of the cases were male. We evaluated 15 machine learning models to predict the binary outcome of best response to chemotherapy, among which LightGBM demonstrated the highest baseline accuracy of 0.71. Further tuning of the LightGBM model improved accuracy to 0.79, with a macro average AUC value of 0.82. Age at diagnosis, maximum metastatic dimension of cancer, and metastatic status at diagnosis were identified as the three most important features. Genetic variables did not establish significant feature importance for response analysis. Survival analysis revealed an association between PFS and p53 mutation status (Exp(B) = 0.52, Wald = 6.98, P = 0.008) and smoking pack years (Exp(B) = 0.99, Wald = 4.28, P = 0.039). Discussion: Utilizing LightGBM as a machine learning method, we developed a predictive model with good accuracy for assessing response to first-line treatment. If confirmed and further improved, such a model could aid in identifying responders to first-line chemotherapy in metastatic colorectal cancer patients and suggesting alternative chemotherapy options for non-responders. Furthermore, our findings highlight the prognostic importance of genetic features, particularly p53 mutation status, and smoking pack years for PFS duration in this context. metastatic colorectal cancer smoking status p53 mutation progression-free survival Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Introduction Colorectal cancer ranks as the third most prevalent cancer globally and is the third leading cause of cancer-related mortality. Annually, 1.3 million new cases are diagnosed, with approximately 20% of patients identified at the metastatic stage. The 5-year survival rates are reported at 14%. The cornerstone of treatment for metastatic CRC lies in systemic therapy, including cytotoxic chemotherapy, immunotherapy, biological therapy such as antibodies against cellular growth factors, and their combinations. Recent clinical studies conducted within the past five years underscore the significance of tailoring treatment regimens based on pathological and molecular characteristics. This patient cohort exhibits susceptibility to various molecular alterations that can be effectively targeted. Notably, advancements achieved over the last two decades have resulted in a significant increase in overall survival times, progressing from 10 to 20 months[ 1 – 4 ]. The aim of this article is to underscore the progress made in the treatment of metastatic colorectal cancer and the significance of personalized treatment approaches. The Light Gradient Boosting Machine (LightGBM) utilizes tree-based learning algorithms and distinguishes itself with the implementation of two innovative techniques: Gradient-based One-Side Sampling (GOSS) and Exclusive Feature Bundling (EFB). GOSS selectively excludes small gradient data samples, focusing instead on large gradient data to improve information gain estimation with a reduced dataset. EFB, on the other hand, combines mutually exclusive features to effectively reduce their number. The 'Light' designation signifies its exceptional speed. LightGBM has emerged as a preferred choice due to its accelerated training speed, heightened efficiency, reduced memory utilization, superior accuracy, and adept handling of large-scale datasets [ 5 ]. Machine learning models are designed with the objective of discerning structures and relationships within data, particularly in domains such as data analysis and pattern recognition. These models undergo training for specific tasks and are subsequently deployed to make predictions on novel data. Machine learning models serve various purposes, including predicting future events based on historical data, identifying patterns within datasets, examining the impact of independent variables on dependent variables, utilizing feature selection and dimensionality reduction techniques to mitigate complexity in large datasets and enhance the model's generalization capability, as well as detecting anomalies by identifying deviations from normal behavior[ 5 – 8 ]. Our objective is to provide an overview of the Light Gradient Boosting Machine (LightGBM) and its innovative techniques, Gradient-Based One-Side Sampling (GOSS), and Exclusive Feature Bundling (EFB), as well as to discuss the broader context of machine learning models and their applications. This study aims to assess the correlation between patient outcomes and the P53 mutation status of tumors. The influence of P53 genetic analysis on treatment modalities and patient outcomes will be examined, emphasizing its significance in guiding treatment decisions. Material and methods The study investigated patients diagnosed with metastatic colorectal cancer who received treatment at the Department of Medical Oncology, Necmettin Erbakan University Faculty of Medicine Hospital between July 7, 2017, and July 1, 2023. Inclusion criteria comprised individuals aged 18 or above who had previously undergone screening for the P53 mutation and were diagnosed with metastatic colorectal cancer. Patient records retrieved from the archives were analyzed to evaluate the response to initial chemotherapy and associated factors using various machine learning models. The study employed a diverse array of machine learning models, including LightGBM, Ridge Regression, XGBoost, Decision Tree, Gradient Boosting Classifier, Linear Discriminant Analysis (LDA), Random Forest, AdaBoost, Logistic Regression, Naive Bayes, Extra Trees, K-Nearest Neighbors (KNN), Dummy Classifier/Regressor, Quadratic Discriminant Analysis (QDA), and Support Vector Machines (SVM). The most accurate model underwent further optimization, considering common clinical features as well as the status of MMR, P53, and RAS for correlation with outcomes. Feature importance and calibration plots were generated, alongside other analyses. Additionally, univariate and multivariate Cox models were developed to explore the determinants of progression-free survival (PFS). Ethical approval for the study was obtained from the ethics committee of Necmettin Erbakan University Medical Faculty. Results Baseline charecteristics The study included a total of 101 newly diagnosed metastatic colorectal cancer patients who initiated first-line chemotherapy. Among the 101 eligible patients, 34 were categorized as P53 wild type, whereas 67 were identified as P53 mutant. The median age of the cohort was 62, with males comprising 69% of the cases. At the time of diagnosis, metastasis was observed in 59 patients, while 52 were found not to have metastasis. Table 1 presents an overview of the baseline characteristics and clinicopathological features. We assessed 15 machine learning models with the aim of predicting the binary outcome of the best response to chemotherapy. Among these models, LightGBM exhibited the highest baseline accuracy of 0.71 (Table 2). The LightGBM model was subsequently fine-tuned, leading to an enhanced accuracy figure of 0.79 and a macro average AUC value of 0.82 (Table 3). The feature importance plot assessed various factors, including age at diagnosis, maximum metastatic dimension, metastatic status, number of metastatic systems, gender, comorbidity, metastatic involvement limited to the liver, cumulative cigarette package years, comorbidity count, MMR status, RAS positivity, tumor location, ECOG performance status, P53 mutation status, and metastatic status at diagnosis. According to the plot, the three most significant features were age at diagnosis, maximum metastatic dimension, and metastatic status at diagnosis, with variable importance scores of 148, 90, and 58, respectively (Table 4). The study examined the correlates of progression-free survival (PFS), considering factors such as age, sex, ECOG performance status, comorbidity, number of comorbidities, cumulative smoking pack years, metastatic status at diagnosis, cancer location, maximum metastatic dimension, number of metastatic organ systems, presence of only liver metastases, RAS positivity, P53 mutation status, and MMR status (Table 5). Genetic variables did not demonstrate significant feature importance for response analysis. However, in survival analysis, progression-free survival (PFS) was found to be associated with p53 mutation status (Exp(B) = 0.52, Wald = 6.98, P = 0.008) and smoking pack years (Exp(B) = 0.99, Wald = 4.28, P = 0.039). (Table 6) The median progression-free survival (mPFS) was 11.6 months (7.7–15.4 months) across all groups. Specifically, the mPFS was 16.6 months (14.1–19.1 months) in the P53 mutant group, while it was 10 months (9.5–10.6 months) in the P53 wild-type group (Table 7). The median progression-free survival (mPFS) was 17.6 months (14.8–20.4 months) in the group of smokers, whereas it was 10.7 months (9.1–12.3 months) in the group of non-smokers (Table 8). Discussion Globally, there has been a significant increase in the incidence of colorectal cancer (CRC) due to lifestyle changes. Although there have been advancements in comprehensive treatments for CRC, such as surgery, chemotherapy, and radiotherapy, the 5-year overall survival (OS) and disease-free survival (DFS) rates remain at approximately 45% and 40%, respectively. This highlights the continued significance of CRC as a major public health concern. The importance of identifying new prognostic factors is emphasized by the heterogeneity of tumors. This enables the development of precise treatment strategies that target unfavorable prognostic indicators at the molecular level[ 9 ]. Machine learning algorithms, including advanced techniques such as LightGBM, play a crucial role in cancer treatment. For instance, Derara Ruba Rufo et al. utilized the LightGBM model to accurately diagnose early diabetes mellitus with an AUC (Area Under the Curve) of 98.1%. Our study demonstrates the effectiveness of LightGBM in predicting first-line chemotherapy response in metastatic CRC patients, providing a valuable tool to assist in treatment selection and optimization. Additionally, the study contributes to the growing body of evidence aimed at improving prognostication and personalizing treatment. Moreover, it may suggest alternative treatment options for non-responders. Future research should focus on conducting prospective clinical trials to improve the accuracy of prediction models and understand the molecular mechanisms underlying treatment resistance. We aim to enhance outcomes for patients grappling with this challenging disease by leveraging advancements in molecular profiling, machine learning, and clinical data analysis.[ 10 – 12 ] TP53 variations are present in 43.2% of colorectal cancers. As a transcription factor, p53 regulates various cellular responses, including DNA repair, cell cycle control, senescence, metabolism, and cell death. The frequent mutations of p53 pose challenges to therapeutic targeting, as standard chemotherapies rely on wild-type p53 function to trigger apoptosis and arrest the cell cycle. The study reported a notably prolonged median progression-free survival (mPFS) of 16.6 months in patients with P53 mutations compared to those with P53 wild-type status, which was 10 months. In 2003, Barry Iacopetta's meta-analysis revealed a potential association between p53 mutation and poorer survival in the majority of analyzed studies, although some studies contradicted this, reporting no association or even better outcomes. Rabih et al. observed an extended median progression-free survival (mPFS) with bevacizumab treatment in patients with P53 mutations. In contrast, Peng Wang et al. demonstrated significantly shorter disease-free survival and overall survival in patients with p53-positive metastatic colorectal cancer compared to those who were p53-negative. According to AJ Munro et al.'s extensive review, abnormal p53 did not affect outcomes in patients receiving chemotherapy. However, Kyoung Min Kim et al.'s study on metastatic colorectal cancer patients indicated shorter overall survival in the p53 mutation group, although there were no significant effects on recurrence-free survival. These findings highlight the varying impact of P53 mutations on the prognosis and treatment response of colorectal cancer patients in different studies and clinical contexts.[ 9 , 13 – 16 ]. Studies conducted by Calengeo et al. and Phipss AI et al. have reported a significant association between smoking and increased colorectal cancer (CRC)-related mortality among patients with metastatic colorectal cancer (mCRC). It is important to consider these findings when discussing the potential risks of smoking with patients who have been diagnosed with mCRC. Additionally, Munro AJ et al. demonstrated that current smokers face more than twice the risk of CRC-related death compared to former and never smokers. However, according to a meta-analysis by Yang, Lu-Ping, and colleagues, a history of smoking did not significantly affect the prognosis of mCRC, despite previous findings suggesting otherwise. Furthermore, while left-sided tumor location has been identified as a positive prognostic indicator in mCRC, a meta-analysis of a large cohort of Chinese colorectal cancer patients suggested a potential extension of overall survival (OS) in smoking patients with right-sided colon tumors. These results suggest that the prognosis in mCRC may be influenced by factors such as tumor location and patient demographics, which may affect the relationship between smoking and prognosis. In addition, our study found that first-line chemotherapy treatment significantly extended the median progression-free survival (mPFS) of smoking mCRC patients. Although some studies have suggested that smoking may enhance the effectiveness of immunotherapy in non-small cell lung cancer patients, we did not find any analysis of overall survival (OS) in smoking mCRC patients. While the findings suggest a potential carcinogenic effect of smoking, it is important to note that they do not necessarily indicate a worse prognosis or reduced treatment efficacy. Further comprehensive analyses and prospective clinical trials will help us better understand the complexity of smoking behavior, tumor biology, and treatment outcomes. The results of this study contribute to the growing evidence for the development of more personalized approaches in the treatment of metastatic colorectal cancer[ 17 – 20 ]. Overall, our study contributes to the increasing body of evidence aimed at improving prognostication and personalizing treatment for metastatic CRC. We aim to enhance outcomes for patients grappling with this challenging disease by leveraging advancements in molecular profiling, machine learning, and clinical data analysis. Declarations Author Contribution Author contribution Concept — OY, HSK,MA. Design — OY, HSK,MA. Supervision — all authors. Data collection and/or processing — all authors . Analysis and/or interpretation — OY, HSK. Literature search — OY, AFG, MKE. Writing — OY, HSK,MA.. Critical reviews — all authors.Oğuzhan Yıldız1, Ali Fuat Gürbüz1, Melek Karakurt Eryılmaz1, Murat Araz1, Mahmut Selman Yıldırım2, Hakan Şat Bozcuk3, Mehmet Artaç1 Acknowledgement Best Regards Data Availability Ethical approval for the study was obtained from the ethics committee of Necmettin Erbakan University Medical Faculty. References Biller, L.H. and D. Schrag, Diagnosis and treatment of metastatic colorectal cancer: a review . Jama, 2021. 325(7): p. 669–685. Reinert, T., et al., Analysis of plasma cell-free DNA by ultradeep sequencing in patients with stages I to III colorectal cancer . JAMA oncology, 2019. 5(8): p. 1124–1131. Atlanta, G., American cancer society. Cancer facts and Figs. 2013 . Amer. Cancer Soc., 2013. 7. BM, W., Systemic treatment of colorectal cancer . Gastroenterology, 2008. 134: p. 1296–1310. e1. Barhak, J., Visualization and pre-processing of intensive care unit data using python data science tools. Proceedings from MODSIM World 2018, 2018. Jain, V., Everything you need to know about “Activation Functions” in Deep learning models. url : https://towardsdatascience.com/everything-you-need-to-knowabout-activation-functions-in-deep-learning-models-84ba9 f82c253. Dernière consultation le. 15: p. 02–22. Brownlee, J., What is a Confusion Matrix in Machine Learning [Electronic resource]. Access mode : https://machinelearningmastery.com/confusion-matrixmachine-learning/Chicco D., Jurman G. The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation. BMC Genomics, 2020. 21: p. 6. Serghiou, S. and K. Rough, Deep Learning for Epidemiologists: An introduction to neural networks . American Journal of Epidemiology, 2023. 192(11): p. 1904–1916. Wang, P., et al., The prognostic value of p53 positive in colorectal cancer: A retrospective cohort study . Tumor Biology, 2017. 39(5): p. 1010428317703651. Ke, G., et al., Lightgbm: A highly efficient gradient boosting decision tree . Advances in neural information processing systems, 2017. 30. Basha, S.M., D.S. Rajput, and V. Vandhan, Impact of Gradient Ascent and Boosting Algorithm in Classification . International Journal of Intelligent Engineering & Systems, 2018. 11(1). Rufo, D.D., et al., Diagnosis of diabetes mellitus using gradient boosting machine (LightGBM) . Diagnostics, 2021. 11(9): p. 1714. Munro, A., et al., Smoking compromises cause-specific survival in patients with operable colorectal cancer . Clinical Oncology, 2006. 18(6): p. 436–440. Yang, L.-P., et al., Association between cigarette smoking and colorectal cancer sidedness: A multi-center big-data platform-based analysis . Journal of Translational Medicine, 2021. 19: p. 1–11. Iacopetta, B., TP53 mutation in colorectal cancer . Human mutation, 2003. 21(3): p. 271–276. Said, R., et al., P53 mutations in advanced cancers: clinical characteristics, outcomes, and correlation between progression-free survival and bevacizumab-containing therapy . Oncotarget, 2013. 4(5): p. 705. Carr, P., et al., Lifestyle factors and risk of sporadic colorectal cancer by microsatellite instability status: a systematic review and meta-analyses . Annals of Oncology, 2018. 29(4): p. 825–834. Limsui, D., et al., Cigarette smoking and colorectal cancer risk by molecularly defined subtypes . JNCI: Journal of the National Cancer Institute, 2010. 102(14): p. 1012–1022. Colangelo, L.A., et al., Cigarette smoking and colorectal carcinoma mortality in a cohort with long-term follow‐up . Cancer, 2004. 100(2): p. 288–293. Phipps, A.I., J. Baron, and P.A. Newcomb, Prediagnostic smoking history, alcohol consumption, and colorectal cancer survival: the Seattle Colon Cancer Family Registry . Cancer, 2011. 117(21): p. 4948–4957. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4265594","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":291443072,"identity":"b3934d3d-0cde-4edf-8adc-2e9a2e9e5e92","order_by":0,"name":"Oğuzhan Yıldız","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA4ElEQVRIiWNgGAWjYDCCAxAqgYGBGcS0YGDgIV4LGxAnSJCkhceAOC18tw8wPvzaVpfHP7vn42PeHxJy/DwHGD98zMGtRfJcArOxbBtbscSds5uNeRIkjCV7G5glZ27DrcXgDAObtGQbT2LDjdxt0kAtiRvOM7Ax8xLWIpE4/0bOM+K1SH5sM0jccCOHDaLlbAN+LZJnGJuNGc4lJG68kWZsOCcN6Jeeg814/cJ3hvngwx9ldYnzbiQ/fPDGxgYYYskHP3zEo4WBgbGBmQddBJ96iJIfBJWMglEwCkbBiAYAM+RN/LKWsP0AAAAASUVORK5CYII=","orcid":"","institution":"Department of Medical Oncology, Necmettin Erbakan University School of Medicine, Konya, Turkey","correspondingAuthor":true,"prefix":"","firstName":"Oğuzhan","middleName":"","lastName":"Yıldız","suffix":""},{"id":291443073,"identity":"6cdf8797-9d55-46ca-ade8-d78e7b1c87cc","order_by":1,"name":"Ali Fuat Gürbüz","email":"","orcid":"","institution":"Department of Medical Oncology, Necmettin Erbakan University School of Medicine, Konya, Turkey","correspondingAuthor":false,"prefix":"","firstName":"Ali","middleName":"Fuat","lastName":"Gürbüz","suffix":""},{"id":291443074,"identity":"025f2a16-422d-4bc9-80cd-0a5671158be2","order_by":2,"name":"Melek Karakurt Eryılmaz","email":"","orcid":"","institution":"Department of Medical Oncology, Necmettin Erbakan University School of Medicine, Konya, Turkey","correspondingAuthor":false,"prefix":"","firstName":"Melek","middleName":"Karakurt","lastName":"Eryılmaz","suffix":""},{"id":291443076,"identity":"10151b63-78c0-4aff-80ce-17109d4c0256","order_by":3,"name":"Murat Araz","email":"","orcid":"","institution":"Department of Medical Oncology, Necmettin Erbakan University School of Medicine, Konya, Turkey","correspondingAuthor":false,"prefix":"","firstName":"Murat","middleName":"","lastName":"Araz","suffix":""},{"id":291443078,"identity":"c4bf6ac5-19fa-4141-b06b-248ceb4a5758","order_by":4,"name":"Mahmut Selman Yıldırım","email":"","orcid":"","institution":"Department of Medical Genetics, Necmettin Erbakan University School of Medicine, Konya, Turkey","correspondingAuthor":false,"prefix":"","firstName":"Mahmut","middleName":"Selman","lastName":"Yıldırım","suffix":""},{"id":291443080,"identity":"74759c6b-7142-41bc-bdef-d6d21e71a220","order_by":5,"name":"Hakan Şat Bozcuk","email":"","orcid":"","institution":"Department of Medical Oncology, Medical Park Antalya Hospital, Antalya, Turkey","correspondingAuthor":false,"prefix":"","firstName":"Hakan","middleName":"Şat","lastName":"Bozcuk","suffix":""},{"id":291443081,"identity":"d9be521a-5631-4dbb-8555-49f4ea269506","order_by":6,"name":"Mehmet Artaç","email":"","orcid":"","institution":"Department of Medical Oncology, Necmettin Erbakan University School of Medicine, Konya, Turkey","correspondingAuthor":false,"prefix":"","firstName":"Mehmet","middleName":"","lastName":"Artaç","suffix":""}],"badges":[],"createdAt":"2024-04-14 15:44:23","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4265594/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4265594/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":55063099,"identity":"5a862995-7483-45fa-b3eb-36b6f962fe35","added_by":"auto","created_at":"2024-04-22 03:03:06","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":92430,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable 1: The baseline demographic and clinical characteristics of the patients \u003c/strong\u003epr: Partial Response, cr: complete response, sd: stabledisease, progressive disease, MMR: mismatch repair, dmmr:deficient mismatch repair, pmmr: procifient mismatch repair\u003c/p\u003e","description":"","filename":"Picture1.png","url":"https://assets-eu.researchsquare.com/files/rs-4265594/v1/9f485e24f71a94a531ec0220.png"},{"id":55063098,"identity":"2bf693a1-5b57-409f-8ff4-229d6b8f6bd4","added_by":"auto","created_at":"2024-04-22 03:03:06","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":108661,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable 2: 15 machine learning models (\u003c/strong\u003eLightGBM, Ridge Regression, XGBoost, Decision Tree(dt), Gradient Boosting Classifier(gbc), Linear Discriminant Analysis (LDA), Random Forest(rf), AdaBoost(ada), Logistic Regression(lr), Naive Bayes(nb), Extra Trees(et), K-Nearest, Neighbors (knn), Dummy Classifier/Regressor, Quadratic Discriminant Analysis (qda), Support Vector Machines (svm)\u003cstrong\u003e)\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture2.png","url":"https://assets-eu.researchsquare.com/files/rs-4265594/v1/62f55df7fd68d74be0c008ff.png"},{"id":55063567,"identity":"f057ad80-e24d-43f7-899d-8b61f5da4c08","added_by":"auto","created_at":"2024-04-22 03:11:05","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":49828,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable 3: Receiver Operating Characteristic Curves for Light Gradient Boosting Machine Classifier, \u0026nbsp;\u003c/strong\u003eAUC: Area Under the Curve\u003c/p\u003e","description":"","filename":"Picture3.png","url":"https://assets-eu.researchsquare.com/files/rs-4265594/v1/65a5f6dbf225c57d42a366c8.png"},{"id":55063096,"identity":"c117e60d-6407-4f2e-9ab6-a17fe1121378","added_by":"auto","created_at":"2024-04-22 03:03:05","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":5713,"visible":true,"origin":"","legend":"\u003cp\u003eThis image is not available with this version.\u003c/p\u003e","description":"","filename":"placeholderimage.png","url":"https://assets-eu.researchsquare.com/files/rs-4265594/v1/3afc4a56855168357975b75e.png"},{"id":55063097,"identity":"df762d9e-4e71-49fd-9a7f-e262560c2b61","added_by":"auto","created_at":"2024-04-22 03:03:06","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":33390,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable5: Feature Importance Plot\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture5.png","url":"https://assets-eu.researchsquare.com/files/rs-4265594/v1/11ea5da2b13c62923c0fba9f.png"},{"id":55063094,"identity":"de0d828c-affd-482d-ba14-82d1cae8700e","added_by":"auto","created_at":"2024-04-22 03:03:05","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":55006,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable6: Correlates of Progression-Free Survival \u003c/strong\u003eECOG:Eastern Cooperative Oncology Group, MMR: mismatch repair, dmmr:deficient mismatch repair, pmmr: procifient mismatch repair\u003c/p\u003e","description":"","filename":"Picture6.png","url":"https://assets-eu.researchsquare.com/files/rs-4265594/v1/1b01e8e997105b69f2ffb91e.png"},{"id":55063100,"identity":"4045d762-1af8-45e6-bc4e-913e3b4151f7","added_by":"auto","created_at":"2024-04-22 03:03:06","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":28456,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable7: p53-\u003c/strong\u003e \u003cstrong\u003eProgression-Free Survival\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture7.png","url":"https://assets-eu.researchsquare.com/files/rs-4265594/v1/de5b20fd0c71ff508c4caf1b.png"},{"id":55063101,"identity":"b4981f1f-2e04-4c9d-9739-f07eb4420e48","added_by":"auto","created_at":"2024-04-22 03:03:06","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":32857,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable8: Cigarette- Progression-Free Survival\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture8.png","url":"https://assets-eu.researchsquare.com/files/rs-4265594/v1/bd477799caf9a567987d633f.png"},{"id":56876931,"identity":"1b94f914-4e7b-4799-93d9-e8f7f620c082","added_by":"auto","created_at":"2024-05-21 15:17:51","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":787941,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4265594/v1/38385eb9-4bff-4bfc-8562-f94542aebfd0.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Gradient Boosting Machine based prediction of chemotherapy response and role of p53 mutational and smoking status for progression free survival in metastatic colorectal cancer","fulltext":[{"header":"Introduction","content":"\u003cp\u003eColorectal cancer ranks as the third most prevalent cancer globally and is the third leading cause of cancer-related mortality. Annually, 1.3\u0026nbsp;million new cases are diagnosed, with approximately 20% of patients identified at the metastatic stage. The 5-year survival rates are reported at 14%. The cornerstone of treatment for metastatic CRC lies in systemic therapy, including cytotoxic chemotherapy, immunotherapy, biological therapy such as antibodies against cellular growth factors, and their combinations. Recent clinical studies conducted within the past five years underscore the significance of tailoring treatment regimens based on pathological and molecular characteristics. This patient cohort exhibits susceptibility to various molecular alterations that can be effectively targeted. Notably, advancements achieved over the last two decades have resulted in a significant increase in overall survival times, progressing from 10 to 20 months[\u003cspan additionalcitationids=\"CR2 CR3\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. The aim of this article is to underscore the progress made in the treatment of metastatic colorectal cancer and the significance of personalized treatment approaches.\u003c/p\u003e \u003cp\u003eThe Light Gradient Boosting Machine (LightGBM) utilizes tree-based learning algorithms and distinguishes itself with the implementation of two innovative techniques: Gradient-based One-Side Sampling (GOSS) and Exclusive Feature Bundling (EFB). GOSS selectively excludes small gradient data samples, focusing instead on large gradient data to improve information gain estimation with a reduced dataset. EFB, on the other hand, combines mutually exclusive features to effectively reduce their number. The 'Light' designation signifies its exceptional speed. LightGBM has emerged as a preferred choice due to its accelerated training speed, heightened efficiency, reduced memory utilization, superior accuracy, and adept handling of large-scale datasets [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Machine learning models are designed with the objective of discerning structures and relationships within data, particularly in domains such as data analysis and pattern recognition. These models undergo training for specific tasks and are subsequently deployed to make predictions on novel data. Machine learning models serve various purposes, including predicting future events based on historical data, identifying patterns within datasets, examining the impact of independent variables on dependent variables, utilizing feature selection and dimensionality reduction techniques to mitigate complexity in large datasets and enhance the model's generalization capability, as well as detecting anomalies by identifying deviations from normal behavior[\u003cspan additionalcitationids=\"CR6 CR7\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eOur objective is to provide an overview of the Light Gradient Boosting Machine (LightGBM) and its innovative techniques, Gradient-Based One-Side Sampling (GOSS), and Exclusive Feature Bundling (EFB), as well as to discuss the broader context of machine learning models and their applications.\u003c/p\u003e \u003cp\u003eThis study aims to assess the correlation between patient outcomes and the P53 mutation status of tumors. The influence of P53 genetic analysis on treatment modalities and patient outcomes will be examined, emphasizing its significance in guiding treatment decisions.\u003c/p\u003e"},{"header":"Material and methods","content":"\u003cp\u003eThe study investigated patients diagnosed with metastatic colorectal cancer who received treatment at the Department of Medical Oncology, Necmettin Erbakan University Faculty of Medicine Hospital between July 7, 2017, and July 1, 2023. Inclusion criteria comprised individuals aged 18 or above who had previously undergone screening for the P53 mutation and were diagnosed with metastatic colorectal cancer. Patient records retrieved from the archives were analyzed to evaluate the response to initial chemotherapy and associated factors using various machine learning models. The study employed a diverse array of machine learning models, including LightGBM, Ridge Regression, XGBoost, Decision Tree, Gradient Boosting Classifier, Linear Discriminant Analysis (LDA), Random Forest, AdaBoost, Logistic Regression, Naive Bayes, Extra Trees, K-Nearest Neighbors (KNN), Dummy Classifier/Regressor, Quadratic Discriminant Analysis (QDA), and Support Vector Machines (SVM). The most accurate model underwent further optimization, considering common clinical features as well as the status of MMR, P53, and RAS for correlation with outcomes. Feature importance and calibration plots were generated, alongside other analyses. Additionally, univariate and multivariate Cox models were developed to explore the determinants of progression-free survival (PFS). Ethical approval for the study was obtained from the ethics committee of Necmettin Erbakan University Medical Faculty.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eBaseline charecteristics\u003c/h2\u003e \u003cp\u003eThe study included a total of 101 newly diagnosed metastatic colorectal cancer patients who initiated first-line chemotherapy. Among the 101 eligible patients, 34 were categorized as P53 wild type, whereas 67 were identified as P53 mutant. The median age of the cohort was 62, with males comprising 69% of the cases. At the time of diagnosis, metastasis was observed in 59 patients, while 52 were found not to have metastasis. Table\u0026nbsp;1 presents an overview of the baseline characteristics and clinicopathological features.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWe assessed 15 machine learning models with the aim of predicting the binary outcome of the best response to chemotherapy. Among these models, LightGBM exhibited the highest baseline accuracy of 0.71 (Table\u0026nbsp;2).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe LightGBM model was subsequently fine-tuned, leading to an enhanced accuracy figure of 0.79 and a macro average AUC value of 0.82 (Table\u0026nbsp;3).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe feature importance plot assessed various factors, including age at diagnosis, maximum metastatic dimension, metastatic status, number of metastatic systems, gender, comorbidity, metastatic involvement limited to the liver, cumulative cigarette package years, comorbidity count, MMR status, RAS positivity, tumor location, ECOG performance status, P53 mutation status, and metastatic status at diagnosis. According to the plot, the three most significant features were age at diagnosis, maximum metastatic dimension, and metastatic status at diagnosis, with variable importance scores of 148, 90, and 58, respectively (Table\u0026nbsp;4).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe study examined the correlates of progression-free survival (PFS), considering factors such as age, sex, ECOG performance status, comorbidity, number of comorbidities, cumulative smoking pack years, metastatic status at diagnosis, cancer location, maximum metastatic dimension, number of metastatic organ systems, presence of only liver metastases, RAS positivity, P53 mutation status, and MMR status (Table\u0026nbsp;5).\u003c/p\u003e \u003cp\u003eGenetic variables did not demonstrate significant feature importance for response analysis. However, in survival analysis, progression-free survival (PFS) was found to be associated with p53 mutation status (Exp(B)\u0026thinsp;=\u0026thinsp;0.52, Wald\u0026thinsp;=\u0026thinsp;6.98, P\u0026thinsp;=\u0026thinsp;0.008) and smoking pack years (Exp(B)\u0026thinsp;=\u0026thinsp;0.99, Wald\u0026thinsp;=\u0026thinsp;4.28, P\u0026thinsp;=\u0026thinsp;0.039). (Table\u0026nbsp;6)\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe median progression-free survival (mPFS) was 11.6 months (7.7\u0026ndash;15.4 months) across all groups. Specifically, the mPFS was 16.6 months (14.1\u0026ndash;19.1 months) in the P53 mutant group, while it was 10 months (9.5\u0026ndash;10.6 months) in the P53 wild-type group (Table\u0026nbsp;7).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe median progression-free survival (mPFS) was 17.6 months (14.8\u0026ndash;20.4 months) in the group of smokers, whereas it was 10.7 months (9.1\u0026ndash;12.3 months) in the group of non-smokers (Table\u0026nbsp;8).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eGlobally, there has been a significant increase in the incidence of colorectal cancer (CRC) due to lifestyle changes. Although there have been advancements in comprehensive treatments for CRC, such as surgery, chemotherapy, and radiotherapy, the 5-year overall survival (OS) and disease-free survival (DFS) rates remain at approximately 45% and 40%, respectively. This highlights the continued significance of CRC as a major public health concern. The importance of identifying new prognostic factors is emphasized by the heterogeneity of tumors. This enables the development of precise treatment strategies that target unfavorable prognostic indicators at the molecular level[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eMachine learning algorithms, including advanced techniques such as LightGBM, play a crucial role in cancer treatment. For instance, Derara Ruba Rufo et al. utilized the LightGBM model to accurately diagnose early diabetes mellitus with an AUC (Area Under the Curve) of 98.1%. Our study demonstrates the effectiveness of LightGBM in predicting first-line chemotherapy response in metastatic CRC patients, providing a valuable tool to assist in treatment selection and optimization. Additionally, the study contributes to the growing body of evidence aimed at improving prognostication and personalizing treatment. Moreover, it may suggest alternative treatment options for non-responders. Future research should focus on conducting prospective clinical trials to improve the accuracy of prediction models and understand the molecular mechanisms underlying treatment resistance. We aim to enhance outcomes for patients grappling with this challenging disease by leveraging advancements in molecular profiling, machine learning, and clinical data analysis.[\u003cspan additionalcitationids=\"CR11\" citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]\u003c/p\u003e \u003cp\u003eTP53 variations are present in 43.2% of colorectal cancers. As a transcription factor, p53 regulates various cellular responses, including DNA repair, cell cycle control, senescence, metabolism, and cell death. The frequent mutations of p53 pose challenges to therapeutic targeting, as standard chemotherapies rely on wild-type p53 function to trigger apoptosis and arrest the cell cycle. The study reported a notably prolonged median progression-free survival (mPFS) of 16.6 months in patients with P53 mutations compared to those with P53 wild-type status, which was 10 months. In 2003, Barry Iacopetta's meta-analysis revealed a potential association between p53 mutation and poorer survival in the majority of analyzed studies, although some studies contradicted this, reporting no association or even better outcomes. Rabih et al. observed an extended median progression-free survival (mPFS) with bevacizumab treatment in patients with P53 mutations. In contrast, Peng Wang et al. demonstrated significantly shorter disease-free survival and overall survival in patients with p53-positive metastatic colorectal cancer compared to those who were p53-negative. According to AJ Munro et al.'s extensive review, abnormal p53 did not affect outcomes in patients receiving chemotherapy. However, Kyoung Min Kim et al.'s study on metastatic colorectal cancer patients indicated shorter overall survival in the p53 mutation group, although there were no significant effects on recurrence-free survival. These findings highlight the varying impact of P53 mutations on the prognosis and treatment response of colorectal cancer patients in different studies and clinical contexts.[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan additionalcitationids=\"CR14 CR15\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eStudies conducted by Calengeo et al. and Phipss AI et al. have reported a significant association between smoking and increased colorectal cancer (CRC)-related mortality among patients with metastatic colorectal cancer (mCRC). It is important to consider these findings when discussing the potential risks of smoking with patients who have been diagnosed with mCRC. Additionally, Munro AJ et al. demonstrated that current smokers face more than twice the risk of CRC-related death compared to former and never smokers. However, according to a meta-analysis by Yang, Lu-Ping, and colleagues, a history of smoking did not significantly affect the prognosis of mCRC, despite previous findings suggesting otherwise. Furthermore, while left-sided tumor location has been identified as a positive prognostic indicator in mCRC, a meta-analysis of a large cohort of Chinese colorectal cancer patients suggested a potential extension of overall survival (OS) in smoking patients with right-sided colon tumors. These results suggest that the prognosis in mCRC may be influenced by factors such as tumor location and patient demographics, which may affect the relationship between smoking and prognosis.\u003c/p\u003e \u003cp\u003eIn addition, our study found that first-line chemotherapy treatment significantly extended the median progression-free survival (mPFS) of smoking mCRC patients. Although some studies have suggested that smoking may enhance the effectiveness of immunotherapy in non-small cell lung cancer patients, we did not find any analysis of overall survival (OS) in smoking mCRC patients. While the findings suggest a potential carcinogenic effect of smoking, it is important to note that they do not necessarily indicate a worse prognosis or reduced treatment efficacy. Further comprehensive analyses and prospective clinical trials will help us better understand the complexity of smoking behavior, tumor biology, and treatment outcomes. The results of this study contribute to the growing evidence for the development of more personalized approaches in the treatment of metastatic colorectal cancer[\u003cspan additionalcitationids=\"CR18 CR19\" citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eOverall, our study contributes to the increasing body of evidence aimed at improving prognostication and personalizing treatment for metastatic CRC. We aim to enhance outcomes for patients grappling with this challenging disease by leveraging advancements in molecular profiling, machine learning, and clinical data analysis.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eAuthor contribution Concept \u0026mdash; OY, HSK,MA. Design \u0026mdash; OY, HSK,MA. Supervision \u0026mdash; all authors. Data collection and/or processing \u0026mdash; all authors . Analysis and/or interpretation \u0026mdash; OY, HSK. Literature search \u0026mdash; OY, AFG, MKE. Writing \u0026mdash; OY, HSK,MA.. Critical reviews \u0026mdash; all authors.Oğuzhan Yıldız1, Ali Fuat G\u0026uuml;rb\u0026uuml;z1, Melek Karakurt Eryılmaz1, Murat Araz1, Mahmut Selman Yıldırım2, Hakan Şat Bozcuk3, Mehmet Arta\u0026ccedil;1\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eBest Regards\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eEthical approval for the study was obtained from the ethics committee of Necmettin Erbakan University Medical Faculty.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eBiller, L.H. and D. Schrag, \u003cem\u003eDiagnosis and treatment of metastatic colorectal cancer: a review\u003c/em\u003e. Jama, 2021. 325(7): p. 669\u0026ndash;685.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eReinert, T., et al., \u003cem\u003eAnalysis of plasma cell-free DNA by ultradeep sequencing in patients with stages I to III colorectal cancer\u003c/em\u003e. JAMA oncology, 2019. 5(8): p. 1124\u0026ndash;1131.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAtlanta, G., \u003cem\u003eAmerican cancer society. Cancer facts and Figs. 2013\u003c/em\u003e. Amer. Cancer Soc., 2013. 7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBM, W., \u003cem\u003eSystemic treatment of colorectal cancer\u003c/em\u003e. Gastroenterology, 2008. 134: p. 1296\u0026ndash;1310. e1.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarhak, J., \u003cem\u003eVisualization and pre-processing of intensive care unit data using python data science tools.\u003c/em\u003e Proceedings from MODSIM World 2018, 2018.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJain, V., \u003cem\u003eEverything you need to know about \u0026ldquo;Activation Functions\u0026rdquo; in Deep learning models. url\u003c/em\u003e: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://towardsdatascience.com/everything-you-need-to-knowabout-activation-functions-in-deep-learning-models-84ba9\u003c/span\u003e\u003cspan address=\"https://towardsdatascience.com/everything-you-need-to-knowabout-activation-functions-in-deep-learning-models-84ba9\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003cem\u003ef82c253.\u003c/em\u003e Derni\u0026egrave;re consultation le. 15: p. 02\u0026ndash;22.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrownlee, J., \u003cem\u003eWhat is a Confusion Matrix in Machine Learning [Electronic resource]. Access mode\u003c/em\u003e: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://machinelearningmastery.com/confusion-matrixmachine-learning/Chicco\u003c/span\u003e\u003cspan address=\"https://machinelearningmastery.com/confusion-matrixmachine-learning/Chicco\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e \u003cem\u003eD., Jurman G. The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation.\u003c/em\u003e BMC Genomics, 2020. 21: p. 6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSerghiou, S. and K. Rough, \u003cem\u003eDeep Learning for Epidemiologists: An introduction to neural networks\u003c/em\u003e. American Journal of Epidemiology, 2023. 192(11): p. 1904\u0026ndash;1916.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, P., et al., \u003cem\u003eThe prognostic value of p53 positive in colorectal cancer: A retrospective cohort study\u003c/em\u003e. Tumor Biology, 2017. 39(5): p. 1010428317703651.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKe, G., et al., \u003cem\u003eLightgbm: A highly efficient gradient boosting decision tree\u003c/em\u003e. Advances in neural information processing systems, 2017. 30.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBasha, S.M., D.S. Rajput, and V. Vandhan, \u003cem\u003eImpact of Gradient Ascent and Boosting Algorithm in Classification\u003c/em\u003e. International Journal of Intelligent Engineering \u0026amp; Systems, 2018. 11(1).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRufo, D.D., et al., \u003cem\u003eDiagnosis of diabetes mellitus using gradient boosting machine (LightGBM)\u003c/em\u003e. Diagnostics, 2021. 11(9): p. 1714.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMunro, A., et al., \u003cem\u003eSmoking compromises cause-specific survival in patients with operable colorectal cancer\u003c/em\u003e. Clinical Oncology, 2006. 18(6): p. 436\u0026ndash;440.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang, L.-P., et al., \u003cem\u003eAssociation between cigarette smoking and colorectal cancer sidedness: A multi-center big-data platform-based analysis\u003c/em\u003e. Journal of Translational Medicine, 2021. 19: p. 1\u0026ndash;11.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIacopetta, B., \u003cem\u003eTP53 mutation in colorectal cancer\u003c/em\u003e. Human mutation, 2003. 21(3): p. 271\u0026ndash;276.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaid, R., et al., \u003cem\u003eP53 mutations in advanced cancers: clinical characteristics, outcomes, and correlation between progression-free survival and bevacizumab-containing therapy\u003c/em\u003e. Oncotarget, 2013. 4(5): p. 705.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCarr, P., et al., \u003cem\u003eLifestyle factors and risk of sporadic colorectal cancer by microsatellite instability status: a systematic review and meta-analyses\u003c/em\u003e. Annals of Oncology, 2018. 29(4): p. 825\u0026ndash;834.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLimsui, D., et al., \u003cem\u003eCigarette smoking and colorectal cancer risk by molecularly defined subtypes\u003c/em\u003e. JNCI: Journal of the National Cancer Institute, 2010. 102(14): p. 1012\u0026ndash;1022.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eColangelo, L.A., et al., \u003cem\u003eCigarette smoking and colorectal carcinoma mortality in a cohort with long-term follow‐up\u003c/em\u003e. Cancer, 2004. 100(2): p. 288\u0026ndash;293.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePhipps, A.I., J. Baron, and P.A. Newcomb, \u003cem\u003ePrediagnostic smoking history, alcohol consumption, and colorectal cancer survival: the Seattle Colon Cancer Family Registry\u003c/em\u003e. Cancer, 2011. 117(21): p. 4948\u0026ndash;4957.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"metastatic colorectal cancer, smoking status, p53 mutation, progression-free survival","lastPublishedDoi":"10.21203/rs.3.rs-4265594/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4265594/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground:\u003c/strong\u003e Identifying predictors of response or progression after first-line chemotherapy for stage 4 colorectal cancer remains a challenge. This study aims to evaluate the correlation between patient outcomes and the p53 mutational status and smoking status of tumors using various machine learning methods.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMaterial and methods:\u003c/strong\u003e We consecutively recruited all patients diagnosed with metastatic colorectal cancer at an academic center within a specified time period. Response to first-line chemotherapy and associated factors were assessed using various machine learning models. The most accurate model was further optimized. Additionally, common clinical features, MMR, p53, and RAS status were tested for correlation with the outcome. Feature importance and calibration plots were generated, and univariate and multivariate Cox models were utilized to analyze associates of progression-free survival (PFS).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults:\u003c/strong\u003e A total of 101 newly diagnosed metastatic colorectal cancer patients initiating first-line chemotherapy were included. The median age was 62, and 69% of the cases were male. We evaluated 15 machine learning models to predict the binary outcome of best response to chemotherapy, among which LightGBM demonstrated the highest baseline accuracy of 0.71. Further tuning of the LightGBM model improved accuracy to 0.79, with a macro average AUC value of 0.82. Age at diagnosis, maximum metastatic dimension of cancer, and metastatic status at diagnosis were identified as the three most important features. Genetic variables did not establish significant feature importance for response analysis. Survival analysis revealed an association between PFS and p53 mutation status (Exp(B) = 0.52, Wald = 6.98, P = 0.008) and smoking pack years (Exp(B) = 0.99, Wald = 4.28, P = 0.039).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDiscussion:\u003c/strong\u003e Utilizing LightGBM as a machine learning method, we developed a predictive model with good accuracy for assessing response to first-line treatment. If confirmed and further improved, such a model could aid in identifying responders to first-line chemotherapy in metastatic colorectal cancer patients and suggesting alternative chemotherapy options for non-responders. Furthermore, our findings highlight the prognostic importance of genetic features, particularly p53 mutation status, and smoking pack years for PFS duration in this context.\u003c/p\u003e","manuscriptTitle":"Gradient Boosting Machine based prediction of chemotherapy response and role of p53 mutational and smoking status for progression free survival in metastatic colorectal cancer","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-04-22 03:03:01","doi":"10.21203/rs.3.rs-4265594/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"3fcd40f4-3465-417e-9a4e-19cbecea3d29","owner":[],"postedDate":"April 22nd, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-11-04T00:38:06+00:00","versionOfRecord":[],"versionCreatedAt":"2024-04-22 03:03:01","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4265594","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4265594","identity":"rs-4265594","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}