Application of parametric Accelerated Failure Time (AFT) model in early stage breast cancer patients

Abstract

ABSTRACT Purpose: Cox Proportional Hazard (CPH) model is the most commonly used multivariate regression model for survival analysis. However, it is not always possible to obtain the proportional hazard (PH) assumption. In this case, Parametric Accelerated Failure Time (AFT) models may be applied. In this study, AFT and CPH models were applied to patients with early stage breast cancer and the results were compared. Materials and Methods: Retrospective survival data of 697 patients with early stage breast cancer were analyzed in this study. 13 independent variables and overall survival time as the dependent variable were tested. Multiple CPH regression analysis were performed for significant variables . For the AFT model, hazard functions of the data belonging to log-normal, log-logistic, Weibull and of exponential distributions were examined. Results: Although age groups, tumor grade, neural invasion and extra capsule involvement did not provide the assumption of COH, when statistically significant 9 independent variables were applied to the multivariate COX model, metastatic lymph nodes and menopausal status were found to be significant. According to AIC value and hazard function distributions, the most appropriate AFT model was with log-logistic. AFT model with log-logistic regression, number of metastatic lymph nodes, menopausal status and tumor size were found to be significant. Conclusion: In the literature, the CPH model is one of the most commonly used survival models. In cases where the assumption of the proportionality of hazards were violated; it may be more appropriate to use alternative models such as the AFT regression model.

Keywords

• Survival Analysis
• Cox Proportional Hazard Model
• Accelerated Failure Time (AFT) Model
• Parametric Proportional Model
• log-logistic

Parametrik hızlandırılmış başarısızlık (ölüm) zamanı (HBZ) modelleri ile erken evre meme kanseri hastalarında uygulama

Abstract

ÖZET Amaç: Cox Orantılı Hazard (COH) modeli sağkalım analizinde en sık kullanılan çok değişkenli regresyon modelidir. Ancak uygulamada orantılı hazard (OH) varsayımını sağlamak her zaman mümkün olmamaktadır. Bu durumda Parametrik Hızlandırılmış Başarısızlık (Ölüm) Zamanı (HBZ) modelleri uygulanır. Bu çalışmada HBZ ve COH modelleri meme kanseri tanısı olan hastalara uygulanarak sonuçlar karşılaştırılmıştır. Gereç ve Yöntem: Bu çalışmada 697 erken evre meme kanserli hastanın retrospektif sağkalım verileri analiz edilmiştir. Bağımlı değişken olan genel sağkalım ve 13 bağımsız değişkenle test edilmiştir. Anlamlı çıkan değişkenler çoklu regresyon analizi olan COH modeli ile yapılan regresyonla tahmin edilmeye çalışılmıştır. HBZ modeli için verilerin olası log-normal, log-lojistik, Weibull ve üstel dağılımlarına ait Hazard Fonksiyonlarına bakılmıştır. Bulgular: Yaş grupları, tümör derecesi, nöral invazyon ve ekstra kapsül tutulumu COH varsayımını sağlamamasına rağmen, istatistiksel olarak anlamlı çıkan 9 bağımsız değişken çok değişkenli COX modeline girdiğinde sadece metastatik lenf nodu sayısı ve menopoz değişkenleri anlamlı bulunmuştur. HBZ modeli için, bağımlı değişken olan genel sağkalıma ait, AIC değerine ve hazard fonksiyonu dağılımlarına göre en uygun HBZ modelinin Log-lojistik olduğu görülmüştür. Log-lojistik regresyonu sonucunda ise metastatik lenf nodu sayısı, menopoz ve tümör boyutu değişkenleri anlamlı çıkmıştır. Sonuç: Literatürde COH modeli en sık kullanılan sağkalım modellerinin başında gelir. Hazardların orantılı olmadığı durumlarda COH regresyon modeli yerine HBZ regresyon modeli gibi modellerin kullanımı daha uygun olabilir.

Keywords

• Sağkalım Analizi
• Cox Orantılı Hazard Modeli
• Parametrik Hızlandırılmış Başarısızlık (Ölüm) Zamanı (HBZ)
• Parametrik Orantılı Modeli
• log-lojistik

References

1. KAYNAKLAR 1) Cox DR. Regression Models and Life-Tables. Breakthroughs in statistics. Springer Series in Statitstics, New York, NY, 1992. 527-541. https://doi.org/10.1007/978-1-4612-4380-9_37
2. 2) Lee ET and Go OT. Survival Analysis in Public Health Research. Annual Review of Public Health, 1997, 18.1, 105-134. https://doi.org/10.1146/annurev.publhealth.18.1.105
3. 3) Karimi A, Delpisheh A, Sayehmiri K. Application of accelareted failure time models for breast cancer patients’ survival in Kurdistan Province of Iran. Journal of Cancer Research and Therapeutics, 2016, 12(3), 1184-1148. doi: 10.4103/0973-1482.168966
4. 4) Keene ON. Alternatives to the Hazard Ratio in Summarizing Efficacy in Time‐to‐Event Studies: An Example from Influenza Trials. Statistics in Medicine, 2002, 21(23), 3687-3700. doi:10.1002/sim.1312
5. 5) Khanal SP, Sreenivas V, Acharya SK. Accelarated Failure Time Models: An Application in The Survival of Acute Liver Failure Patients İn India. International Journal of Sciences and Research. 2014, 31(6), 161-166 ISSN:2319-7064:161-166.
6. 6) Bakhshi E, et al. Survival Analysis of Thalassemia Major Patients Using Cox, Gompertz Proportional Hazard and Weibull Accelerated Failure Time Models. Medical Journal of the Islamic Republic of Iran, 2017, 31, 97. doi:10.14196/mjiri.31.97
7. 7) Ghadimi et al. Family History of the Cancer on the survival of the Patients with Gastrointestinal Cancer in Northern Iran, Using Frailty Models. BioMed Central Gastroenterology 2011, 1, 104
8. 8) Brandon G, Seals S, Aban I; Survival Analysis and regression models J Nucl Cardiol. 2014 Aug;21(4):686-94. doi: 10.1007/s12350-014-9908-2

9. 9) Wei LJ. The Accelerated Failure Time Model: A Useful Alternative To The Cox Regression Model in Survival Analysis. Statistics in Medicine, 1992, 11(14-15), 1871-1879. doi: 10.1002/sim.4780111409
10. 10) Pourhoseingholi MA, et al. Comparing Cox Regression and Parametric Models for Survival of Patients with Gastric Carcinoma. Asian Pacific Journal Cancer Prevention, 2007, 8, 412-416
11. 11) Swindell WR. Accelerated Failure Time Models Provide a Useful Statistical Framework for Aging Research." Experimental Gerontology, 2009, 44(3), 190-200. doi: 10.1016/j.exger.2008.10.005
12. 12) Kwong GPS and Hutton JL. Choice of Parametric Models in Survival Analysis: Applications to Monotherapy for Epilepsy and Cerebral Palsy. Journal of the Royal Statistical Society: Series C (Applied Statistics), 2003, 52(2), 153-168.
13. 13) Seyoum D, et al. Risk Factors for Mortality among Adult HIV/AIDS Patients Following Antiretroviral Therapy in Southwestern Ethiopia: An Assessment through Survival Model. International Journal of Environmental Research and Public Health, 2017, 14, 296. doi:10.3390/ijerph14030296
14. 14) Teshnizi SH and Ayatollahi SMT. Comparison of Cox Regression and Parametric Models: Application for Assessment of Survival of Pediatric Cases of Acute Leukemia in Southern Iran. Asian Pacific Journal of Cancer Prevention, 2017, 18(4), 981 – 985. doi:10.22034/APJCP.2017.18.4.981
15. 15) Bradburn MJ, Clark TG, Love SB, Altman DG. Survival Analysis Part II: Multivariate Data Analysis – An Introduction to Concepts and Methods. British Journal of Cancer, 2003, 89, 431 – 436. doi:10.1038/sj.bjc.6601119
16. 16) Akaikei H. "Information Theory and An Extension of Maximum Likelihood Principle." Proc. 2nd Int. Symp. on Information Theory. 1973.
17. 17) Vaida F, Blanchard S. Conditional Akaike Information for Mixed-Effects Models. Biometrika, 2005, 92(2), 351-370.
18. 18) Vahedi M et al. What is the best parametric survival models for analyzing hemodialysis data? Global Journal of Health Science. 2016. Vol. 8, No. 10; 2016. doi: 10.5539/gjhs.v8n10p118.
19. 19) Bekiroğlu N, Cebeci IA, Hüseyin A, Acar M; Parametrik Hızlandırılmış Başarısızlık (ölüm) Zamanı (HBZ) Modelleri ile Akciğer Kanseri Hastalarında Uygulama (Poster sunumu). XX. Ulusal ve III. Uluslararası Biyoistatistik Kongresi. 26-29 Ekim 2018, Gaziantep.
20. 20) Global Burden of Disease Cancer Collaboration. Fitzmaurice C, Allen C, Barber RM, Barregard L, Bhutta ZA, et al. Global, regional, and national cancer incidence, mortality, years of life lost, years lived with disability, and disability-adjusted life-years for 32 cancer groups, 1990 to 2015: A systematic analysis for the global burden of disease study. JAMA Oncol. 2017;3:524–48. doi: 10.1001/jamaoncol.2016.5688
21. 21) Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2018;68:394–424. doi: 10.3322/caac.21492
22. 22) Fitzmaurice C, Allen C, Barber R, et al. Global, regional, and national cancer incidence, mortality, years of life lost, years lived with disability, and disability-adjusted life-years for 29 cancer groups, 1990 to 2016: A systematic analysis for the global burden of disease study. JAMA Oncol. 2017;3:524–48. doi: 10.1001/jamaoncol.2016.5688
23. 23) Kasza J, Wraith D., Lamb K, Wolfe R; Survival analysis of time‐to‐event data in respiratory health research studies. Asian Pacific Society of Respirology. 2014. 19, 483–492doi: 10.1111/resp.12281.
24. 24) Rubio FJ, Remontet L, Jewell NP, Belot A; On a general structure for hazard-based regression models: An application to population-based cancer research. Stat Methods Med Res. 2019. Volume: 28 issue: 8, page(s): 2404-2417. doi:10.1177/0962280218782293.
25. 25) Ghorbani N, Yazdani-Charati J, Anvari K, Ghorbani N; Application of Weibull Accelerated Failure Time Model on the Disease-free Survival Rate of Breast Cancer. Iranian Journal of Health Sciences 2016; 4(2): 11-18. http://jhs.mazums.ac.ir.
26. 26) Iraji Z, Koshki TJ, Dolatkhah R, Jafarabadi MA; Parametric survival model to identify the predictors of breast cancer mortality: An accelerated failure time approach. J Res Med Sci. 2020 Apr 13;25:38. doi: 10.4103/jrms.JRMS_743_19
27. 27) Alfonso AG, de Oca NA. Application of hazard models for patients with breast cancer in Cuba. Int J of Clin Exp Med. 2011; 4:148-56.
28. 28) Khan HM, Saxena A, Gabbidon K, Stewart TS, Bhatt C. Survival analysis for white non-Hispanic female breast cancer patients. Asian Pac J Cancer Prev. 2014; 15:4049-54.