Abstract

Identifying patient subgroups with heterogeneous clinical outcomes and varying treatment-biomarker interactions is crucial in clinical research. These two types of subgroups are referred to as subgroup analysis (SA) and homogeneity fusion (HF), respectively. Unlike existing methods that conduct SA or HF separately, simultaneously identifying two-way subgroup structures improves estimation and clustering accuracy, particularly in the presence of cluster heterogeneity and weak signals across high-dimensional biomarkers. In this paper, we propose a novel Double Centre-Augmented Penalties Fixed Effect Model(DCAF) for the simultaneous identification of two-way subgroups based on clustered survival data. Instead of conventional pairwise penalties, we introduce a differentiable centre-augmented harmonic-type penalty to facilitate subgroup identification. We establish theoretical results on estimation and selection consistency. Numerical studies, including simulations and two real data examples, demonstrate the effectiveness of the proposed DCAF method in accurately identifying two-way subgroups, as well as in selecting important biomarkers for each subgroup.

Key words and phrases: Double-centre-augmented, Subgroup analysis, Homogeneity fusion, High dimensional biomarkers, Theoretical property

Information

Preprint No.SS-2025-0230
Manuscript IDSS-2025-0230
Complete AuthorsYe He, Dongsheng Tu, Liu Liu, Ling Zhou
Corresponding AuthorsLing Zhou
Emailszhouling@swufe.edu.cn

References

  1. Ballman, K. (2015). Biomarker: Predictive or prognostic? J Clin Oncol, 33(33):3968–3971.
  2. Boyd, S., Parikh, N., Chu, E., Peleato, B., and Eckstein, J. (2010). Distributed optimization and statistical learning via the alternating direction method of multipliers. Foundations and Trends in Machine Learning, 3(1):1–122.
  3. Breheny, P. and Huang, J. (2011). Coordinate descent algorithms for nonconvex penalized regression, with applications to biological feature selection. Ann Appl Stat, 5(1):232–253.
  4. Caroni, C., Crowder, M., and Kimber, A. (2010). Proportional hazards models with discrete frailty. Lifetime Data Analysis, 16:374–384.
  5. Chen, E, X., Jonker, D. J., Loree, J. M., Kennecke, H. F., Berry, S. R., Couture, F., et al.
  6. (2020). Effect of Combined Immune Checkpoint Inhibition vs Best Supportive Care Alone in Patients With Advanced Colorectal Cancer: The Canadian Cancer Trials Group CO.26 Study. Jama Oncol, 6(6):831–838.
  7. Cheng, X., Lu, W., and Liu, M. (2015). Identification of homogeneous and heterogeneous variables in pooled cohort studies. Biometrics, 71(2):397–403.
  8. Cooper, W. A. and Tan, P. H. (2024). Predictive and prognostic biomarkers in solid tumours. Pathology, 56(2):145–146.
  9. Eng, K. H. and Hanlon, B. M. (2014). Discrete mixture modeling to address genetic heterogeneity in time-to-event regression. Bioinformatics, 30(12):1690–1697.
  10. Fan, J. and Lv, J. (2008). Sure independence screening for ultrahigh dimensional feature space (with discussion). J R Stat Soc Ser B (Stat Methodol), 70:849–911.
  11. Fei, Z., Zheng, Q., Hong, H. G., and Li, Y. (2021). Inference for high-dimensional censored quantile regression. J Am Stat Assoc, 118(542):898–912.
  12. Gasperoni, F., Ieva, F., Paganoni, A. M., Jackson, C., and Sharples, L. (2020). Non-parametric frailty cox models for hierarchical time-to-event data. Biostatistics, 21(3):531–544.
  13. Guo, G. and Rodrlguez, G. (1992). Estimating a multivariate proportional hazards model for clustered data using the em algorithm, with an application to child survival in guatemala. J Am Stat Assoc, 87(420):969–976.
  14. Han, M., Lin, Y., Liu, W., and Wang, Z. (2024). Robust inference for subgroup analysis with general transformation models. Journal of Statistical Planning and Inference, 229.
  15. He, Y., Zhou, L., Xia, Y., and Lin, H. (2023). Center-augmented ℓ2-type regularization for subgroup learning. Biometrics, 79(3):2157–2170.
  16. Heckmanb, J. and Singer, B. (1984). A method for minimizing the impact of distributional assumptions in econometric models for duration data. Econometrica, 52(2):271–320.
  17. Hougaard, P. (1986). A class of multivariate failure time distributions. Biometrika, 73(3):671– 678.
  18. Hu, X., Huang, J., Liu, L., Sun, D., and Zhao, X. (2021). Subgroup analysis in the heterogeneous Cox model. Stat Med, 40(3):739–757.
  19. Ke, Z. T., Fan, J., and Wu, Y. (2015). Homogeneity pursuit. J Am Stat Assoc, 110(509):175– 194.
  20. Lange, K. (2016). MM optimization algorithm. SIAM.
  21. Legrand, C., Duchateau, L., Sylvester, R., Janssen, P., van der Hage, J. A., van de Velde, C. J.,
  22. et al. (2006). Heterogeneity in disease free survival between centers: lessons learned from an EORTC breast cancer trial. Clin Trials, 3(1):10–18.
  23. Liu, L., He, K., Wang, D., Ma, S., Qu, A., Luan, Y., et al. (2024). Health care provider clustering using fusion penalty in quasi-likelihood. Biometrical J, 66(2).
  24. Ma, S. and Huang, J. (2017). A concave pairwise fusion approach to subgroup analysis. J Am Stat Assoc, 112(517):410–423.
  25. Moore, M. J., Goldstein, D., Hamm, J., Figer, A., Hecht, J. R., Gallinger, S., et al. (2007). Erlotinib plus gemcitabine compared with gemcitabine alone in patients with advanced pancreatic cancer: a phase III trial of the National Cancer Institute of Canada Clinical Trials Group. J Clin Oncol, 25(15):1960–1966.
  26. Peikari, M., Salama, S., Nofech-Mozes, S., and Martel, A. L. (2018). A cluster-then-label semisupervised learning approach for pathology image classification. Sci Rep-uk, 8(1):7193.
  27. Rand, W. M. (1971). Objective criteria for the evaluation of clustering methods. J Am Stat Assoc, 66(336):846–850.
  28. Robins, J. M. and Rotnitzky, A. (1992). Recovery of information and adjustment for dependent censoring using surrogate markers. pages 297–331. Birkh¨auser, Boston, MA.
  29. Shen, X. and Huang, H. C. (2010). Grouping pursuit through a regularization solution surface. J Am Stat Assoc, 105(490):727–739.
  30. Shultz, D., Pai, J., Chiu, W., Ng, K., Hellendag, M., Heestand, G., et al. (2016). A novel biomarker panel examining response to gemcitabine with or without erlotinib for pancreatic cancer therapy in ncic clinical trials group pa.3. PLoS One, 11(1).
  31. Tang, X., Xue, F., and Qu, A. (2020). Individualized multi-directional variable selection. J Am Stat Assoc, 116(535):1280–1296.
  32. Therneau, T. M., Grambsch, P. M., and Pankratz, V. S. (2003). Penalized survival models and frailty. J Comput Graph Stat, 12(1):156–175.
  33. Tibshirani, R., Saunders, M., Rosset, S., Zhu, J., and Knight, K. (2005). Sparsity and smoothness via the fused lasso. J R Stat Soc Ser B (Stat Methodol), 67(1):91–108.
  34. Vaupel, J. W., Manton, K. G., and Stallard, M. E. (1979). The impact of heterogeneity in individual frailty on the dynamics of mortality. Demography, 16(3):439–454.
  35. Werner, E., Clark, J. N., Hepburn, A., Bhamber, R. S., Ambler, M., Bourdeaux, C. P., et al.
  36. (2023). Explainable hierarchical clustering for patient subtyping and risk prediction. Exp Biol Med, 248(244):2457–2559.
  37. Witten, D. M., Shojaie, A., and Zhang, F. (2014). The cluster elastic net for high-dimensional regression with unknown variable grouping. Technometrics, 56(1):112–122.
  38. Xu, J. and Lange, K. (2019). Power k-means clustering. Proceedings of Machine Learning Research, 97:6921–6931.
  39. You, N., He, S., Wang, X., Zhu, J., and Zhang, H. (2018). Subtype classification and heterogeneous prognosis model construction in precision medicine. Biometrics, 74(3):814–822.
  40. Zhang, B., Hsu, M., and Dayal, U. (1999). K-harmonic means - a data clustering algorithm. Hewlett-Packard Technical Report.
  41. Zhang, J., He, X., and Wang, J. H. (2021). Directed community detection with network embedding. J Am Stat Assoc, 117(540):1809–1819.
  42. Zhu, Y., Shen, X., and Pan, W. (2013). Simultaneous grouping pursuit and feature selection over an undirected graph. J Am Stat Assoc, 108(502):713–725.
  43. Zou, H. (2006). The adaptive lasso and its oracle properties. J Am Stat Assoc, 101(476):1418– 1429.

Acknowledgments

The authors gratefully acknowledge National Key R&D Program of China (No. 2022YFA1003702), National Natural Science Foundation of China (No. 12426309) and Guanghua Talent Project of SWUFE, Natural Sciences and Engineering Research Council of Canada, and Sichuan Sichuan Science and Technology Program (Grant No. 2024NSFSC0040, 2024ZYD0115).

Supplementary Materials

The proofs and more simulation results are presented in the Supp.

Supplementary materials are available for download.