TO THE EDITOR:
I read with interest Benkhoff et al’s1 recent article on thromboinflammation as a determinant of major adverse cardiac events (MACE) after ST-elevation myocardial infarction, which highlights the potential of this biomarker for guiding precision medicine in this patient population.
Although the study provides valuable insights, there is room for potential refinements to the statistical analysis, particularly regarding the univariate and multivariate logit models. Given the continuous nature of several of the determinants (ie, age, body mass index, total cholesterol, high-density lipoprotein, low-density lipoprotein, glomerular filtration rate, C-reactive protein, and troponin T), generalized additive logit models (GALMs) could offer a more flexible and nuanced approach to examining their effects on MACE.2,3 Unlike linear logit models, GALMs allow for the estimation of data-driven, nonlinear relationships between determinants and the log odds of the outcome, which may better capture the complexities of these effects.
In future work, it is recommended to use GALMs to identify inflection points (sharp bends) in curvilinear relationships between these determinants and MACE, as has been done in some recent studies on other health conditions.4,5 Although not yet part of everyday clinical practice, such inflection points may provide valuable insights into optimal treatment thresholds and facilitate the targeting of interventions. For instance, if certain levels of glomerular filtration rate and C-reactive protein are associated with sharp increases in MACE risk, it might be suggested that clinicians prioritize more aggressive interventions for patients exceeding these thresholds. Furthermore, although the authors' identification of an optimal cutpoint for H3Cit-DNA of >219.3 μg/L based on receiver operating characteristic analysis is a valuable contribution, applying GALMs to the continuous version of H3Cit-DNA could provide additional insights on rapid changes in risk that might lead to even more refined cutpoints.
In sum, incorporating GALMs into future research in this area could enhance the statistical robustness of the findings and contribute to a more comprehensive understanding of the factors influencing outcomes after ST-elevation myocardial infarction.
Conflict-of-interest disclosure: C.N.M. declares no competing financial interests.
Acknowledgments
Contribution: C.N.M. wrote the manuscript as sole author.
References
- 1.Benkhoff M, Alde K, Ehreiser V, et al. Thromboinflammation is associated with clinical outcome after ST-elevation myocardial infarction. Blood Adv. 2024;8(21):5581–5589. doi: 10.1182/bloodadvances.2024014273. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Wood SN. Inference and computation with generalized additive models and their extensions. Test. 2020;29(2):307–339. [Google Scholar]
- 3.Stasinopoulos MD, Kneib T, Klein N, et al. Cambridge University Press; 2024. Generalized additive models for location, scale and shape. [Google Scholar]
- 4.Ye H, Chen Z, Li K, Zhang Y, Li H, Tian N. Non-linear association of the platelet/high-density lipoprotein cholesterol ratio with bone mineral density a cross-sectional study. Lipids Health Dis. 2024;23(1) doi: 10.1186/s12944-024-02291-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Zhou P, Liu RL, Wang FX, Hu HF, Deng Z. Blood urea nitrogen has a nonlinear association with 3-month outcomes with acute ischemic stroke: a second analysis based on a prospective cohort study. Clin Nutr ESPEN. 2024;59:140–148. doi: 10.1016/j.clnesp.2023.12.001. [DOI] [PubMed] [Google Scholar]