Abstract
Recent work describing longitudinal soluble programmed death-ligand 1 (sPD-L1) trajectories in critically ill patients with COVID-19 provides valuable insight into immune checkpoint dynamics and mortality risk. However, interpretation of these findings warrants caution. Longitudinal analyses are vulnerable to survivorship and informative censoring bias, potentially overstating declining biomarker trends among survivors. In addition, the absence of time-varying adjustment for evolving disease severity and organ support limits causal inference regarding sPD-L1 as an independent prognostic marker. Finally, renal dysfunction may confound circulating sPD-L1 concentrations through impaired clearance. Addressing these methodological issues is essential for validating sPD-L1 as a robust prognostic biomarker in critical illness.
Dear Editor,
We read with great interest the study by Takeuchi et al., which offers valuable longitudinal insight into soluble programmed death-ligand 1 (sPD-L1) dynamics and their association with mortality in critically ill patients with COVID-19 [1]. The authors should be commended for addressing an important gap in the literature. Our intent is not to dispute the biological plausibility of sPD-L1 dynamics, but to highlight methodological challenges in interpreting biomarker trajectories during acute critical illness. Nevertheless, several methodological considerations merit discussion, as they may affect interpretation and generalizability.
First, longitudinal in-hospital analyses of sPD-L1 trajectories are susceptible to survivorship and informative censoring biases. Patients who died or were discharged were progressively excluded from later ICU sampling, potentially exaggerating the observed decline in sPD-L1 among remaining patients. Such bias is well recognized in longitudinal studies where attrition is linked to disease severity and may lead to misleading inferences, whereby apparent biomarker trends reflect selective loss of the most severely ill rather than true biological resolution [2].
Second, the longitudinal and survival models did not sufficiently account for organ support requirements or time-varying disease severity. Biomarkers such as sPD-L1 are likely to fluctuate alongside dynamic changes in organ dysfunction and the evolving need for vasopressors, renal replacement therapy, or mechanical ventilation. Methodological studies have consistently demonstrated that failure to model time-dependent confounders can substantially bias associations between longitudinal exposures and mortality, particularly in ICU settings where illness severity changes rapidly [3, 4]. As with established biomarkers such as creatinine or albumin, temporal changes in sPD-L1 may therefore primarily reflect evolving organ dysfunction rather than disease-specific immune modulation. As a result, it remains unclear whether sPD-L1 independently predicts mortality or primarily reflects progressive multiorgan failure.
Third, impaired renal function complicates interpretation of sPD-L1 as a marker of immunological exhaustion. In this study, sPD-L1 correlated closely with estimated glomerular filtration rate and creatinine. Prior evidence suggests that circulating immune checkpoint molecules may accumulate in acute kidney injury due to reduced clearance rather than increased production [5]. Without explicit adjustment for renal function or biomarker kinetics, elevated sPD-L1 levels may partially reflect impaired elimination, limiting biological interpretability.
Taken together, these considerations warrant cautious interpretation of longitudinal immunological biomarkers in critically ill patients. Future multicenter studies incorporating adjustment for time-varying disease severity, renal function-informed modeling, and strategies to address survivorship bias are needed to establish sPD-L1 as an independent and clinically actionable predictive biomarker.
Author contributions
All authors showed equal contributions in the preparation of this manuscript.
Funding
No funding was received for this work.
Data availability
No datasets were generated or analysed during the current study.
Declarations
Ethics approval and consent to participate
Not applicable.
Informed consent
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Takeuchi S, Kawamoto E, Matsusaki T, Ono D, Sakakura Y, Gaowa A, et al. Use of dynamic soluble programmed death-ligand 1 trajectories to identify early organ dysfunction and predict mortality in critical coronavirus disease. J Intens Care. 2019. 10.1186/s40560-025-00846-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Czeisler MÉ, Wiley JF, Czeisler CA, Rajaratnam SM, Howard ME. Uncovering survivorship bias in longitudinal mental health surveys during the COVID-19 pandemic. Epidemiol Psychiatr Sci. 2021;30:e45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Mansournia MA, Etminan M, Danaei G, Kaufman JS, Collins G. Handling time varying confounding in observational research. BMJ. 2017;16:359. [DOI] [PubMed] [Google Scholar]
- 4.Petersen ML, Porter KE, Gruber S, Wang Y, Van Der Laan MJ. Diagnosing and responding to violations in the positivity assumption. Stat Methods Med Res. 2012;21(1):31–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Pickering JW, Endre ZH. Acute kidney injury urinary biomarker time-courses. PLoS ONE. 2014;9(7):e101288. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
No datasets were generated or analysed during the current study.