PLOS ONE Article

\documentclass[10pt,letterpaper]{article}
\usepackage[top=0.85in,left=2.75in,footskip=0.75in]{geometry}
\usepackage{times}
\usepackage{graphicx}
\usepackage{amsmath,amssymb}
\usepackage{lineno}
\usepackage{natbib}
\usepackage{hyperref}

\linenumbers

\title{Title of the Manuscript (Up to 250 Characters)}
\date{}

\begin{document}
\begin{flushleft}
{\Large\textbf\newline{Longitudinal Analysis of Biomarker Expression and Clinical Outcomes in a Multi-Center Cohort Study}}\newline\\
First A.~Author\textsuperscript{1,*},
Second B.~Author\textsuperscript{2},
Third C.~Author\textsuperscript{1,2}\\
\bigskip
\textbf{1} Department of Epidemiology, School of Public Health, University of California, Berkeley, CA, USA\\
\textbf{2} Division of Biostatistics, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA\\
\bigskip
* [email protected]
\end{flushleft}

\section*{Abstract}
Chronic inflammatory conditions remain a leading cause of morbidity worldwide, yet reliable prognostic biomarkers are lacking. We conducted a prospective, multi-center cohort study across four academic medical centers to evaluate the predictive capacity of serum interleukin-6 (IL-6), C-reactive protein (CRP), and tumor necrosis factor alpha (TNF-$\alpha$) for disease progression over a 24-month follow-up period. A total of 1{,}247 participants aged 30--75 years with confirmed diagnoses were enrolled between January 2021 and June 2022. Serial blood samples were collected at baseline, 6, 12, and 24 months. Mixed-effects logistic regression models adjusted for age, sex, body mass index, and comorbidity burden were used to assess associations between biomarker trajectories and clinical endpoints. Elevated baseline IL-6 levels ($>$15~pg/mL) were associated with a 2.3-fold increased risk of disease progression (95\% CI: 1.7--3.1, $p < 0.001$). CRP showed moderate predictive value (AUC = 0.72), while TNF-$\alpha$ did not reach significance after adjustment. These findings support the integration of IL-6 monitoring into routine clinical practice for risk stratification.

\section*{Author summary}
Inflammation plays a central role in many chronic diseases, but clinicians lack simple blood tests that reliably predict which patients will worsen over time. We followed more than 1{,}200 patients at four hospitals for two years, measuring three common markers of inflammation in their blood at regular intervals. We found that one marker, interleukin-6, was strongly associated with disease worsening, even after accounting for other patient characteristics. A second marker, C-reactive protein, showed moderate utility, while a third marker did not provide additional predictive value. Our results suggest that routine measurement of interleukin-6 could help doctors identify high-risk patients earlier and tailor treatment plans accordingly.

\section*{Ethics statement}
This study was approved by the Institutional Review Boards of all participating institutions (IRB Protocol \#2021-0347). All participants provided written informed consent prior to enrollment. The study was conducted in accordance with the Declaration of Helsinki and registered at ClinicalTrials.gov (NCT04XXXXXXX).

\linenumbers

\section*{Introduction}
Chronic inflammatory diseases affect an estimated 40\% of the global population and are a leading driver of healthcare expenditure~\cite{ex1}. Despite decades of research, prognostic tools that enable early identification of patients at highest risk of progression remain limited~\cite{ex2}. Serum biomarkers of systemic inflammation, including interleukin-6 (IL-6), C-reactive protein (CRP), and tumor necrosis factor alpha (TNF-$\alpha$), have been individually studied, but their comparative predictive performance in longitudinal settings is poorly characterized~\cite{ex3}.

Prior single-center studies have reported conflicting results regarding the prognostic value of CRP and TNF-$\alpha$, in part due to heterogeneous patient populations and short follow-up durations~\cite{ex4}. A meta-analysis of 15 observational studies found pooled AUCs ranging from 0.55 to 0.78 for CRP, highlighting the need for standardized multi-center investigations with serial biomarker measurements to resolve these discrepancies and establish clinically actionable thresholds.

The pathophysiological rationale for monitoring IL-6 is supported by its upstream role in the inflammatory cascade. Unlike CRP, which is a downstream hepatic acute-phase protein, IL-6 directly activates the JAK-STAT signaling pathway and has been shown to correlate with tissue-level inflammation in biopsy studies~\cite{ex5}. TNF-$\alpha$, while central to innate immune activation, exhibits rapid circadian fluctuations that may limit its utility as a single-timepoint predictor.

In this study, we address these gaps by evaluating the longitudinal trajectories of three inflammatory biomarkers in a well-characterized cohort of 1{,}247 patients followed for 24~months across four academic medical centers. Our primary objectives are to (1)~compare the predictive accuracy of IL-6, CRP, and TNF-$\alpha$ for disease progression, (2)~identify optimal biomarker thresholds for clinical use, and (3)~assess whether serial measurements improve prediction beyond single baseline values.

\section*{Materials and methods}

\subsection*{Study design}
We conducted a prospective, observational cohort study at four academic
medical centers in the United States between January 2021 and December 2023.
The study protocol was designed in accordance with the STROBE guidelines for
observational research. Sites were selected to ensure geographic and
demographic diversity, representing the Northeast, Southeast, Midwest, and
West Coast regions.

\subsection*{Participants}
Adults aged 30--75 years with a confirmed clinical diagnosis based on standardized criteria were eligible for enrollment. Exclusion criteria included active malignancy, pregnancy, immunosuppressive therapy within the preceding 6~months, and estimated glomerular filtration rate below 30~mL/min/1.73~m$^2$. A total of 1{,}247 participants were enrolled; 1{,}112 (89.2\%) completed all follow-up visits.

\subsection*{Data collection}
Venous blood samples were collected at baseline, 6, 12, and 24~months into EDTA and serum separator tubes. IL-6 and TNF-$\alpha$ were quantified using high-sensitivity ELISA kits (R\&D Systems, Minneapolis, MN), and CRP was measured via immunoturbidimetric assay on the Roche Cobas c501 platform. Inter-assay coefficients of variation were 6.2\% for IL-6, 4.8\% for CRP, and 7.1\% for TNF-$\alpha$. Samples with hemolysis indices exceeding 100~mg/dL were excluded from analysis. Demographic and clinical data were abstracted from electronic health records using a standardized case report form. Comorbidities were coded using ICD-10 and summarized by the Charlson comorbidity index.

\subsection*{Statistical analysis}
Mixed-effects logistic regression models were fitted with disease progression as the binary outcome. Random intercepts accounted for within-site clustering. Fixed effects included biomarker concentrations (log-transformed), age, sex, BMI, smoking status, and Charlson comorbidity index. Receiver operating characteristic (ROC) curves were constructed for each biomarker, and AUCs were compared using the DeLong test. All analyses were performed in R~v4.3.1 with the \texttt{lme4} and \texttt{pROC} packages. A two-sided $p < 0.05$ was considered statistically significant.

\section*{Results}

\begin{table}[!h]
\centering
\caption{{\bf Baseline characteristics by disease progression status.}}
\label{tab:baseline}
\begin{tabular}{lcc}
\hline
Characteristic & Progressors ($n=312$) & Non-progressors ($n=935$) \\
\hline
Age, years (mean $\pm$ SD) & $58.4 \pm 11.2$ & $52.1 \pm 13.5$ \\
Female, \% & 54.8 & 51.2 \\
BMI, kg/m$^2$ (mean $\pm$ SD) & $29.7 \pm 5.1$ & $27.3 \pm 4.8$ \\
Current smoker, \% & 18.3 & 14.7 \\
Charlson index (mean $\pm$ SD) & $3.1 \pm 1.8$ & $1.9 \pm 1.4$ \\
IL-6, pg/mL (median [IQR]) & 18.3 [12.1--27.6] & 8.7 [5.2--14.1] \\
CRP, mg/L (median [IQR]) & 6.8 [3.4--12.5] & 3.1 [1.4--6.2] \\
TNF-$\alpha$, pg/mL (median [IQR]) & 4.2 [2.8--6.9] & 3.5 [2.1--5.8] \\
Statin use, \% & 42.6 & 31.8 \\
\hline
\end{tabular}
\end{table}

Subgroup analyses stratified by sex revealed consistent associations between IL-6 and progression in both women (OR = 2.18; 95\% CI: 1.42--3.34) and men (OR = 2.47; 95\% CI: 1.63--3.74), with no significant interaction ($p_{\text{interaction}} = 0.61$). Sensitivity analyses excluding patients with baseline CRP $>$ 50~mg/L (suggestive of acute infection) did not materially alter the results (OR for IL-6 = 2.25; 95\% CI: 1.66--3.05).

\begin{figure}[!h]
\centering
\fbox{\parbox[c][6cm][c]{0.8\linewidth}{\centering ROC curves for IL-6, CRP, and TNF-$\alpha$ as predictors of 24-month disease progression}}
\caption{{\bf Receiver operating characteristic curves for inflammatory biomarkers.} IL-6 yielded the highest AUC (0.81; 95\% CI: 0.77--0.85), followed by CRP (0.72; 95\% CI: 0.67--0.76) and TNF-$\alpha$ (0.58; 95\% CI: 0.53--0.63). The diagonal dashed line represents chance performance.}
\label{fig1}
\end{figure}

\begin{figure}[!h]
\centering
\fbox{\parbox[c][6cm][c]{0.8\linewidth}{\centering Kaplan--Meier curves stratified by baseline IL-6 quartile}}
\caption{{\bf Time-to-progression stratified by baseline IL-6 quartile.} Patients in the highest IL-6 quartile ($>$22~pg/mL) had significantly shorter median time to progression compared to the lowest quartile ($<$7~pg/mL; log-rank $p < 0.001$). Shaded regions indicate 95\% confidence intervals.}
\label{fig2}
\end{figure}

Elevated baseline IL-6 was the strongest independent predictor of 24-month disease progression (OR = 2.31; 95\% CI: 1.72--3.11; $p < 0.001$). CRP retained significance but with a smaller effect size (OR = 1.48; 95\% CI: 1.12--1.95; $p = 0.006$). TNF-$\alpha$ was not significantly associated with the outcome after multivariable adjustment (OR = 1.09; 95\% CI: 0.84--1.42; $p = 0.52$). Serial IL-6 measurements at 6~months improved the AUC from 0.81 to 0.86 compared to baseline alone ($p = 0.01$).

\section*{Discussion}
Our findings establish IL-6 as a robust prognostic biomarker for disease progression in a large, multi-center cohort with extended follow-up. The observed AUC of 0.81 exceeds those reported in prior single-center studies, likely reflecting the benefits of standardized measurement protocols and rigorous quality control across sites. The improvement in discrimination achieved by incorporating 6-month follow-up measurements suggests that biomarker trajectories capture clinically relevant pathophysiological dynamics beyond what a single snapshot can provide.

The modest predictive value of CRP is consistent with its role as a non-specific acute-phase reactant. While CRP remains useful for monitoring acute inflammatory flares, our results suggest it is insufficient as a standalone prognostic marker in chronic disease settings. The lack of association between TNF-$\alpha$ and disease progression aligns with recent meta-analytic evidence indicating high inter-individual variability in circulating TNF-$\alpha$ levels~\cite{ex4}. These findings have implications for clinical guidelines that currently recommend TNF-$\alpha$ testing in routine practice.

\section*{Conclusion}
In a prospective cohort of 1{,}247 patients, IL-6 demonstrated strong and consistent predictive performance for 24-month disease progression. Integration of serial IL-6 monitoring into clinical workflows may improve risk stratification and enable earlier therapeutic intervention. Future studies should evaluate whether biomarker-guided treatment strategies lead to improved patient outcomes.

Several strengths of this study warrant emphasis. First, the multi-center design with standardized collection protocols minimizes pre-analytical variability that has confounded prior single-site investigations. Second, the 24-month follow-up period captures clinically meaningful endpoints rather than relying on surrogate markers. Third, the use of mixed-effects models appropriately accounts for within-site correlation.

This work has limitations. The cohort is drawn exclusively from academic medical centers, which may limit generalizability to community practice settings. Additionally, we did not assess novel biomarkers such as galectin-3 or pentraxin-3 that have shown promise in recent smaller studies. Finally, the observational design precludes causal inference regarding the relationship between biomarker levels and disease progression.

\section*{Data availability statement}
The de-identified dataset analyzed in this study is available from the corresponding author upon reasonable request and execution of a data use agreement. Summary statistics are provided in S1~Table.

\section*{Supporting information}
\paragraph*{S1 Fig.} {\bf Receiver operating characteristic curves by study site.} Site-specific ROC curves for IL-6 demonstrating consistent performance across the four participating centers (AUC range: 0.78--0.84).

\paragraph*{S1 Table.} {\bf Complete regression model output.} Full multivariable logistic regression results including all covariates, interaction terms, and sensitivity analyses.

\section*{Acknowledgments}
We gratefully acknowledge the clinical research coordinators at each participating site: M.~Johnson (Site~1), S.~Patel (Site~2), R.~Chen (Site~3), and L.~Garcia (Site~4). We thank Dr.~Jane Smith for statistical advice and Dr.~Robert Lee for critical review of the manuscript. This work was supported by NIH grants R01-AI142857 and UL1-TR001863, and the Burroughs Wellcome Fund.

\nolinenumbers

\bibliographystyle{plos2015}
\begin{thebibliography}{5}
\bibitem{ex1} Author A, Author B. Global burden of chronic inflammatory disease: a systematic review. \emph{Lancet}. 2023;401(10382):1120--1135.
\bibitem{ex2} Chen Y, Patel R, Thompson S. Biomarkers for disease progression: challenges and opportunities. \emph{Nature Reviews Immunology}. 2022;22(5):310--325.
\bibitem{ex3} Williams K, Nakamura T. Interleukin-6 as a prognostic indicator: meta-analysis of 42 studies. \emph{JAMA Internal Medicine}. 2021;181(8):1044--1056.
\bibitem{ex4} Rodriguez M, Fischer A, Gupta S. Variability in circulating TNF-$\alpha$ levels across populations: implications for clinical utility. \emph{Annals of Internal Medicine}. 2023;178(3):345--358.
\bibitem{ex5} Lee H, Park J, Kim D. Longitudinal biomarker trajectories and clinical outcomes in chronic disease cohorts. \emph{BMJ Open}. 2024;14(2):e072341.
\end{thebibliography}

\end{document}