Establishing next-generation reference intervals for pro-gastrin-releasing peptide using a dynamic modeling approach

Abstract Background Serum pro-gastrin-releasing peptide (ProGRP) levels significantly vary with age. However, conventional partitioned reference intervals (RIs) fail to reflect the continuous and subtle physiological changes associated with aging. To address this limitation, we investigated the age- and sex-specific dynamics of ProGRP levels and innovatively developed next-generation RIs for adults and elderly individuals in North China. Methods A total of 4136 rigorously screened individuals (aged 20–80 years) were analyzed. The effects of age and sex on ProGRP levels were assessed using Mann–Whitney U tests, Spearman’s correlation, and Kruskal–Wallis tests. The Harris–Boyd method was used to evaluate the necessity of sex or age partitioning. Age-partitioned RIs were established using a nonparametric method. Next-generation RI models were developed using generalized additive models for location, scale, and shape, with age-related dynamics visualized. The clinical applicability of both RI types was evaluated by comparing upper reference limit flagging rates across age subgroups; rates closer to the theoretical 2.5% were considered indicative of superior RI accuracy. Results Sex had a minimal and clinically insignificant effect on ProGRP levels. In contrast, age had a significant effect: concentrations remained stable until approximately 40 years of age, followed by progressive increases. The partitioned RIs were defined as 0–52.77 pg/mL (20–49 years) and 0–63.68 pg/mL (≥ 50 years). Next-generation RIs were quantified and visualized. Compared with partitioned RIs, next-generation RIs yielded reference limit flagging rates more closely aligned with the theoretical 2.5% across most age groups and demonstrated significantly greater stability (1.83%–3.74% vs. 1.07%–7.10%). Conclusions Compared with partitioned RIs, next-generation RIs for ProGRP improve the accuracy of laboratory result interpretation. With advances in laboratory informatics, broader clinical implementation of next-generation RIs is anticipated.