Non-invasive methods for β-cell function assessment using a urine C-peptide model corrected for urine volume

To the Editor: Progressive decline of pancreatic β-cell insulin secretion, with or without insulin resistance, is a central feature of type 2 diabetes pathophysiology. Accurate assessment of β-cell function is therefore essential for appropriate diagnosis and treatment selection. Several methods exist, including the homeostasis model assessment (HOMA),^[1] which uses fasting glucose and insulin and is widely used in epidemiological studies, but lacks accuracy. Load tests remain the gold standard for precise assessment of islet function, including oral glucose tolerance tests (OGTT),^[2] intravenous glucose tolerance tests (IVGTT),^[3] mixed-meal tolerance tests (MMTT),^[4] and hyperglycemic clamps. These methods require repeated venous sampling for insulin or C-peptide measurements, thereby increasing patient discomfort and restricting their frequent use. Clinically, non-invasive methods for β-cell function assessment are needed.

C-peptide is secreted in equimolar amounts with insulin by pancreatic β-cells and is excreted in urine at 5–10% of secreted amounts, compared with only 0.1% for insulin.^[5] Consequently, urinary C-peptide (UCP) serves as a non-invasive marker of endogenous insulin secretion. Although 24-h UCP collection accurately reflects β-cell function,^[6] it remains impractical. The UCP creatinine ratio (UCPCR) provides an alternative approach, and the 2-h post-meal UCPCR has been widely studied. However, the optimal timing and reliability of UCP measurement and UCPCR after tolerance testing remain unclear. In this study, we aimed to identify the optimal urinary indicator and collection time for assessing β-cell secretion, determine factors influencing UCP, and develop predictive equations for serum C-peptide from UCP.

The study included 2 cohorts: individuals with normal glucose tolerance (NGT) and patients with diabetes mellitus (DM). NGT subjects were selected using strict metabolic criteria, including age >18 years, body mass index (BMI) between 18 and 24 kg/m², normal fasting and 2-h post prandial plasma glucose levels, normal lipid profile, blood pressure <140/90 mmHg, and no family history of diabetes. A total of 54 NGT participants were recruited through advertisements. Between 2017 and 2018, 270 patients with diabetes aged 18–79 years were enrolled at the First Affiliated Hospital with Nanjing Medical University. Additionally, from August 2023 to February 2024, 700 patients with or without diabetes admitted to the Endocrinology Department underwent a 75-g OGTT or an MMTT using steamed bread (100 g wheat flour), followed by blood and urine sampling. The whole study was approved by the Ethics Committee of the First Affiliated Hospital with Nanjing Medical University (No.2021-SR-298). Written informed consent was obtained from all participants.

After an evening meal, participants fasted overnight while collecting all urine, including the morning void. Subjects with NGT and patients with DM underwent a 75 g OGTT or MMTT, respectively, with blood samples drawn at baseline and 0.5 h, 1 h, 2 h, and 3 h post-load to measure glucose, insulin, and C-peptide. Urine was collected hourly during tolerance tests for C-peptide and creatinine measurements. Volumes were recorded, and cumulative urinary indices were calculated for the 0–2 h and 0–3 h periods [Supplementary Figure 1, http://links.lww.com/CM9/C955].

The Botnia clamp was used to assess insulin secretion and sensitivity independently. This combined an intravenous glucose tolerance test (IVGTT) with a 0.3 g/kg glucose bolus and frequent blood sampling over 60 min to assess first- and second-phase insulin secretion (SPIR), and a hyperinsulinemic-euglycemic clamp, in which insulin was infused continuously at 2 mU·kg⁻¹·min⁻¹ while glucose infusion rates were adjusted to maintain glycemia at 5 mmol/L. Insulin sensitivity indices were calculated from glucose uptake rates and insulin concentrations.

Biochemical assays included hexokinase determination of plasma glucose, automatic biochemical analysis of lipid and renal markers (HITACHI 7020, Japan), and electrochemiluminescence immunoassays for insulin and C-peptides (Roche Diagnostics GmbH, German). Glycated hemoglobin (HbA1c) was measured using high-performance liquid chromatography (Bio-Rad D10, United States). Data were analyzed using Student’s t-test, chi-square test, Pearson correlation, and linear regression in R software (version 4.1.0, New Zealand), with significance set at P <0.05.

The characteristics of subjects with NGT and DM are presented in Supplementary Table 1, http://links.lww.com/CM9/C955. Age, BMI, and metabolic indicators, including HbA1c, urea, triglycerides, high-density lipoprotein, alanine aminotransferase (ALT), total bilirubin, and direct bilirubin levels, differed significantly between groups. In NGT subjects, mean glucose was 4.45 mmol/L (95% CI: 4.31–4.58 mmol/L) and peaked at 7.98 mmol/L (95% CI: 7.54–8.41 mmol/L) at 0.5 h during OGTT [Supplementary Figure 2A, http://links.lww.com/CM9/C955]. Fasting insulin (INS0) and C-peptide (CP0) were 49.03 pmol/L and 597.98 pmol/L, respectively. After glucose loading, insulin and C-peptide peaked at 1 h (589.28 and 3306 pmol/L) before returning to baseline after 180 min [Supplementary Figure 2B, C, http://links.lww.com/CM9/C955]. Always, patients had higher glucose levels, peaking at 2 h [Supplementary Figure 2A, http://links.lww.com/CM9/C955]. CP0 levels were lower than in NGT subjects, whereas insulin levels were similar, possibly reflecting exogenous insulin use. Insulin and C-peptide responses were delayed, rising gradually from 0.5 to 2 h and remaining below NGT curves, but exceeding NGT levels at 3 h [Supplementary Figure 2B, C, http://links.lww.com/CM9/C955].

UCP levels during tolerance tests paralleled serum C-peptide levels, rising from baseline to a peak at 1–2 h and then returning to fasting levels. Subjects with DM had significantly lower fasting UCP levels than NGT subjects, although levels were comparable by 2–3 h [Supplementary Figure 2D, http://links.lww.com/CM9/C955]. Urinary creatinine (UCR) was also lower in the DM group, except at 2–3 h [Supplementary Figure 2E, http://links.lww.com/CM9/C955], whereas the urine volumes were comparable [Supplementary Figure 2F, http://links.lww.com/CM9/C955]. After adjusting for urine volume and creatinine, the total UCR excretion quantity of urinary excreted creatinine (UCRQ) was similar between groups, except at 0–1 h [Supplementary Figure 3, http://links.lww.com/CM9/C955]. UCP quantity (UCPQ) and UCPCR were higher in NGT subjects during the first 2 h but were similar at 2–3 h [Supplementary Figure 3A–C, http://links.lww.com/CM9/C955]. The fold differences in UCPQ between the NGT and DM groups were greater than those observed in UCPCR. Insulin and C-peptide area under curve (AUC) during tolerance tests were approximately twice as high in the NGT group compared to those in the DM group, consistent with the UCPQ changes.

UCPQ indices showed stronger correlations with fasting and AUC insulin and C-peptide levels than with UCPCR in both groups [Supplementary Table 2, http://links.lww.com/CM9/C955]. In the NGT group, the strongest correlations were observed with quantity of urinary excreted C-peptide 0–2h (UCPQ0–2h) and quantity of urinary excreted C-peptide 0–3h (UCPQ0–3h) (r = 0.627), whereas in the DM, UCPQ0-3h showed the highest correlation (r = 0.648). UCPCR correlations were weaker but remained significant. Metabolic factors, including age, BMI, HbA1c, creatinine, triglycerides, uric acid, and ALT levels, were correlated with insulin secretion indices [Supplementary Table 3, http://links.lww.com/CM9/C955]. Multivariate regression incorporating these factors improved prediction of insulin and C-peptide AUCs, with UCPQ0–3h showing higher coefficients than UCPCR0–3h [Supplementary Tables 4 and 5, http://links.lww.com/CM9/C955].

Botnia clamp tests in 12 NGT subjects confirmed strong correlations between UCPQ0–3h and first- and SPIR (R² = 0.443 and 0.750, respectively), as well as insulin sensitivity [Supplementary Figure 4, http://links.lww.com/CM9/C955]. UCPCR0–3h showed weaker correlations, supporting the superiority of UCPQ as the superior indicator of β-cell function.

To validate these findings, 700 hospitalized patients undergoing MMTT were analyzed [Supplementary Table 6, http://links.lww.com/CM9/C955]. Insulin and C-peptide levels peaked at 120 min [Supplementary Figure 5A,B, http://links.lww.com/CM9/C955]. UCPQ0–3h correlated more strongly with C-peptide AUC (r = 0.558) than with UCPCR0–3h (r = 0.497), suggesting that volume-adjusted UCP is a superior marker [Supplementary Table 7, http://links.lww.com/CM9/C955].

Furthermore, factors influencing the relationship between UCPQ0-3h and area under curve of serum C-peptide (AUCcp) were analyzed using residuals. Regression analysis yielded

AUCcp = exp[0.86049 × ln(UCPQ0–3h) – 0.03369 × (ln[UCPQ0–3h])² + 7.59646]

(UCPQ0–3h, pmol/1000) (R² = 0.328, P <0.001) [Supplementary Figure 5C, Supplementary Table 8, http://links.lww.com/CM9/C955].

Residual analysis showed that estimated glomerular filtration rate (eGFR), HbA1c, and weight influenced the UCPQ0-3h–AUCcp relationship, with lower eGFR and higher HbA1c being linked to lower UCPQ0–3h at a given AUCcp [Supplementary Figure 5D–F, http://links.lww.com/CM9/C955]. Urinary albumin and 24-h proteinuria were not affected [Supplementary Figure 6, http://links.lww.com/CM9/C955]. Incorporating eGFR, HbA1c, weight, and sex improved the predictive model:

AUCcp = exp[0.844 × ln(UCPQ0–3h) – 0.038 × ln(UCPQ0–3h)² + 0.835 × ln(Weight) – 0.448 × ln(eGFR) + 0.215 × Sex + 7.533]

(UCPQ0–3h, pmol/1000; Weight, kg; HbA1c,%; eGFR, mL·min⁻¹·1.73 m⁻²; Sex, male = 1, female = 2) (R² = 0.516, P <0.001) [Supplementary Table 9, http://links.lww.com/CM9/C955].

C-peptide is generated by cleavage of proinsulin in β-cells and secreted with insulin in equimolar amounts. Unlike insulin, which is largely metabolized by the liver and peripheral tissues, C-peptide undergoes minimal hepatic metabolism and is predominantly degraded in the kidneys. It is more stable in the circulation and is cleared by the kidneys approximately 30 times faster than insulin,^[7] resulting in substantial urinary excretion. In insulin-treated patients with diabetes, circulating insulin antibodies may artificially elevate plasma C-peptide levels, whereas UCP more accurately reflects β-cell secretion.

C-peptide is less stable in blood due to protease activity, but remains stable in urine at room temperature for 24 h without preservative and for up to 72 h with boric acid or sodium carbonate. It also withstands long-term storage and multiple freeze–thaw cycles without degradation. UCP has long been recognized as a non-invasive marker of β-cell function; although 24-h collections accurately reflect secretory capacity, they are limited by practicality. UCPCR after glucose or meal tolerance correlates well with serum C-peptide,^[8] similar to the albumin-to-creatinine ratio in nephropathy assessment.

Our data showed that hourly UCP excretion after oral tolerance paralleled changes in serum C-peptide and insulin levels. Cumulative post-tolerance UCPQ correlated more strongly with insulin and C-peptide AUCs than with spot samples. Notably, UCPQ0–3h demonstrated a stronger association than UCPCR, as confirmed by the Botnia clamp, showing closer correlations with both insulin secretion phases. Despite the strong correlation between UCPQ0–3h and UCPCR, 3-h urine collection is constrained by clinical scheduling. Consequently, this approach is likely feasible mainly in inpatient settings and requires thorough patient education and cooperation.

Although UCPCR has been validated in pregnancy, children, and diabetes subtypes,^[9] further studies are needed to evaluate UCPQ0–3h in these contexts. UCP excretion may be affected by hyperglycemia-related changes in glomerular filtration rate (GFR), renal impairment, treatment, and certain medications, requiring cautious interpretation of results. Although some reports suggest that UCP is less influenced by reduced GFR,^[10] our findings link β-cell secretion with creatinine and eGFR, but not to proteinuria.

In conclusion, 3-h cumulative UCP excretion after tolerance testing provides a more reliable, non-invasive indicator of endogenous insulin secretion than spot UCP or UCPCR in normal subjects and patients with diabetes.

Funding

This study was supported by grants from Non-communicable Chronic Diseases- National Science and Technology Major Project (No. 2023ZD0507000), the National Key Research and Development Program of China (No. 2024YFA1802900), and the National Natu­ral Science Foundation of China (No. 82200888).

Conflicts of interest

None.

Supplementary Material

cm9-139-2830-s001.pdf ^{(3MB, pdf)}