Markers of Collagen Formation and Degradation Reflect Renal Function and Predict Adverse Outcomes in Patients With Type 1 Diabetes

Despite improvements in the medical care of patients with type 1 diabetes (T1D), the risk of developing chronic kidney disease (CKD) and cardiovascular events (CVEs) and of premature death remains higher in these patients compared with the general population (1–4). There is growing evidence that dysregulation of extracellular matrix (ECM) turnover is associated with accelerated systemic and renal fibrosis and modulation of inflammation (5–7). Regardless of the underlying etiology, CKD is believed to be driven by renal fibrosis (8). Collagens are an essential part of the fibrotic structure, acting as a scaffold for a range of ECM molecule interactions and cell adhesion (6,9,10). Because increased collagen formation and reduced degradation are closely linked with the development of fibrosis, assessment of collagen formation and degradation may identify patients with active fibrosis at higher risk of CVEs and heart failure, mortality, and rapid deterioration of kidney function.

Collagen type VI (COL VI) is located in the interface between the basement membrane and interstitial matrix in most connective tissues where it provides structural support for cells (11). We have recently shown that serum PRO-C6 (sPRO-C6), a marker for COL VI formation, was independently associated with adverse outcomes in patients with type 2 diabetes (12). These findings support the potential role of PRO-C6 as a marker that may reflect COL VI formation.

Collagen type III (COL III) is one of the most prevalent collagens in the fibrotic kidney (13). In a recent study, fragments of COL III were identified in the urine of patients with CKD and could help to discriminate between patients with CKD and healthy individuals (14). The specific C3M fragment, a COL III fragment, has been previously associated with disease severity and CKD progression (15). These findings suggest that C3M can be used as a marker of renal fibrosis.

We hypothesized that PRO-C6 and C3M reflect pathophysiological processes associated with COL VI formation and COL III degradation and, therefore, are 1) associated with kidney function and 2) risk markers for mortality, CVEs, decline in estimated glomerular filtration rate (eGFR), and end-stage renal disease (ESRD) in patients with T1D. This study is the first to our knowledge to investigate the biomarkers PRO-C6 and C3M in a T1D cohort with various degrees of diabetic nephropathy and to assess the association of the biomarkers with relevant clinical outcomes.

Between 2009 and 2011, adult patients with T1D were recruited to enter a study at Steno Diabetes Center Copenhagen. The original study has been previously described (16,17). Patients with ESRD, defined as receiving dialysis or renal transplantation, or eGFR <15 mL/min/1.73 m² were not included. All participants gave written informed consent, and the study was approved by the ethics committee of the capital region of Copenhagen.

Three 24-h urine collections were obtained to measure urinary albumin excretion rate (UAER) using an enzyme immunoassay (VITROS; Raritan, NJ). Participants were categorized as normoalbuminuric if UAER was <30 mg/24 h, as microalbuminuric if UAER was or previously had been recorded as between 30 and 299 mg/24 h, and as macroalbuminuric if UAER was or previously had been recorded as ≥300 mg/24 h in two of three consecutive measurements. All patients classified as normoalbuminuric did not have any history of micro- or macroalbuminuria before enrollment in the study.

HbA_1c was measured by high-performance liquid chromatography (Bio-Rad Laboratories, Munich, Germany) and serum creatinine concentration by an enzyme method (Hitachi 912; Roche Diagnostics, Mannheim, Germany). The eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation (18).

Serum samples from 663 patients and urine from 609 patients were available for the current study. PRO-C6 and C3M were measured in serum (sPRO-C6, sC3M) and urine (uPRO-C6, uC3M) using competitive ELISAs developed by Nordic Bioscience (Herlev, Denmark). The monoclonal antibodies used in the PRO-C6 and C3M ELISAs specifically detect the last 10 amino acids of the COOH-terminal of the α-3 chain of COL VI (3168′KPGVISVMGT′3177) (19) and the neoepitope generated by matrix metalloproteinase-9–mediated cleavage of COL III (610′KNGETGPQGP′619) (20), respectively. Intra- and interassay variations of the ELISAs were <10 and 15%, respectively. To normalize for urine output, urinary levels of the markers were normalized to urinary creatinine levels. Urinary creatinine was measured using the ADVIA 1800 Chemistry System (Bayer HealthCare). The ELISAs were carried out as previously described (12,15,19–21).

On 31 December 2016, we traced all patients through the Danish National Death Register and the Danish National Health Register (22,23). For deceased patients, information was obtained on the date and cause of death. Data on cause of death up to 31 December 2015 were available to us. All deaths were classified as cardiovascular unless an unambiguous noncardiovascular cause was diagnosed, a previously recognized approach (24). Cause of death was not definitive in only three participants. Information about hospital admission and related ICD-10 diagnoses (https://icd.who.int/browse10/2016/en) and procedural codes (according to the Nordic Classification of Surgical Procedures, www.sst.dk) were obtained from the Danish National Health Register. The combined cardiovascular end point was defined as cardiovascular death, nonfatal acute myocardial infarction (ICD-10 codes I21–I24), nonfatal stroke (ICD-10 codes I61–I66), and coronary interventions (procedural codes KFNA-G) or as peripheral arterial interventions, including amputations. Hospitalization because of heart failure comprised the ICD-10 code I50.

Information on eGFR during follow-up was obtained at outpatient visits and was traced through electronic laboratory records. On the basis of this information, we defined the renal end point eGFR decline of ≥30%.

Time to ESRD was defined as time to development of CKD stage 5 (ICD-10 code N18.5), chronic dialysis (procedural code BJFD2), kidney transplantation (procedural codes KKAS 00, 10, and 20), eGFR <15 mL/min/1.73 m², or renal failure as cause of death.

For participants who experienced multiple end points, the analysis included only the first event. Median follow-up was 5.1 years (interquartile range [IQR] 4.7–5.6) for the combined cardiovascular end point, 5.2 years (4.8–5.7) for heart failure, 6.2 years (5.6–6.7) for mortality, 5.2 years (4.8–5.7) for ESRD, and 5.1 years (2.1–6.1) for decline in eGFR ≥30%.

All normally distributed variables are presented as mean ± SD and categorical variables as total numbers with corresponding percentages. The distribution of serum and urinary PRO-C6 and C3M as well as UAER was skewed, and these variables were log₂ transformed in all analyses and summarized as medians with IQRs. Clinical characteristics of the population at baseline were stratified by median concentrations of sPRO-C6 and sC3M (Supplementary Table 1) and by median concentrations of uPRO-C6 and uC3M (Supplementary Table 2) and compared using a t test for continuous variables and χ² test for categorical variables.

First, we performed unadjusted and adjusted linear regression analyses. We used the unadjusted models to determine whether any association existed between level of serum and urinary PRO-C6 and C3M and renal function (eGFR and UAER). The subsequent multivariable adjustment included sex, age, diabetes duration, LDL cholesterol, smoking, HbA_1c, systolic blood pressure, UAER (in analyses of eGFR), and baseline eGFR (in analyses of UAER).

Next, for the analysis of follow-up data, we applied Cox proportional hazards analyses to compute hazard ratios (HRs) with 95% CIs per doubling of serum and urinary PRO-C6 and C3M for all specified end points. First, we applied unadjusted models to determine whether any association existed between the markers and the predefined end points. The subsequent adjustment included sex, age, diabetes duration, LDL cholesterol, smoking, HbA_1c, systolic blood pressure, UAER, and eGFR. To further investigate the results, we did a sensitivity analysis to assess whether BMI and history of cardiovascular disease at baseline influenced our findings.

We evaluated the added prognostic impact to the traditional risk prediction methods for the significant findings in the adjusted model. We calculated the integrated discrimination improvement (IDI), which has been suggested as a more powerful method to demonstrate improved diagnostic performance of a biomarker (25). For a more comprehensive measure, the relative IDI (rIDI) was calculated, which is the relative contribution of the new biomarker for improved risk prediction compared with the standard approach. To investigate the potential of the biomarker after multivariable adjustment in distinct subgroups of patients, we investigated the association of biomarkers with eGFR >60, 45, and 30 mL/min/1.73 m² and in patients with normoalbuminuria, microalbuminuria, and macroalbuminuria. Adjustments in this analysis included sex, age, diabetes duration, LDL cholesterol, smoking, HbA_1c, systolic blood pressure, UAER, and eGFR. Patients were excluded from the analysis if data were missing for variables used in the adjustments.

For database management and statistical analysis, we used SAS 9.4 software (SAS Institute, Cary, NC) and MedCalc version 16.8.4 (www.medcalc.org). Statistical significance was an α-level of 0.05 on two-sided tests.

Of the original cohort (n = 676), measurements of sPRO-C6 and sC3M were available in 663 (98.1%) patients, and measurements of uPRO-C6 and uC3M were available in 609 (90.1%) patients. Clinical characteristics of the study population are shown in Table 1. In the cohort, 384 (56%) were male; mean ± SD age was 55 ± 13 years, and mean baseline eGFR was 82 ± 26 mL/min/1.73 m² (Table 1). Median (IQR) baseline UAER was 17 (8–65) mg/24 h. The clinical characteristics in the 663 participants below and above the median of serum and urinary PRO-C6 and C3M are shown in Supplementary Tables 1 and 2, respectively. Patients with sPRO-C6 above the median differed in several characteristics: They were older and had longer diabetes duration, lower eGFR, and higher albuminuria level, HbA_1c, HDL cholesterol, blood pressure, and BMI. They were also more frequently prescribed antihypertensive treatment and statins, and more had a history of cardiovascular disease at baseline. In contrast, the patients with sC3M above the median did not differ from the patients with levels below the median, with the exception that they had lower eGFR and higher albuminuria and LDL cholesterol levels.

Patients with uPRO-C6 above the median differed in a few characteristics from those with levels below the median: They were more often female and older and had longer diabetes duration and higher HDL cholesterol and systolic blood pressure. In contrast to this, the patients with uC3M above the median differed in several characteristics compared with patients with levels below the median: They were more often female and younger and had shorter diabetes duration, higher eGFR, and lower UAER, HbA_1c, and BMI, and fewer were treated with antihypertensive drugs and statins and had a history of cardiovascular disease at baseline.

The number of patients with eGFR >90, 60–90, and <60 mL/min/1.73 m² and the levels of the four biomarkers are shown in Supplementary Table 3. The number of patients with normo-, micro-, and macroalbuminuria at baseline and the level of the four biomarkers in these subgroups are shown in Supplementary Table 4.

To assess whether the biomarkers in circulation and urine were associated, we compared levels of PRO-C6 and C3M in serum and urine. There was a significant, but weak, positive correlation between serum and urinary PRO-C6 (r² = 0.14, P < 0.001) (Supplementary Fig. 1). Serum and urinary C3M had a weak negative correlation (r² = 0.01, P = 0.006) (Supplementary Fig. 1B).

To investigate whether the biomarkers were associated with kidney function, we applied linear regression analysis between biomarkers and eGFR at baseline. Higher levels of sPRO-C6, sC3M, and uPRO-C6 were associated with lower eGFR in both the unadjusted and the adjusted analysis (P ≤ 0.001) (Table 2). Higher levels of uC3M were associated with higher eGFR in both the unadjusted and the adjusted analysis (P < 0.001) (Table 2).

Next, we investigated whether the biomarkers were associated with albuminuria by applying linear regression analysis. Higher levels of sPRO-C6, sC3M, and uPRO-C6 were associated with higher UAER in both the unadjusted and the adjusted analysis (P ≤ 0.007) (Table 2). Higher uC3M was associated with lower UAER in the unadjusted analysis (P < 0.001) (Table 2), but significance was lost after adjustment (P = 0.56).

Finally, we investigated the association between duration of diabetes and the various markers in linear regression analysis. Higher levels of sPRO-C6 were associated with longer diabetes duration in both the unadjusted and the adjusted analyses (P < 0.001). Higher levels of uPRO-C6 were associated with longer diabetes duration in the unadjusted analyses (P < 0.001) but not after adjustment (P = 0.45). Higher uC3M was associated with shorter duration of diabetes in the unadjusted analysis (P < 0.001), but significance was lost in adjusted analyses (P = 0.55). sC3M was not associated with duration of diabetes (P = 0.18).

To assess the prognostic potential of the investigated biomarkers, we applied Cox proportional hazard regression analysis. In the follow-up period, 9% of participants (n = 58) died, 14% (n = 94) experienced a CVE, 4.2% (n = 28) were hospitalized because of heart failure, 14% (n = 93) had a decline in eGFR of ≥30%, and 3% (n = 21) were diagnosed with ESRD.

Higher sPRO-C6 was a predictor of mortality in the unadjusted and adjusted analysis (per doubling: HR 2.26 [95% CI 1.31–3.87], P = 0.003) (Table 3). To investigate whether there were distinct subgroups of patients with a more severe outcome, we stratified the patients on the basis of a range of eGFR cutoffs (Table 4). All the investigated cutoffs for eGFR yielded a significant association of the adjusted sPRO-C6 with mortality (all P ≤ 0.005) (Table 4). To further investigate potential patient subgroups with a worse outcome, we also stratified patients on the basis of albuminuria stage. We found that adjusted sPRO-C6 was significantly associated with mortality in patients with normoalbuminuria (UAER <30 mg/24 h; HR 4.16 [1.69–10.24], P = 0.002) (Table 4). The addition of sPRO-C6 to the adjusted model containing conventional risk factors improved the rIDI by 11.2% (P = 0.04) for mortality. Higher sC3M was associated with a higher risk of mortality in the unadjusted analysis (P = 0.004) (Table 3), but the association was attenuated after adjustment (P = 0.09) (Table 3). uPRO-C6 and uC3M were not associated with risk of mortality (P ≥ 0.07) (Table 3).

Higher levels of sPRO-C6 and sC3M were associated with higher risk of CVEs in the unadjusted analysis (both P ≤ 0.003). After adjustment, higher sPRO-C6 and sC3M were no longer associated with higher risk of CVEs (P > 0.16) (Table 3). Higher levels of uC3M were associated with lower risk of cardiovascular events in the unadjusted analysis (P = 0.004) (Table 3), but the association was lost after adjustment (P = 0.8). uPRO-C6 was not associated with CVEs.

Higher levels of sPRO-C6 and sC3M were associated with a higher risk of heart failure in the unadjusted analysis (P ≤ 0.037) (Table 3), but the association was lost after adjustment (P ≥ 0.15) (Table 3). Higher levels of uC3M were associated with a lower risk of heart failure in the unadjusted analysis (P = 0.01) (Table 3), but the association was lost in the adjusted analysis (P = 0.64) (Table 3). uPRO-C6 was not associated with the risk of hospitalization for heart failure.

Higher levels of sPRO-C6 and sC3M were associated with a higher risk of decline in eGFR of ≥30% (n = 93) in the unadjusted analysis (P ≤ 0.002), but the associations were lost after adjustment (P ≥ 0.30). However, sPRO-C6 was significantly associated with decline in eGFR ≥30% in analyses restricted to participants with eGFR >45 and >30 mL/min/1.73 m² (both P ≤ 0.009) (Table 4) but only borderline for participants with eGFR >60 mL/min/1.73 m² (P = 0.06) (Table 4). Higher levels of uPRO-C6 were associated with a higher risk of decline in eGFR of ≥30% in the unadjusted analysis (P = 0.004), and after adjustment, higher uPRO-C6 was associated with a lower risk of decline in eGFR of ≥30% (P = 0.02). Higher levels of uC3M were associated with a lower risk of decline in eGFR of ≥30% in the unadjusted analyses (HR 0.53 [95% CI 0.42–0.68], P < 0.001) but just lost significance after adjustment (HR 0.72 [0.51–1.02], P = 0.07). When stratifying patients by albuminuria stage, we found that there was a significant association between the adjusted sPRO-C6 and decline in eGFR of ≥30% (per doubling: HR 2.82 [1.08–7.33], P = 0.03) (Table 4) in patients with normoalbuminuria (UAER <30 mg/24 h).

Higher levels of sPRO-C6 and sC3M were associated with a higher risk of ESRD in the unadjusted analysis (P ≤ 0.02). Higher sPRO-C6 remained associated with a higher risk of ESRD in the adjusted analysis (P = 0.008) (Table 3). In the investigated subgroups stratified by eGFR and albuminuria, higher sPRO-C6 was significantly associated with ESRD in patients with eGFR >30 mL/min/1.73 m² after adjustment (HR 23.14 [95% CI 1.47–365.5], P = 0.03) (Table 4). sC3M was not associated with ESRD after adjustment (P = 0.24) (Table 3). Higher levels of uPRO-C6 were associated with higher risk of ESRD in the unadjusted analysis (P < 0.001) but lost the association after adjustment (P = 0.57). Higher levels of uC3M were associated with lower risk of ESRD in the unadjusted analysis (P < 0.001) (Table 3), but significance was lost after adjustment (P = 0.96) (Table 3).

All the significant findings remained significant when BMI was included in the adjusted analyses. Moreover, analyses of the cardiovascular outcome excluding participants with a previous diagnosis (n = 154) were confirmatory, with the exception that sPRO-C6 and sC3M were not associated with CVEs in the unadjusted analysis (P ≥ 0.20).

To our knowledge, this study is the first to examine the association between PRO-C6 and C3M and renal function as well as the risk of hard renal and cardiovascular end points, including risk of heart failure and all-cause mortality, in a cohort of patients with T1D. In the current study of 663 patients with T1D and various degrees of albuminuria, we demonstrated that sPRO-C6 was an independent predictor of mortality and ESRD and that it was associated with lower eGFR and higher UAER at baseline. Furthermore, we demonstrated that higher uPRO-C6 was associated with lower eGFR and higher UAER at baseline and with lower risk of decline in eGFR of ≥30% in the follow-up analysis when adjusted for other risk factors. We found a positive correlation between serum and urinary PRO-C6, which may explain the results seen in the cross-sectional analysis. However, it does not account for the unexpected findings in the follow-up analysis, where adjustments for potential confounders showed that higher uPRO-C6 was associated with a lower risk of decline in eGFR of ≥30%. The rIDI has been suggested as a relevant addition to the conventional methods of determining incremental added value of a marker to risk prediction (25). Using rIDI, we showed that the addition of sPRO-C6 to the model increased the prediction of mortality by 11.2% beyond the conventional risk markers. Several studies have assessed PRO-C6, a marker for COL VI formation, in patients with CKD of different etiology and in patients with type 2 diabetes. Fenton et al. (26) demonstrated that high levels of sPRO-C6 were independently associated with increased mortality in patients with high-risk CKD of different etiology. They also demonstrated that higher sPRO-C6 was associated with increased progression to ESRD, but the association was lost after adjustments for eGFR and urinary albumin-to-creatinine ratio (26). This is interesting because we found that high sPRO-C6 was only significantly associated with mortality in patients with normoalbuminuria when grouping the patients on the basis of albuminuria stage, suggesting that high sPRO-C6 is an earlier marker for worse outcome than endothelial dysfunction. We also demonstrated that higher levels of sPRO-C6 were significantly associated with an increased risk of development of ESRD after full adjustment, including UAER and eGFR, which is in contrast to the findings of Fenton et al. When patients were stratified on the basis of a range of eGFR cutoffs, we found that high sPRO-C6 was significantly associated with a risk of developing ESRD in patients with eGFR >30 mL/min/1.73 m². This could indicate that sPRO-C6 is of more prognostic value in patients with mild kidney disease as opposed to patients with severe kidney disease. It could also be due to the fact that we have limited events in our cohort, and this should be reproduced in a larger cohort.

Rasmussen et al. (21) demonstrated that high levels of uPRO-C6 were independently associated with 1-year progression of the CKD in patients with CKD of different etiology independently of traditional risk factors, which is partly in contrast to our findings. It is important to keep in mind that both the results from the study by Fenton et al. (26) and those by Rasmussen et al. have been obtained from analysis of samples from participants from the Renal Impairment in Secondary Care (RIISC) study, a cohort with moderate to advanced CKD (27). Rasmussen et al. also demonstrated that even though PRO-C6 was not a specific marker for the kidneys, biopsies from nonfibrotic kidneys portrayed low levels of COL VI and no PRO-C6 staining, whereas they observed intense staining for both COL VI and PRO-C6 in biopsies from fibrotic kidneys, suggesting that uPRO-C6 reflects disease activity in patients with CKD. Finally, in a study on PRO-C6 in patients with type 2 diabetes with microalbuminuria and without symptoms or prior history of coronary artery disease, we showed that high levels of sPRO-C6 at baseline were independently associated with mortality, CVEs, and decline in eGFR of ≥30% (12). In the current study, we also found high sPRO-C6 to be independently associated with mortality, but we did not find an independent association with CVEs or an increased risk of heart failure. This may in part be due to the difference in study populations. We did, however, show that high levels of sPRO-C6 were significantly associated with a decline in eGFR ≥30% in patients with eGFR >30 (i.e., early to moderate disease), which is in line with the published results in patients with type 2 diabetes (12). We also demonstrated that the association between higher sPRO-C6 and decline in eGFR ≥30% was present in patients with normoalbuminuria, suggesting that high sPRO-C6 is an earlier marker for poor outcome than albuminuria. The cohort of the current study consists of patients with T1D with both normoalbuminuria (<30 mg/24 h) and albuminuria (>30 mg/24 h) compared with the previous study of patients with type 2 diabetes with albuminuria (12). One reason for the different results in the two populations could be that many of the patients in the present cohort had not developed albuminuria, reflecting endothelial dysfunction, a prominent risk factor for developing both kidney and cardiovascular disease. Another possible reason for this difference is the nature of the two types of diabetes. Type 2 diabetes is usually diagnosed after several years of living with the disease, making the patients more likely to develop complications of the disease, and in addition to elevated glucose, as in T1D, clusters with cardiovascular risk factors, including obesity, dyslipidemia, and hypertension.

C3M is the degradation product of COL III and is a measure of COL III turnover in the interstitial matrix. In our study, high levels of sC3M were associated with lower eGFR and higher UAER in the cross-sectional analysis, but in the follow-up analysis, the association with an increased risk of mortality was not independent of conventional risk factors. Higher levels of uC3M were associated with higher eGFR and lower risk of decline in eGFR of ≥30%. Stribos et al. (28) demonstrated that uC3M is lower in patients with impaired kidney function. However, they found no association between plasma C3M levels and eGFR, which is in contrast to our finding. It should be noted that Stribos et al. collected samples from a small cohort of renal transplant recipients as part of a routine check, giving rise to a very diverse cohort. Genovese et al. (15) demonstrated that in patients with IgA nephropathy, lower levels of uC3M were associated with lower eGFR, but the level in serum was not. This is partly in contrast to our finding where higher sC3M levels were associated with lower eGFR. Our findings for uC3M are in line with the previous findings. It should be taken into account that the authors also found that patients treated with immunosuppressive drugs exhibited lower sC3M concentrations, whereas uC3M was not affected. When we adjusted for eGFR in our analysis there was the risk of overadjustment (e.g., a high level of C3M indicates high levels of fibrosis in the kidney, a high level of fibrosis gives a low GFR). Alternatively, a high level of C3M in the serum could be due to accumulation mediated by lack of filtration as seen for creatinine. However, the lack of accumulation of sC3M seen in patients with IgA nephropathy at low GFR (15), along with the very low correlation between C3M in urine and serum, suggests that sC3M is not a marker for filtration but, rather, a marker for fibrosis.

We demonstrated a negative correlation between serum and urinary C3M, suggesting that uC3M reflects a lower remodeling of the renal stroma. sC3M, however, may reflect a systemic profibrotic and proinflammatory environment, whereas other organs contribute to the pool of C3M measured in serum and not just the kidneys. This is a possible explanation of our finding that a higher level of sC3M was detrimental, whereas a higher level of uC3M was beneficial. These findings indicate that changes in PRO-C6 and C3M may reflect pathophysiological alterations of the ECM, and thus disease activity, which would explain the association with adverse outcomes.

CKD is characterized by a progressive reduction of GFR as a result of loss of functional filtration surface area, a process that is assumed to be driven by fibrosis (5,8). Even though sPRO-C6 is a predictor of poor outcome, no causal relationship between PRO-C6 and renal disease has been shown. Recent studies have indicated that collagens are more than just structural entities and can interact with their immediate surrounding and distant sites through the release of bioactive fragments, such as endotrophin, that can stimulate signaling pathways (19,29,30). Endotrophin has important biological effects, such as attracting macrophages, enhancing tumor growth factor-β signaling, and promoting fibrosis and metabolic dysfunction (31). Biomarkers reflecting pathological alterations to the ECM may identify patients with fast progression of CKD who may be superior responders to treatment (6,9,29). In this cohort, we were able to demonstrate that COL VI formation and COL III degradation act as an early marker for poor prognosis. This is important because therapeutic interventions are more likely to be effective during the early period of kidney disease because the reversibility of fibrosis in advanced stages becomes doubtful as a result of varied processes, such as cross-linking of the ECM (32).

Strengths of the study include that the cohort represents 20% of the patients with T1D followed in our outpatient clinic at Steno Diabetes Center Copenhagen, thus representing a broad segment of the population of adult patients with T1D in the region. Moreover, the cohort is large and well-described, and all stages of albuminuria are represented. The cohort therefore consists of patients from normal to moderate stages of kidney disease. A limitation is that the measured biomarkers are not unique for renal tissue, and fragments from other tissues and organs could affect the levels.

In conclusion, higher levels of serum markers representing COL VI formation (PRO-C6) and COL III degradation (C3M) were independently associated with the presence of diabetic kidney disease. Moreover, higher sPRO-C6 was an independent predictor of mortality and adverse renal outcome, and higher uPRO-C6 and uC3M were associated with a lower risk of decline in eGFR in patients with T1D.

Acknowledgments. The authors thank all the participants of the PROFIL cohort and acknowledge the work of study nurse Lone Jelstrup and laboratory technicians Anne G. Lundgaard, Berit R. Jensen, Tina R. Juhl, and Jessie A. Hermann, employees at Steno Diabetes Center Copenhagen, Gentofte, Denmark.

Funding. The authors thank The Danish Research Foundation and the Innovation Fund Denmark for financial support.

Duality of Interest. D.G.K.R., S.H.N., M.A.K., and F.G. are full-time employees of Nordic Bioscience. Nordic Bioscience is a privately owned, small- to medium-sized enterprise partly focused on the development of biomarkers. M.A.K. and F.G. hold stocks in Nordic Bioscience. The patent for the ELISAs used to measure PRO-C6 and C3M levels are owned by Nordic Bioscience. The funder provided support in the form of salaries for D.G.K.R., S.H.N., M.A.K., and F.G. Outside this work, P.R. has received research grants from Novo Nordisk and AstraZeneca; has acted as consultant for Novo Nordisk, Bayer, Astellas, Boehringer Ingelheim, AbbVie, and AstraZeneca (honoraria to institution); and has shares in Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.

None of the authors received fees, bonuses, or other benefits for the work described in the article. The funder did not play any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Author Contributions. S.P.-L. analyzed and interpreted the data and wrote the manuscript. D.G.K.R. and T.W.H. helped with analyzing the data and reviewed and edited the manuscript. D.G.K.R. and S.H.N. measured the biomarkers. N.T. and S.A.W. collected all data regarding the follow-up. N.T., S.H.N., M.A.K., and F.G. critically reviewed and revised the manuscript. S.T. performed the original study, including data collection, and reviewed and edited the manuscript. P.R. reviewed and edited the manuscript. P.R. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.