Albuminuria Regression and All-Cause Mortality among Insulin-Treated Patients with Type 2 Diabetes: Analysis of a Large UK Primary Care Cohort

Background: Overt albuminuria (urinary albumin-creatinine ratio [ACR] > 300 mg/g) is an established risk factor for progression of nephropathy and total mortality. However, whether a reduction in ACR translates into a reduction in mortality and/or cardiovascular (CV) events among insulin-treated patients with type 2 diabetes (T2D) in routine practice is currently not known. Methods: We obtained data on a large cohort of insulin users with T2D and nephropathy (baseline ACR ≥300 mg/g) from UK general practices between 2007 and 2014. Their corresponding ACR values after 1year of follow-up were thereafter categorized into: (1) < 300 mg/g (i.e., albuminuria regression) or (2) > 300 mg/g (i.e., nonregression of albuminuria), and the cohort was followed-up for 5 years for all-cause mortality and CV events. Cox proportional hazard models were fitted to estimate the risk of all-cause death. Results: A total of 11,074 patients with insulin-treated T2D met the inclusion criteria. Their mean age was 62.3 (13.6) years; mean HbA1c: 8.7 (1.8) and 53% were male. A total of 682 deaths occurred after a follow-up period of 43,393 person-years with a mortality rate of 16 per 1,000 person-years. Five-year survival was markedly reduced in the group whose proteinuria persisted or progressed (91 vs. 95%; log-rank p value < 0.001). Compared to patients whose ACR levels remained above 300 mg/g, all-cause mortality and CV events were 31 and 27% lower in those whose albuminuria regressed to < 300 mg/g (adjusted hazard ratio [aHR] 0.69; 95% CI 0.52–0.91; p = 0.008 and aHR 0. 73; 95% CI 0.54–0.98; p = 0.041), respectively. Conclusion: In patients with insulin-treated T2D and nephropathy in routine practice, a regression in albuminuria (e.g., via better BP or glycemic control) is associated with a significant reduction in all-cause mortality. Thus, albuminuria is not only simply a risk marker of renal and CV disease but also an independent target for therapy. Albuminuria reduction should be viewed as a goal for renal and CV protection.


Introduction
Albuminuria is a strong predictor of adverse renal and cardiovascular (CV) outcomes in patients with type 2 diabetes (T2D), hypertension, and in the general population [1][2][3][4]. Albuminuria is typically assessed by measuring urinary albumin to creatinine ratio (ACR). ACR levels between 30 and 300 mg/g represent moderately increased levels of albuminuria, known as microalbumin-uria, while levels of more than 300 mg/g are associated with overt proteinuria. Several therapeutic strategies are available to reduce ACR, namely, via interruption of the renin-angiotensin system (RAS) with either angiotensinconverting enzyme (ACE) inhibitors or angiotensin II receptor blocker [5,6]; strict BP control; strategies to achieve tight glucose control [7]; and more recently, the sodium glucose co-transporter (SGLT)-2 inhibitor [8,9]. While these strategies are also associated with improvement in renal and CV outcomes, the precise impact of ACR reduction, independent of conventional CV risk parameters, in mediating the beneficial effect of cardio-renal outcome remains unclear.
For many patients with T2D, insulin treatment will be required to control hyperglycemia and to reduce the risk of long-term vascular complications in patients with T2D [10][11][12]. However, insulin therapy is known to induce ∼4 to 9 kg weight gain in the first year of treatment. [13] This is relevant within the context of diabetic nephropathy since obesity is also a significant risk factor for the appearance of proteinuria and ESRD [14]. Furthermore, recent evidence from randomized controlled trial, epidemiological and observational studies have implicated insulin therapy in patients with T2D with increased CV risk and mortality of [15][16][17][18], possibly due to weight gain, recurrent hypoglycemia, potential adverse effects of iatrogenic hyperinsulinemia, as well as a surrogate marker of increased diabetes duration [19,20]. Thus, a cohort of insulin-treated patients with T2D represents a complex heterogeneous challenging group of patients, many of whom have significant comorbidities and high CV disease risk.
Although previous studies have shown that the reduction in albuminuria achieved in the first months of RAS blockade predicts long-term renal and CV risk reduction [21,22] -implying a causal association between albuminuria with cardiorenal risks [23], no previous study has been performed to investigate the independent effects of reduction in albuminuria on CV events and total mortality among insulin-treated patients with T2D in real-world primary care, given that insulin initiators usually have poor glycemic control, longer disease duration or may be more advanced in age, and a reduction in ACR among them could be associated with marked reduction in these events.

Study Design
Using a large UK Primary Care database -The Health Improvement Network (THIN) Database -we conducted a historic cohort study among patients with T2D currently on insulin. THIN is a large UK electronic Primary Care database with clinical details of over 12.4 million patients, of which about 3.61 million are active. Data are imputed with the longitudinal records from about 587 General Practices, which are obtained and updated regularly.
The routine clinical information of these patients is constantly and systematically entered into this database by trained doctors and specialist nurses. Data of specialist or primary care consultations, diagnoses, laboratory results, prescriptions, referrals, hospital admissions, immunizations, to important clinical measures such as body weight, height, and body mass index, as well as information on the patients' demography (e.g., age, and gender), lifestyle characteristics (e.g., alcohol use and smoking), and socioeconomic status (measured by the Townsend deprivation scores) are also included.
THIN database has been validated by various studies and shown to be demographically representative of the UK population in terms of indices of diseases and patients' demography [24]. Our research group has extensively used the THIN database in evaluating diabetes-related outcomes in routine clinical practice [25,26].

Study Participants
We obtained routine clinical data on 11,074 patients with a diagnosis of T2D who met our inclusion criteria. These were adults (aged 18 years and above) who were on insulin therapy between December 2006 and May 2014, irrespective of the use of other glucose-lowering therapies (GLTs). These patients must have recorded values of ACR at baseline and 1 year after insulin initiation. Only those with nephropathy (ACR levels above 300 mg/g) were included. Urinary ACR in the dataset was measured from a single voided urine sample by a central laboratory, with the lowest detectable and reportable level of 1.0 mg/g.
Excluded from the study were those with missing baseline data of ACR and those with medical codes for type 1 or gestational diabetes, or other forms of diabetes, alongside those with no continuous regular prescriptions for insulin in their records for more than 6 months.

Follow-up and Endpoints
The selected study participants with nephropathy at insulin initiation (baseline date) were followed-up from this date for 1 year. Post-1 year ACR levels were estimated, and patients were grouped into 2: those with ACR levels now < 300 mg/g vs. those with ACR ≥300 mg/g. Patients were then followed-up from this point until occurrence of the primary or secondary endpoint, or loss to followup, or discontinuation of insulin therapy, or at the end of the 5-year follow-up period.
The primary endpoints were all-cause mortality and CV events (a composite of nonfatal myocardial infarction [MI] and stroke).
The secondary endpoint was a 3-composite of major adverse CV events (MACE), which includes all-cause mortality or nonfatal MI or stroke, and the component parts of non-fatal MI and stroke.
All these outcomes were identified using their appropriate Read Codes in the database.

Baseline and Endpoint Characteristics
To be able to adjust for the effect of possible a priori confounders on the study endpoints; data were also obtained on important baseline clinical covariates. Among these were demographic variables such as age, gender, socioeconomic status, alcohol and smoking status; important clinical measures such as body weight, height, SBP, and DBP; biochemical parameters, for DOI: 10.1159/000496276 example, baseline HbA1c, lipid-profile, use of other medications including other GLTs; as well as comorbidity status, duration of diabetes treatment, and duration of insulin use. We computed the change in mean arterial pressure and HbA1c after 1 year of insulin initiation; and alongside baseline covariates with significant differences between the 2 groups, these were adjusted for in the final Cox model.

Statistical Analysis
Multiple imputations using the chained equation model were used to input missing data for some important baseline clinical covariates such as weight, HbA1c, eGFR, weight, SBP, and DBP, which were found to be completely missing at random.
Baseline data were summarized for the 2 groups using mean with SDs for continuous variables and absolute numbers with proportions (%) for categorical variables. The differences in baseline characteristics between the 2 groups were estimated using Pearson's chi-square test for categorical and Student t test for continuous variables.
Kaplan-Meier survival curves were estimated separately for both treatment groups. From these survival functions, we calculated the absolute reduction in the probability of an event occurring within the 5-year follow-up period.
Using the Cox regression model, the marginal hazard ratios (HRs) were estimated in order to quantify the adjusted HR (aHR) of an event occurring in the "ACR below 300 mg/g" group compared with the "ACR 300 mg/g and above" group. We tested for violations of the proportional hazard assumption of the Cox regression model, by adding an interaction term of the predictor; and by log-minus-log survival curves; and finally confirmed the proportional hazards assumptions were through Schoenfeld residuals test.
Point estimates were computed with 95% CIs at the conventional statistical significance level of 0.05. Stata Software version 15 was used for all the analyses.

Ethical Approval
This was obtained from the South-East Research Ethics Committee, UK.

Patient Characteristics
A total of 11,074 new insulin initiators met our inclusion criteria. Of this, 1,552 (14%) had reduction in albuminuria, as against 9,522 (86%) who had persistent or progressive albuminuria after 1 year of intensive glucose control with insulin.
Their overall mean age was 62.3 ± 14 years; mean baseline HbA1c and eGFR were 8.7 ± 1.8% and 59.9 ± 21.2 mLs/min/1.73 m 2 , respectively. Table 1 is a summary of the baseline characteristics of the study participants, stratified by the 2 treatment groups.
From Table 1, it can be seen that at baseline, those with their post-1 year ACR less than 300 mg/g were younger (p < 0.001), less obese, and overweight (p = 0.002); had less comorbidities (p < 0.05); shorter duration of diabetes before insulin initiation (p = 0.004); higher eGFR (p < 0.001); and lower systolic and diastolic BP (p < 0.05). Conversely, the use of other GLTs, baseline HbA1c, and lipid profile were similar in both groups (all p ≥ 0.05).

Primary Endpoint
All-Cause Mortality: Survival analyses at 5 years were 91 vs. 95% for patients with ACR above 300 mg/g (group 1) vs. those with ACR below 300 mg/g (group 2); log-rank test p value < 0.001 (Fig. 1). Overall, there were 682 deaths with a crude incidence rate of 15.7 per 1,000 person-years (95% CI 14.6-17.0) within a total follow-up period of 43,393 person-time. There were 621 vs. 61 deaths in group 1 vs. 2 with unadjusted mortality rates of 16.7 vs. 9.7 per 1,000 person-years ( Table 2).

Secondary Endpoint
Composite MACE: The probability of survival for composite MACE fell from 98% in both groups at year 1 to 84 vs. 90% in groups 1 and 2, respectively, at 5 years (log-rank p value < 0.001) Figure 3a. Overall, there were 1,173 composite MACE (1,062 in group 1 vs. 111 in group 2) with a crude event rate of 30.7 per 1,000 person-years (95% CI 30.0-32.5) Table Figure 3b and c. In both groups, there was a total of 69 and 419 events of nonfatal MI and stroke, respectively (event rates = 1.6 vs. 10.6 per 1,000 person-years). The events between the 2 treatment groups are summarized in Table 3.

Discussion
The present study showed that albuminuria is an independent marker for subsequent CV events and total mortality in insulin-treated patients with T2D. Regression of albuminuria as a result of multifactorial intervention in routine clinical care was associated with a further significant reduction in CV events and total mortality events. This association appears to be marginally stronger for total mortality than for CV endpoints. In addition, we found that regression of albuminuria was associated with reduction in nonfatal MI (though not statistically significant) but, interestingly, was associated with a significant reduction in stroke event, compared with the cohort whose albuminuria level either increased or did not regress. Thus, levels of albuminuria should be considered not only as an important risk marker but also an important therapeutic target for CV and mortality prevention in patients with T2D and should be a key consideration when determining drug choice irrespective of blood pressure and glucose levels. Albuminuria has long been recognized to be a risk marker for the severity of kidney disease. Although early opinion suggests that albuminuria is simply a surrogate marker of renal injury, evidence in the last 12 years, however, has shown the cause-effect relationship between albuminuria and progressive kidney damage [21][22][23]. Our present study, obtained in a large retrospective cohort of patients with insulin-treated T2D undergoing routine care in UK general practices, shows that this phenomenon may also apply to CV event and total mortality. This is akin to the impact of blood pressure and serum cholesterol, where therapeutic strategies have been designed and with the aim to lower blood pressure and serum cholesterol, respectively. Indeed, there are recognized therapeutic strategies that can reduce the degree of albuminuria, namely, antihypertensive agents such as ACE inhibitor and angiotensin-2 receptor blocker [5,6], tight glucose control [7], SGLT2 inhibitors [8,9], and low protein diet [27]. Since RAS blockade and SGLT2 inhibitor also lower blood pressure, it remained speculative whether the changes in albuminuria per se affect the CV/mortality endpoints independently of blood pressure. However, individual variations to RAS blockade are well described; that is, patients can have a systolic blood pressure reduction without a simultaneous reduction in albuminuria or Estimated probability of survival 8,564 1,409 No at risk (failure) ACR 300 and above ACR below 300  vice versa [28]. This discordance in response to RAS blockade has been reported in clinical trial [29] as well as in population treated in routine clinical practice [30] and that the beneficial impact of albuminuria reduction was reported to be independent of blood pressure reduction. A particularly important finding from this present study was the fact that the CV and total mortality reduction were not observed in the cohort whose albuminuria did not regress, following adjustment for conventional CV factors. The current study for the first time shows on a large scale, in a real-world practice, that the degree of albuminuria reduction is directly related to the subsequent CV and mortality protection in a high-risk group of patients with T2D.
In the absence of a prospectively designed clinical study, with appropriate treatment arms and robust endpoints, the mechanism linking albuminuria and excess mortality remain speculative. The steno hypothesis suggested that urinary protein excretion reflects generalized vascular endothelial dysfunction [31], for example, increased circulating von Willebrand factor antigen released [32] and nitric oxide inhibition [33] in response to endothelial cell damage. In addition to von Willebrand Factor, soluble vascular cell adhesion molecule, fibrinogen, high platelet adhesiveness, erythrocytes aggregation, and tissue plasminogen activator have been found to correlate with urinary albumin excretion [34], especially in patients with diabetes, indicating increased thrombosis risk. Proteinuria is also linked with insulin resistance, a recognized marker and mediator of atherogenesis [35]. More recently, a key mechanism that contributes to the link between albuminuria and adverse CV outcome relates to the loss of the glycocalyx -a polysaccharide gel that lines the luminal endothelial surface and that normally acts as a barrier against albumin filtration [36]. Degradation of the glycocalyx in response to endothelial activation can lead to albuminuria and subsequent vascular inflammation, thus providing a pathophysiological framework for the clinical association of albuminuria with renal and CV disease progression.
The main strength of our study derives from the inclusion of a large cohort of patients with T2D receiving insulin therapy in a real-world population, which is largely representative of the UK population. This implies that our findings will be generalizable to various population that share similar demographics. The large cohort of patients studied here provides adequate statistical power, which also enabled the component endpoints to be studied. It also contains information on other time-varying covariates to adjust for possible confounders. We adjusted for a large set of factors that could have differed at the baseline. Nevertheless, some residual confounding in our study could persist. For example, our classification of albuminuria was largely based on a single measurement, in contrast to current recommendation, in which at least 2 measurements are required. In addition, as is the case in all studies of CV or ESRD risk associated with eGFR and albuminuria, the effect of competing hazards may bias estimates of risk. This is because elevated ACR and low eGFR are also risk factors for nonrenal diseases; associated differential mortality in high-risk individuals may confound HR estimates for CV events. Lastly, changes after baseline in medications were not evaluated in this analysis and therefore cannot account for any differences that might influence the association between ACR and outcomes.
In summary, the findings of this study extend the cause-effect relationship between albuminuria and CV risks and that suppressing albuminuria should be an important target of therapy to achieve optimal CV protection in high-risk individuals with T2D. While further prospective interventional studies are required to clarify the cause-effect relationship, we would suggest that CV risk reduction guidelines on patients with T2D should not only view albuminuria as an important risk factor/ marker but should also define albuminuria regression as a target for therapy, similar to lipids, blood pressure, and glucose targets.

Disclosure Statement
None declared for all authors relating to the content of this manuscript.

Funding Source
This work was supported by the Medical Research Council (grant numbers MR/K00414X/1 and MR/P021220/1) and Arthritis Research UK (grant number 19891).

Research in Context
What is already known about this subject? -Albuminuria is a strong predictor of adverse renal and CV outcomes in patients with T2D, hypertension, and in the general population.
-Several therapeutic strategies are available to reduce ACR, namely, via interruption of the RAS with either ACE inhibitors or angiotensin II receptor blocker; strategies to achieve tight glucose control; and the use of SGLT-2 inhibitor.
-Previous subanalysis of studies has shown that the reduction in albuminuria achieved in the first months of RAS blockade predicts long-term renal and CV risk reduction.
What is the key question? -Among patients with insulin-treated T2D with overt proteinuria (ACR > 300 g/g), undergoing routine treatment in primary care, does regression of proteinuria confer an associated reduction in CV events, and total mortality compared with patients whose proteinuria did not regress?
What are the new findings? -Among this cohort of patients with insulin-treated T2D, increased albuminuria is an independent marker for subsequent CV events, and total mortality in real-world setting.
-Regression of albuminuria as a result of multifactorial intervention in routine clinical care was associated with a further significant reduction in CV events and total mortality events compared with patients whose albuminuria did no regress.
-The beneficial impact of albuminuria regression appears to be stronger for total mortality than for CV endpoints, regression of albuminuria is not associated with a significant reduction in nonfatal MI but, was associated with a significant reduction in stroke event, compared with the cohort whose albuminuria level either increased or did not regress.
How might this impact on clinical practice in the foreseeable future? -Thus, levels of albuminuria should be considered not only as an important risk marker but also as an important therapeutic target for CV and mortality prevention in patients with T2D and should be a key consideration when determining drug choice irrespective of blood pressure and glucose levels.