Diabetes mellitus (DM) is one of the most important health problems worldwide. In recent years, the prevalence of type 2 DM has increased in parallel to obesity and unhealthy lifestyles. In Spain, between 6% and 10% of the general population has DM. Between 6.3% and 7.4% of the Spanish healthcare budget is spent on diabetes care; this represents an annual cost of €1290–1476 per patient.1,2

Since the prevalence of DM has increased to epidemic proportions, the number of renal biopsies is also increasing. The lesions observed in these biopsies confirm either isolated diabetic nephropathy (DN), DN with non-diabetic nephropathy (NDN) or isolated NDN. Identification of these conditions affects patient's management and prognosis, especially if the diagnosis of NDN leads to a change in a patient's therapeutic strategy.3,4 Most prevalent NDN conditions in patients with DM are, focal and segmental glomerulosclerosis, IgA nephropathy and the membranous nephropathy.5 Therefore, research work has been conducted to find clinical factors that predict NDN in this population. Previous studies have observed that level of proteinuria, absence of diabetic retinopathy (DR), higher levels of haemoglobin, shorter duration of DM, acute renal failure at the time of renal biopsy, microhaematuria, DM treated with oral hypoglycaemics, lower blood sugar and less intima-media thickness are risk factors for NDN.3,5–9

Several studies have observed that renal survival was decreased in diabetic patients diagnosed of DN compared with patients with NDN.10,11 Moreover, the importance of renal biopsy in diabetic patients lies not only in the fact of differentiating DN from NDN, but also in studying histological lesions within DN.12 Amongst the different types of DN, the most advanced (type IV) presents higher proteinuria with a worse renal prognosis (RRT).13

The main aim of our study is to evaluate the prevalence and types of NDN in a consecutive cohort of biopsied diabetic patients and to find out whether there are factors that predict NDN. In addition, we plan to study whether there are differences in renal and patient survival according to the diagnosis obtained from the renal biopsy.

Isolated DN was detected in 38 of 110 diabetic patients biopsied. In the rest (72 patients), NDN was detected. The most common diagnosis of NDN was IgA nephropathy (n=9; 13.2%), followed by benign nephrosclerosis (n=6; 8.8%), focal and segmental glomerulonephritis (n=5; 7.4%), membranous glomerulonephritis (n=4; 5.9%), chronic interstitial nephropathy (n=4; 5.9%) and other less common ones (Fig. 2). 13.6% (n=15) of the renal biopsies were unclassifiable due to insufficient sample for diagnosis.

Fig. 2.

Diagnoses from renal biopsy: distribution of the number of patients according to diagnosis in renal biopsy. F&S: focal and segmental; GN: glomerulonephritis; MPGN: membranoproliferative glomerulonephritis; NAS: nephroangiosclerosis; AIN: acute interstitial nephritis; CIN: chronic interstitial nephritis; ATN: acute tubular necrosis.

(0.16MB).

Predictive factors for non-diabetic nephropathy

In order to identify predictive factors of NDN, a multivariate binary logistic regression analysis (Table 3) was performed. Risk factors included a high level of serum creatinine, age of the patient, less proteinuria in 24h collection, shorter duration of diabetes and no DR (Table 3). Although we found significant differences in presence of haematuria (DN 9 vs. 32 cases) in the multivariate analysis, it was not significant.

Table 3.

Predictive factors for non-diabetic nephropathy in diabetic patients.

Parameter	OR	CI (95%)	Lateral significance (p)
Creatinine (mg/dL)	1.48	1.01–2.17	0.04
Proteinuria (g/24h)	0.81	0.68–0.97	0.03
Duration of DM (years)	0.992	0.987–0.998	0.004
Age (years)	1.07	1.01–1.13	0.02
Retinopathy (yes/no)	0.23	0.07–0.81	0.02

Multivariate binary logistic regression analysis. Dependent variable: non-diabetic nephropathy.

95% CI: confidence interval of 95%; OR: odds ratio.

The discriminative ability of the model showed an ROC curve 0.805 (95% CI: 0.708–0.902) (Fig. 3).

Fig. 3.

ROC curve: the model's discriminative ability showed an ROC curve with an area under the curve of 0.805 (95% CI: 0.708–0.902).

(0.06MB).

An analysis using ROC curves for each of the variables was performed, showing the cut-off point of each one with greater specificity and sensitivity. We therefore obtained an increased risk of NDN when creatinine was greater than 1.58mg/dL, proteinuria under 3.15g/24, age over 53 years and duration of DM under 10 years.

Renal prognosis and survival

A survival analysis of patients with diabetes and renal prognosis (need for RRT) was performed. The need for RRT was 33%. It was twice as much in NDN compared with DN (11% n=4 vs. 22% n=16), without statistical significance. At the time of renal biopsy, one patient (2.6%) in the DN group and 9 (12.5%) in the NDN group started haemodialysis. During the follow-up period, 5 patients in the DN group required RRT (13.2%), 20 in the NDN group (29.4%), and none in the group with both conditions. Overall patient mortality was 23% (n=25), 5% (n=5) of these in the DN group and 18% (n=20) in the NDN group. No deaths were observed in the DN+NDN group in the follow-up period. No differences in renal or patient survival, or between cohorts studied, were observed (Fig. 4A and B).

Fig. 4.

Renal and patient survival curves in all groups studied. Analysis using Kaplan–Meier curves of the different groups divided according to renal biopsy diagnoses. (A) Start of RRT. (B) Mortality. No statistically significant differences were observed between the groups. DN: diabetic nephropathy; NDN: non-diabetic nephropathy; RRT: renal replacement therapy. Renal and patient survival curve according to proteinuria in patients with DN. Analysis using Kaplan–Meier curves of the different groups of diabetics with diabetic nephropathy divided according to range of proteinuria (nephrotic or non-nephrotic). (C) Start of RRT. (D) Mortality. Statistically significant differences (p=0.004) were observed at start of RRT, observing a higher incidence in the group with nephrotic proteinuria. As for patient survival, no statistically significant differences were observed.

(0.31MB).

Next, the group of patients diagnosed with DN based on Fioretto's scheme was analysed, and no differences between the various categories of renal and patient survival were observed.

Proteinuria at the time of biopsy in patients with DN was then studied (n=41). 58.5% (n=24) had nephrotic-range proteinuria (≥3.5g/24h) and 39.5% (n=17) had non-nephrotic proteinuria. 41.5% (n=17) required RRT. As expected, the patients with nephrotic-range proteinuria had an increased incidence of end-stage kidney disease as compared with the group of patients with non-nephrotic proteinuria (45.8% n=11 vs. 6.6% n=1; p=0.004). No statistically significant differences in patient mortality were observed (12.5% n=3, vs. 11.8% n=2; p=NS) (Fig. 4C and D).

In our population of patients with DM and with a renal biopsy, the majority had NDN. The most common diagnosis of NDN was IgA nephropathy. Elevated serum creatinine, lower degree of 24h proteinuria, shorter duration of diabetes, absence of DR and older age were risk factors for NDN in diabetic patients in whom a renal biopsy was performed. No differences in patient survival or renal prognosis were observed based on the result of the renal biopsy. Patients with nephrotic proteinuria had a worse renal prognosis as compared with the group with non-nephrotic proteinuria.

Sharma et al.3 studied 620 patients with DM biopsied at 2 centres in the United States, and found that 37% showed isolated DN, 26.4% DN and NDN, and 35.4% isolated NDN. In our study, the percentage of patients with isolated DN was similar (34.5%); however, 61.8% had isolated NDN, and the percentage of combined diagnoses of DN and NDN was minimal.

In patients with diabetes, a renal biopsy is usually performed when NDN is suspected.15,16 The most common reasons for indicating renal biopsy in our series were similar to those already known and published: impaired renal function with no apparent cause, presence of microhaematuria, increased proteinuria in the absence of DR, presence of nephrotic syndrome, increased proteinuria with DM of less than 5 years’ history and presence of nephritic syndrome.17–19

As for the results in the renal biopsy, in the studies published, the conditions most frequently found in diabetic patients with NDN are: IgA nephropathy10,11,18,20–22 and interstitial nephritis7,17,19,23 (see Table 4). In our cohort, the most common condition of NDN was IgA nephropathy. However, Sharma et al., who analysed the largest cohort to date with a total of 620 patients, found that the most common finding was acute tubular necrosis. These results may be attributed to the high rate at which renal biopsies are indicated for acute renal failure.3

Several studies have focused on analysing the risk factors for NDN in biopsied diabetic patients and confirmed a shorter duration of diabetes,3,6,7,10,11,17,18,24 lower blood pressure,17 absence of retinopathy,7,10,18,22,23 elevated haemoglobin,7,18 lower glycosylated haemoglobin,18 presence of microhaematuria,9,10,17,18,22 elevated proteinuria,22 decreased serum creatinine,10 or treatment with oral hypoglycaemics.11,24 In other studies, however, NDN was associated with the presence of acute renal failure23 or lower proteinuria.9,10 In our study we observed that NDN risk factors are elevated serum creatinine and advanced age.

We found that independent protective factors against development of NDN were, low degree of proteinuria over 24h, the long duration of diabetes and the presence of DR. Our model obtained an ROC curve with an area under the curve greater than 0.80, thus confirming its reliability. Liang et al. performed a meta-analysis that included a total of 63 studies with 2322 patients to study the risk factors for developing NDN. As with our study, they found that absence of DR, shorter duration of DM and a low level of glycosylated haemoglobin were predictors of NDN. Moreover, in their study, lower blood pressure and lower serum cholesterol also proved to be predictors of NDN.25

The presence of microhaematuria is risk factor of NDN in biopsied patients with DM.9,10,17,18,22 In our study, although the number of patients presenting with haematuria in the NDN group was higher than in the DN group; however by multivariate analysis this factor was not significant as an independent factor in developing NDN.

Several studies have focused on the renal prognosis in diabetic patients with DN vs. NDN.7,10,11,26 In these studies, patients with renal NDN had a better prognosis in terms of progression to chronic renal disease requiring RRT. Our results are the opposed since in our cohort, biopsied diabetic patients diagnosed with NDN had a tendency towards a worse renal prognosis.

In our patients at the time of the biopsy, the creatinine level in the NDN group was higher than in the DN group. This is in contrast to previous studies7,11 in which patients with DN the initial renal function was worse than with NDN. In one of these studies, although the creatinine level in the NDN group was higher than in the DN group, renal prognosis was worse in the DN group.10 One possible explanation might be that the DN group had a higher degree of proteinuria compared with the NDN group, although the differences were not significant.

In the study of differences in patient survival in the NDN group vs. the DN group, no differences were observed. To date, there are no studies focused on analysing survival in diabetic patients based on the diagnosis from a renal biopsy.

One-third of the patients in our cohort were diagnosed with DN. The natural history of DN in type 2 DM is not well established, unlike that of type 1 DM. Changes in indolent glucose metabolism up to the diagnosis of DM are used to establish various periods of time from change to change. This is why at the time of diagnosis of type 2 DM, DN may manifest as micro- or macroalbuminuria.

However, 20 years after the onset of microalbuminuria, only 20% will progress to chronic kidney disease.1 Risk factors for progression to end-stage kidney disease in diabetic kidney disease have been described, and include a family history of DM, smoking, blood glucose and lipid levels, blood pressure and urinary albumin excretion, as well as new emerging biochemical markers.27

In 2010, Taervert et al. performed a histological classification of DM in which a total of 5 classes were differentiated, ranging from absence of lesions under a light microscope to diffuse glomerulosclerosis.28 By specifically analysing which kind of lesions are evident based on degree of proteinuria, it has been shown that an increase in the mesangial area occurs when the degree of proteinuria increases.12 In this regard, in our study, the most common histological lesion was diffuse mesangial expansion. In addition, we observed that patients with DN having a renal biopsy with nephrotic proteinuria showed greater nodular expansion compared with those with non-nephrotic proteinuria in the population diagnosed with DN. All of this is consistent with previous results published by others.12

Several studies have examined the difference in renal prognosis, as classified by Taervert, in patients with DN. Amongst them, Mise et al., with a total of 205 biopsies with a DN diagnosis, and then An et al., with a total of 396 biopsies, showed that patients who in their renal biopsy have DN class IV had a worse renal prognosis than the other classes.13,29

In addition, renal prognosis was analysed according to degree of proteinuria. As with the results of the two studies mentioned above, in our cohort we saw that patients with nephrotic proteinuria have a worse renal prognosis, which probably corresponds to nephropathy class III–IV, where a higher degree of proteinuria is observed. Shimizu et al., in a cohort of 260 diabetic nephropathy cases, analysed the difference in renal prognosis depending on degree of proteinuria, finding that the greater the degree of proteinuria, the higher the incidence of renal events in patients with a glomerular filtration rate greater than 60ml/min as well as in those with less than 60ml/min.30 All of this is consistent with our results. In the same study mentioned above, the differences in mortality were also analysed according to degree of proteinuria, and an increase in mortality with proteinuria in patients with a glomerular filtration rate less than 60ml/min/m2 was observed. However, this increase was not observed in patients with a glomerular filtration rate greater than 60ml/min/m2. In contrast, in our study we observed no differences according to glomerular filtration rate or serum creatinine level.

Finally, biopsies were classified according Fioretto's scheme (categories CI, CII and CIII), but there were no significant differences in either patient or renal survival.

A significant percentage of patients with diabetes (2/3) presented with NDN in renal biopsy. Amongst diabetic patients with renal involvement it is necessary to detect those with NDN, as they have a different prognosis for kidney disease and may benefit from different therapeutic strategies. Classically and according to the guidelines for managing diabetic patients, renal biopsy is indicated when the presence of another condition is suspected according to the clinical course and serum and urinary lab data.31 However, a better understanding of diabetic patients treated by renal biopsy and their clinical and laboratory characteristics may help in decision-making in routine clinical practice. The prognosis for DN differs depending on the type of lesions found in a renal biopsy. Few studies have focused on the study of routine renal biopsy in patients with DM. We believe that further studies are necessary to elucidate the importance and renal prognosis of biopsies in diabetic patients with suspected DN.

SB and MJS have also contributed to the study.

The authors declare that they have no conflicts of interest.