Introduction

The ever-increasing global prevalence of diabetes mellitus, which was estimated to affect over 415 million people worldwide in 2017 [1], is giving rise to serious concern among healthcare providers. Diabetic nephropathy (DN) is a major complication associated with both type 1 and type 2 diabetes (T1DM and T2DM, respectively) and is the leading cause of end-stage renal disease (ESRD) [2]. DN follows distinct phases, wherein glomerular hyperfiltration is followed by a relentless decline in renal function, typically occurring over a 15- to 20-year period [3]. The development of ESRD requires the patient to receive dialysis or undergo renal transplantation, two procedures which are associated with excess morbidity and mortality [4]. The current standard treatment regimen for patients with T2DM involves lifestyle modifications and medical treatment targeted against the fundamental dysregulation of glucose and hypertension [3, 5], but this strategy is unable to affect the underlying pathophysiology of the DN. Although the results of many studies indicate a correlation between the degree of hyperglycemia and progression of DN, diabetic patients receiving intensive glycemic control therapy continue to develop DN. Hyperglycemia can in fact induce modifications in gene expressions which persist even after normoglycemia is restored through a process known as metabolic memory [6]. In a recent study, the risk of development of kidney complications was correlated with a specific cluster of diabetic patients with insulin-resistance, leading the authors to suggest that glucose-lowering therapy is not the optimum management strategy for preventing this complication. Hence, there is a need to focus on new therapeutic targets and initiate treatments at an early stage in order to prevent complications [7]. Treatment options and decisions are often based on the results of clinical trials that meet the highest standards of scientific rigor and ethical oversight [8]. The specific aim of phase III clinical trials is to confirm results obtained in previous experimental trials; as such, phase III clinical trials must test experimental study drugs or treatment in larger populations in order to confirm the effectiveness and safety of use of the drug(s) under study (https://www.fda.gov/). To realize the benefits of a clinical trial, the results must be shared quickly after the study has concluded [9]. However, timely dissemination of clinical trial results continues to be a serious issue. Since favorable results of intervention by drugs are twofold more likely to be published than negative or unfavorable results [10], the efficacy of a drug may be overestimated by the medical community, and trials may be unnecessarily repeated. Beyond the impact on treatment decisions, however, there is an explicit ethical obligation to publish towards study participants, as mandated by the Declaration of Helsinki. Therefore, the non-publication of trial outcome data is against ethical obligations that investigators have towards study participants. In this context, the aim of our study was to assess the public availability of results of phase III clinical trials on DN. Since treatment options and decisions are often based on clinical trials, knowledge on current therapies and their outcome is of utmost importance.

Methods

The analysis was performed according to STROBE (STrengthening the Reporting of Observational studies in Epidemiology) criteria. This article is based on previously conducted studies and does not contain any studies with human participants or animals performed by any of the authors.

Clinical Trials Search

For the cross-sectional analysis, we searched the Clinical Trials.gov registry of clinical trials in June 2017 for clinical trials on DN, with the added restriction of only completed phase III trials. The search was performed by entering the keywords “diabetic nephropathy” and “diabetic renal disease” in the ClinicalTrials.gov search engine. Data on studies obtained from the database were organized in a spreadsheet for analysis. The data available on ClinicalTrials.gov included National Clinical Trial (NCT) number, study title, study description, study design, eligibility criteria, enrollment, arm and interventions, outcome measures, primary completion date, and availability of study results. Following the evaluation of these parameters, we excluded studies that did not meet the following exclusion criteria: no diabetic patients investigated and/or intervention not relevant for diabetic kidney disease (DKD).

Publication Search

A trial was considered to have been published when the results were present in the ClinicalTrials.gov registry or when a journal had published a peer-reviewed manuscript online or in print that included primary or secondary outcome data from the trial in question. When the ClinicalTrials.gov registry did not provide results or links to publications in peer-reviewed journals, we searched the PubMed or/and Google Scholar databases for articles using the study identification number (NCT), the study title, and other study identification numbers. When no published results were found on these latter two databases, the principal investigators (PIs) or sponsors were contacted by email and asked to provide either an article with the study results, which we might have missed, or the reason for the failure to publish the results. Feedback on the missing publications and available data on unpublished clinical trials were analyzed. Clinical trial results that could not be obtained by the preceding described procedure were assessed as unpublished.

Time to Publication

Time to publication refers to the period of time between the primary completion date of the clinical trial and the date of publication of the results either on the ClinicalTrials.gov registry or in peer-reviewed journals. The calculation of the time to publication was performed in accordance with the Food and Drug Administration Amendments Act of 2007 (FDAAA) which requires the publication of results within 1 year after completion of the trial [11] and, therefore, our analysis regarded only studies completed before 2016.

Absolute Risk Reduction Analysis and Patient Number Needed to Treat

The absolute risk reduction (AAR) is the change in the risk of an outcome of a given treatment or activity in relation to a comparison treatment or activity. The number needed to treat (NNT) corresponds to the inverse of the absolute risk reduction.

Statistical Analysis

The following continuous or categorical variables were analyzed: NCT number, study title, gender, age, study phase, study type, study design, condition, intervention, recruitment status, primary completion date and completion date, availability of study results, publication date, time to publication, sponsor, and funding source. Standard methods of descriptive statistics were applied. Two-sided p values 0.05 were considered to be statistically significant.

Results

Publication of Clinical Trials

A total of 49 completed phase III clinical trials were identified from the search of the ClinicalTrials.gov registry in June 2017. Of these studies, 15 were excluded from subsequent analysis since they did not include diabetic patients or any intervention for DKD (Fig. 1). Of the remaining 34 studies, 22 were published and 12 were unpublished. The results of seven studies were recorded in the ClinicalTrials.gov registry, with a direct link provided between these studies and publications in peer-reviewed journals of 11 other studies. Publications on two studies were identified by searching the PubMed/Google Scholar databases, and in two cases, published manuscripts were sent directly to the authors by the PIs. Regarding the unpublished studies, we received answers from six of the 12 PIs or sponsors contacted. Of these, three asserted they were in the process of finalizing the paper or submitting it to journals; two stated that the reasons for failure to publish were “adverse effect” of the testing drug (one case) and “no funds” (one case); and one declared that the results were only available on the sponsor’s website. The FDAAA requires the results of clinical trials to be published within 1 year after the completion of the study; thus, in accordance with the FDAAA, in our analysis of publications we considered only those trials completed before 2016 (n = 32). Of these, 22 studies (69%) had been published, with a mean time to publication of 26.5 (median 23.5, standard deviation [SD] 16.5) months for published studies completed before 2016 and for which the primarly dates were available (n = 18). Thus, only 33% of the studies analyzed met all FDAAA criteria (Fig. 2).

Fig. 1
figure 1

Study flow diagram of the identification of published and unpublished phase III clinical trials on diabetic nephropathy in the ClinicalTrials.gov registry

Full size image
Fig. 2
figure 2

Time to publication of completed phase III clinical trials on diabetic nephropathy (completed before 2016) for which the primary completion dates were available on on ClinicalTrials.gov. Time to publication indicates the number of months between the primary completion date of the clinical trial and the date of publication of the results. FDAAA Timeline mandated by the U.S. Food and Drug Administration Amendments Act of 2007

Full size image

Characteristics of Clinical Trials

The interventions tested were either compared to a placebo control group (20 studies; 59%), to another intervention or to standard care (11 studies; 32%), or to no intervention (3 studies; 9%). Most of the studies (30/34) included renal parameters as primary or secondary outcomes. Overall, 16 different renal parameters were measured to study the effect of interventions on the renal system. Of these, proteinuria and glomerular filtration rate (GFR) were main renal parameters analyzed—in 23 and 14 clinical trials, respectively; progression to ESRD was tested in only three studies (Fig. 3). None of the 34 studies performed a gender-specific analysis; allocation of participants in groups was predominantly randomized (88%). The interventions tested included 26 different categories, with the most represented drug classes being angiotensin-converting enzyme inhibitors (ACEi) and angiotensin-II receptor blockers (ARBs). Data on 11 different interventions, 19 renal outcomes, and more than 1000 patients remained undisclosed (Table 1).

Fig. 3
figure 3

Renal parameters measured as the primary or secondary outcomes in the 34 completed phase II clinical trials on diabetic nephropathy included in our analysis. Bars indicate the number of published (green) and unpublished (grey) studies which measured the outcomes (Y-axis). Four studies did not measure any renal outcome. Proteinuria (single asterisk) was measured by the following different methods: urine-albumin concentration ratio (n = 9 studies); urine protein excretion/24 h (n = 5 studies, urine albumin excretion rate (n = 5 studies), not specified (n = 4 studies). Estimated glomerular filtration rate (eGFR; double asterisk) was measured by the following different methods: creatinine clearance (modification of diet in renal disease [MDRD] study equation or Cockcroft and Gault equation) (n = 9 studies); clearance of iohexol (n = 2 studies); clearance of iothalamate (n = 1 study) ; not specified (n =  2 studies). Inflammatory markers (superscript 1) were: high-sensitivity C-reactive protein; monocyte chemoattractant protein-1 (MCP-1); tumor necrosis factor alpha; interleukin-6; fibrinogen. Endothelial dysfunction markers (superscript 2) were: von Willebrand factor; soluble vascular cell adhesion molecule-1; soluble intercellular adhesion molecule-1; soluble E-selectin. Urine kidney injury markers (superscript 3) were: kidney injury molecule 1; N-acetyl-β-d-glucosaminidase; neutrophil gelatinase-associated lipocalin; liver fatty acid-binding protein. NO Nitric oxide

Full size image
Table 1 Characteristics of the completed phase III clinical trials on diabetic nephropathy that were not published (n = 12)
Full size table

Interventions Efficacy on Renal Outcomes

The effects of interventions on the renal outcomes described in the 22 published studies are summarized in Table 2 [12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32]. Two-thirds (77%) of the published studies versus 25% of the non-published studies compared the interventions to a placebo control group. The most common interventions were ARBs (7 studies) and/or ACEi (5 studies). The cohort size in these studies varied between 22 and 11,140 patients. Whereas most studies included solely patients with T2DM, two studies included patients solely with T1DM [17, 18], and two studies included patients with either T1DM or T2DM [24, 33]. No improvement on renal outcome parameters, such as proteinuria/albuminuria and/or GFR, was reported for most medications [12, 14, 16, 19, 20, 24, 26,27,28,29,30, 32]. Proteinuria/albuminuria was improved only by the addition of ARB treatment to the standard therapeutic regimen [13, 17] or by the addition of vitamin D3 to the standard therapy (study NCT00552409). The addition of sodium–glucose cotransporter 2 inhibitors reduced the urine albumin-to-creatinine ratio (UACR) but not the GFR, but only when added to the standard therapy [31]. Medication with ACEi and a diuretic in addition to standard therapy [25] reduced the risk of DN.

Table 2 Characteristics of the completed phase III clinical trials on diabetic nephropathy that were published (n = 22): effects on renal outcomes
Full size table

Relative and Absolute Risk Reduction

Two published clinical trials obtained significant risk reduction of renal events. Patel et al. described a relative risk reduction of 21% for a combined endpoint (total renal events) [25], and Haller et al. described a relative risk reduction of 16% for new onset of microalbuminuria [26]. For these two trials, absolute risk reduction (ARR) and number needed to treat (NNT) were calculated (Table 3). Combined intervention with perindopril + indapamide in addition to current therapy [25] reduced the relative risk of nephropathy and of new microalbuminuria by 18 and 21%, respectively. This means that 159 patients need to be treated to prevent new onset or worsening nephropathy in one patient (ARR 0.6%), and 25 patients need to be treated to prevent new onset of microalbuminuria in one patient (ARR 4%). In Patel et al.’s study [26], the absolute risk for new microalbuminuria was reduced by 2% with olmesartan; thus, 63 patients need to be treated to avoid the development of new microalbuminuria in one patient. An effect on ESRD was not found in either of these studies.

Table 3 Risk reduction of renal events in two completed phase III clinical trials on diabetic nephropathy
Full size table

Discussion

Since treatment options and decisions are often based on the results of clinical trials, knowledge of the outcome of these studies is of great importance. In our analysis, 31% of the completed phase III clinical trials on DN remained unpublished, which is in line with previous findings on phase III clinical trials on other diseases [34,35,36]. The high number of undisclosed clinical trials may lead to an underestimation of the relevance of DN in the medical literature and thereby hinder a correct risk–benefit assessment of a certain intervention. It is well known that trials showing a benefit of a drug or device have a much greater chance of full publication than trials showing no benefit [37] due to commercial interest and publication strategy of papers since Editors prefer articles that guarantee citations [10]. In addition to publication bias, adverse events are often poorly described. Some studies fail to report the incidences of severe, serious, and fatal adverse events, such as in cancer drug trials [38]. The failure to publish negative results and the underestimation of adverse events lead to an accumulation of literature favoring the benefits of treatments [10].

Further, we found that the time to publication of results was longer than that recommended by FDAAA, with a mean time to publication of 26.5 months compared to the 12 months required by the FDAAA. The effect is a delay in reporting therapeutic strategies. A comparison of results from the various studies, however, remains difficult since study design and the renal outcome parameters vary greatly between the studies.

Overall, the interventions reported in each study, which were aimed at improving renal outcomes, showed low efficacy. The initial clinical evidence of renal involvement in patients with T2DM is usually the appearance of microalbuminuria, which has been defined as a urine albumin excretion rate (UACR) of 30–299 mg/24 h [39]. Patients with diabetes mellitus and microalbuminuria are at high risk of developing overt progressive DN [12]. Urinary albumin concentrations in the upper normal range have been reported to predict both cardiovascular and renal events in both high- and low-risk populations. For these and other reasons, some authors suggest the treatment of urinary albumin excretion as a continuous variable [40]. Proteinuria was measured in many of the clinical trials analyzed (14/34 published trials), indicating that it is used as an important predictor of renal outcome when evaluating DN (Fig. 3). A decline in eGFR was also broadly used to assess the effectiveness of interventions, but only a few studies investigated the risk of ESRD because it requires long-term trials. The measurement of urine protein excretion and eGFR varied greatly among the studies, complicating a reliable comparison of the outcomes [41]. This comparison is further complicated by the fact that current clinical recommendations for the treatment of DN are based on results that in the initial study investigated another primary endpoint (usually glucose therapy), with renal outcome evaluated only secondarily. The DCCT/EDIC study showed that only 24 of 1441 patients developed ESRD after more than 25 years of observation. The ARR was only 1.7% with a NNT of 74 patients due to intensive glucose therapy in patients with T1DM [42]; in patients with T2DM similar results were shown. The mega-trials (ACCORD, ADVANCED, VADT, and UKPDS) only showed weak effects on micro- and macroalbuminuria, with an ARR of 1.0 and 0.3%, respectively, and an NNT of between 91 and 333 patients, without any relevant effect on ESRD by intensive glucose-lowering treatment [43]. In addition, empagliflozin seems to be a new treatment option for DN, but the main effects shown to date are on surrogate parameters, such as creatinine doubling (ARR 1%, NNT 20) and worsening of albuminuria (ARR 5%, NNT 20), while ESRD occurred with an ARR of 0.3% and NNT of 310 [44]. Almost two-thirds of all trials were placebo-controlled, with a higher percentage of placebo-controlled trials in published studies than in unpublished trials (77 vs. 25%, respectively). Placebo-controlled trials produce strong evidence of the effectiveness of a new intervention, limited only by the statistical uncertainty of the outcome [45]. However, knowledge about the relative efficacies between various drugs is also needed for decision-making in clinical practice. In our analysis, only one study [27] compared an intervention to both placebo and another drug(s).

The most represented drug classes in all trials were ACEi and ARBs. Angiotensin-II receptor blockers are renoprotective in hypertensive azotemic patients with T2DM, but their efficacy in early DKD is uncertain. Findings support the current recommendation that inhibitors of the renin–angiotensin–aldosterone system should not be used for primary prevention of DN in normotensive normoalbuminuric persons with diabetes. However, these medications seem to mitigate the progression of DKD when used after the onset of microalbuminuria. ACEi has demonstrated efficacy in reducing cardiovascular risk [27] and, in combination with diuretics, and also shown to correlate with a reduced risk of developing microalbuminuria [25]. However, their effects on preventing DN progression have been less clear, and they have failed to reduce the decline in GFR [21, 23]. A growing body of evidence indicates that the decline in GFR might occur irrespectively of the progression of albuminuria in non-proteinuric DN phenotypes [46]. Altogether, these data call for an early intervention that targets potential mediators of renal dysfunction other than proteinuria to prevent or slow GFR decline already at the stage of normoalbuminuria. In this regard, the role of reactive metabolites [47] and inflammation in the progression to DN is gaining attention. The mechanisms involved are little understood, with evidence of increased inflammatory cytokines (monocyte chemoattractant protein 1 [MCP-1], human tumor necrosis factor alpha [TNF-α)], and mononuclear infiltrates in the glomeruli and tubulointerstitium that would contribute to the progression of DN [45]. Endothelial dysfunction and inflammation markers (von Willebrand factor [vWf], plasma soluble vascular cell adhesion molecule-1 [sVCAM-1], soluble intercellular adhesion molecule-1 [sICAM-1] and interleukin-6 [IL-6]) have been found to be correlated to DN onset in patients with T2DM and microalbuminuria, independently of traditional risk factors [19]. Clinical trials included in our analysis showed poor effects on inflammatory markers. Soy milk showed no significant effect on inflammation [TNF-α, IL-6 and C-reactive protein [CRP]) compared to cow milk [21], while linagliptin, a dipeptidyl peptidase-4 inhibitor, reduced CRP but not MCP-1 [22]. Further, two studies testing the effect of ARBs and liraglutide on kidney inflammation, remained unpublished (Table 2). Curcumin and long-chain –3 polyunsaturated fatty acids are examples of new interventions, as an alternative to RAAS blockade. Unfortunately, in two studies these supplements were not able to reduce proteinuria and to affect GFR [15, 20]. Conversely, vitamin D and its analogs, which activate the vitamin D receptor, were able to reduce proteinuria, inflammation, and glomerulosclerosis in animal models of DKD [48]. One small trial investigating the effect of vitamin D3 in addition to standard therapy in 22 DM patients with early kidney disease, obtained a 17% reduction of mean UACR (ClinicalTrials.gov; NCT00552409). Pentoxifylline (phosphodiesterase inhibitor), ruboxistaurine (protein kinase C inhibitor), and N-acetylcysteine (antioxidant) are promising molecules that showed renoprotective effect in a mouse model and in small trials on humans [48]. The data and outcomes from patients treated with these experimental drugs are as yet not available for assessment due to a delay or failure to publish (Table 1).

This analysis has some limitations. Since ClinicalTrials.gov is considered the most relevant clinical trial registry, we did not investigate other databases. In addition, the investigation of a clinical trial registry implies that only registered trials were included in our analysis. In order to prevent classifying a trial as unpublished, we conducted an exhaustive literature search in two major databases (i.e., PubMed and Google Scholar) with multiple search terms and contacted investigators or sponsors. This analysis assumes that the entries provided in the ClinicalTrials.gov registry are accurate and complete as mandated by the FDAAA. Our data define the current publication bias in phase III clinical trials investigating DN. We hope that the publication efforts will increase over time.

Conclusion

The need for a better publication discipline of clinical trials is obvious based on our study which found that data on 11 different interventions and more than 1000 patients remained undisclosed. Transparency in clinical research has the potential to improve patient care and prevent patients from being exposed to redundant research. The outcome of the phase III clinical trials included in our study was quite limited, and the need for new approaches to prevent or slow the progression of DN is obvious. Several mechanisms underlying DN pathophysiology have been elucidated, which opens new frontiers for the development of specific DKD therapies. Experimental therapies targeting inflammatory, oxidant, or pro-fibrotic pathways activated during DKD are currently under investigation in phase II and III clinical trials [48].