Summary of main results

In this systematic review we could include 30 randomised studies (2039 adults with ADPKD) evaluating 11 interventions (ACEi, ARBs, calcium channel blockers, beta blockers, V2R antagonists, mTOR inhibitors, somatostatin analogues, antiplatelet agents, eicosapentaenoic acids, statins and vitamin D). For most interventions, data were available only from single studies and provided information for surrogate outcomes such as GFR, blood pressure and kidney and cyst volumes, leading to low confidence in estimated treatment effects. Overall, there was little or no evidence that currently available interventions improve patient‐related or kidney health outcomes while evidence for adverse events was sparse and showed potential for increased harm.

ACEi significantly reduced diastolic BP but had uncertain effects on mortality, ESKD, kidney volumes, GFR, creatinine levels and albuminuria. ACEi did not produce different effects on kidney function when compared with beta blockers or ARBs.

In meta‐analyses of data pooled from two studies, V2R antagonists increased thirst and dry mouth. In addition, data from a single RCT (TEMPO 3‐4 Study 2011) showed a greater proportion of patients treated with these drugs had elevations of liver‐enzyme levels. V2R antagonists showed apparent benefits on kidney function and volumes but confident interferences about their impact on the progression to ESKD could not be drawn as such benefits were only shown by a single study (TEMPO 3‐4 Study 2011). Furthermore, in this study data were analysed on an intention‐to‐treat basis. A higher percentage of patients in the intervention than in the placebo arm (22.9% versus 13.8%) discontinued the study, mostly due to low compliance to the treatment. This may introduce a significant attrition bias limiting the confidence and the overall applicability of findings.

mTOR inhibitors had uncertain effects on GFR, total kidney volume, BP and other secondary outcomes (albuminuria, proteinuria) but caused oral ulceration, infection, and diarrhoea. Of note the use of these drugs was associated with a remarkable increase in the risk of angioedema (RR 13.39), although the clinical reliability of this point estimate might be questioned due to the very wide confidence interval observed (2.56 to 70.00). Few data were available on mortality and RRT outcomes and treatment effects were accordingly absent.

When compared with placebo, somatostatin analogues reduced SCr and total kidney volume, but had uncertain benefits on GFR and other secondary outcomes, while causing diarrhoea. As shown in a three‐year duration study (ALADIN Study 2013), the benefits of these drugs on kidney outcomes (particularly, kidney volumes) seemed to be more evident in the early treatment phase (one year) while they tended to dilute at later stages (three years).

Little or no evidence was found to exist for the impact of calcium channel blockers, eicosapentaenoic acids, statins, vitamin D compounds and antiplatelet agents on disease progression and patient outcomes. These treatments were associated with undefined benefits in terms of kidney function and other secondary endpoints, such as BP or proteinuria.

Overall completeness and applicability of evidence

The evidence on available interventions for slowing progression of ADPKD was sparse in both adults and children. Information on ADPKD progression, both in terms of cyst/kidney volumes and deteriorating kidney function was limited, restricted to few interventions and mostly inconclusive because of the availability of single studies only. Most key mortality and cardiovascular outcomes were only marginally addressed. There were few or no data on major patient‐centred outcomes, such as CKD progression, mortality, major morbid events, quality of life and disease‐related symptoms. Conversely, the vast majority of studies only demonstrated sparse benefits in surrogate endpoints (e.g. change in total kidney or cyst volumes, blood pressure control, proteinuria, albuminuria) without evidence of clear benefits on outcomes of CKD progression (e.g. RRT). Surrogate outcomes are indeed useful as proxies for patient‐centred outcomes, particularly in slow‐progressing diseases such as ADPKD. However, the main disadvantage of using surrogate outcomes is that favourable effects of interventions do not always translate into clear benefits to harder endpoints. In some cases, surrogate outcomes may even be "hypothesis‐generating" at best. Although GFR remains the preferable outcome measure for treatment effectiveness, in the majority of ADPKD patients this parameter remains relatively steady until late in the disease. Total kidney volume has been extensively adopted as surrogate outcome measure in ADPKD studies. However, whether a reduced rate of kidney enlargement effectively translates into slowed kidney function deterioration is still object of debate. Accordingly, recently the FDA did not consider the observed improvement in kidney volumes in the TEMPO studies as enough to justify lifelong therapy with the V2R‐antagonist Tolvaptan.

The extreme heterogeneity in study length (ranging from five days to 60 months) also deserves mentioning. Many studies were indeed designed to assess treatment effects in very short time as per their exploratory nature (e.g. pilot or small cross‐over studies: Biao 1997; TEMPO 248 & 249 2005; van Dijk 2003). Short‐time studies preclude interpretations on hard outcomes (death, dialysis), particularly in slowly progressing chronic diseases. In addition, short‐term studies can be powered to investigate the effect of treatments on surrogate endpoints only, showing no proof of significant clinical changes in the long‐term. Performing cumulative outcome analyses with such study heterogeneity in follow‐up duration is potentially unreliable.

All studies were pilot or cross‐over studies conducted on very small populations, with the exception of two multicentre studies (TEMPO 3‐4 Study 2011; Walz 2010). Results from small studies are inconclusive in nature and probably more useful to set the stage for larger confirmation studies rather than for providing definite indications for clinical practice.

The applicability of findings is also limited by the large number of drop‐out found in most studies. Although the overall drop‐out rate varied widely across the studies (1.6% to 33%), this was greater than 10% in nine studies which included all the largest studies conducted on ADPKD patients. Furthermore, in most cases drop‐outs were unbalanced among the study groups, being more frequently observed in the active rather than in the control arm (Cadnapaphornchai 2005; ELATE Study 2011; Melemadathil 2013; Nutahara 2005; RAPYD Study 2012; TEMPO 3‐4 Study 2011; Zeltner 2008). High dropout rates may introduce important attrition bias and limit the internal validity of findings. Per‐protocol analyses can be useful to bypass limitations related with high dropout rates. However, such approaches convey a high risk of bias due to selection of patients and may provide clinically dubious information as they may over‐estimate the benefit or underestimate the harm of an intervention.

Quality of the evidence

Three of the included studies were cross‐over studies (Ruggenenti 2005; SIRENA Study 2010; van Dijk 2001). Most studies focused on small cohorts, were not powered to observe differences in patient‐centred outcomes and did not provide adequate study reporting or information on blinding of patients or investigators or both to assess risks of bias properly.

Limitations in study reporting and design markedly reduced confidence in the results. Actual treatment effects may differ significantly from those calculated from existing studies.

Random sequence generation was adequate in only eight studies (ALADIN Study 2013; Cadnapaphornchai 2005; ELATE Study 2011; Fassett 2010; LOCKCYST Study 2009; RAPYD Study 2012; Ruggenenti 2005; SUISSE ADPKD Study 2007), and in almost half the studies, blinding was not present or not specified.

The overall drop‐out rate was over 10% in nine studies (AIPRI Study 1996; Cadnapaphornchai 2005; ELATE Study 2011; Hogan 2010; Nutahara 2005; SIRENA Study 2010; TEMPO 3‐4 Study 2011; van Dijk 2003; Zeltner 2008) and only six were conducted using intention‐to‐treat analyses (ALADIN Study 2013; Nutahara 2005; RAPYD Study 2012; SUISSE ADPKD Study 2007; TEMPO 3‐4 Study 2011; Walz 2010).

Potential biases in the review process

Despite being the first overall summary of treatment for ADPKD based on a peer‐reviewed protocol, a systematic search of electronic databases including the Cochrane Renal Group’s specialised register of studies, and applying a standardised procedure for data extraction and analysis incorporating assessment of study methodology, the findings of our review should be interpreted with caution. The lack of data in the available studies represents the key limitation. In most cases, the effect of a given intervention was addressed by single studies, which prevented meta‐analyses with sufficient power to draw definitive conclusions on relevant outcomes. Furthermore, data on patient‐level outcomes (such as ESKD, mortality and cardiovascular events and adverse effects) were collectively scarce or absent. Study design was heterogeneous with marked differences among studies with respect to follow up duration, baseline kidney function and methods of assessment of ADPKD severity. Finally, in most cases, ADPKD assessment was made by echo tomography, a technique widely recognised to be inaccurate for identifying small changes in kidney volumes and poorly suited for very expanded kidneys.

Agreements and disagreements with other studies or reviews

After decades of symptomatic treatment for ADPKD, we now have several novel interventions targeting ADPKD biology arising from a wealth of experimental and non‐randomised studies (Chang 2012). Unfortunately, despite great optimism based on preliminary results, to date there has not been sufficient evidence from the available RCTs to demonstrate clear therapeutic benefits that outweigh treatment hazards. In addition, large RCTs are needed before these interventions could be considered as effective treatments to improve outcomes in ADPKD.

The Consortium for Radiologic Imaging for the Study of Polycystic Kidney Disease (Rule 2006) demonstrated that in people with ADPKD, baseline kidney volume values predicted the rate of increase in kidney volumes regardless of age, and accordingly, higher rates of kidney enlargement reflected a faster decline in kidney function. Kidney volumes have therefore been proposed and extensively used in studies as surrogate endpoints of disease progression to overcome the difficulty of following kidney function slope for very long periods of time in RCTs. Despite this, the benefits of some interventions (e.g. mTOR inhibitors) on kidney volumes in ADPKD have not corresponded with substantial changes in kidney function decline. This raises the question as to whether these biomarkers are appropriate as outcomes for assessing treatment effectiveness of novel interventions in ADPKD when used in isolation (Grantham 2011) and whether other surrogates of kidney function or damage would be more appropriate in studies of ADPKD patients (Helai 2012). In this regard, negative results of mTOR inhibitors studies were largely disappointing. These drugs have been demonstrated to be powerful inhibitors of cyst growth in experimental models (Wu 2007) and retrospective observations clearly showed a reduced cystic phenotype in the livers of kidney transplant recipients undergoing immunosuppression with mTOR inhibitors (Qian 2008). Future directions have been hypothesised for exploring whether there is room for mTOR inhibitors in the pathogenetic treatment of ADPKD, including higher doses or longer regimens of treatment, lower doses in combination with other therapeutic approaches (to minimise adverse events) or the use of analogues with better side‐effects profiles or improved kidney penetration (Wüthrich 2009). We suggest that any future study of higher dose mTOR inhibition requires careful systematic measurement of adverse effects and is based on patient‐relevant outcomes.

Although preliminary findings in animal and human studies have suggested that V2R antagonists and somatostatin analogues can be efficacious in slowing cyst growth (Harris 2009), our analyses demonstrated no conclusive effects for V2R and only small effects for somatostatin analogues on kidney function or kidney volumes. On the other hand, concerns might arise concerning the safety profile of these agents because their use is associated with thirst and dry mouth (V2R antagonists) and diarrhoea (somatostatin analogues). Future studies focusing on the effects on kidney function decline rather than kidney volume surrogates are eagerly awaited to confirm and generalise the benefits of these agents in retarding ADPKD progression.

Blood pressure control is currently one of the mainstays of ADPKD management in clinical practice. Hypertensive ADPKD patients have greater and faster annual rates of kidney volumes growth and an increased prevalence of cardiovascular comorbidities and complications with respect to normotensive people (Ecder 2013). Since hypertensive ADPKD patients are at higher risk of kidney disease progression, these people might represent a higher risk population for future studies which would then be powered to capture patient‐centred kidney and mortality outcomes.

In our review, the use of ACEi was associated with significant improvement in BP control. Unfortunately, this benefit was mostly confined to children, and meta‐analyses were underpowered to detect differences in treatment effects on disease progression. Potentially, results from the ongoing HALT‐PKD Study 2008a testing the efficacy of RAAS‐blockade on the progression of cystic disease and decline in kidney function, will clarify whether an intensive (≤ 110/75 mm Hg) versus standard (≤ 130/80 mm Hg) BP control might produce different effects on the disease course in ADPKD patients with both early (GFR > 60 mL/min/1.73 m²) and advanced (GFR 25 to 60 mL/min/1.73 m²) kidney impairment.