Neuropeptide Y and chronic kidney disease progression: a cohort study

ABSTRACT

INTRODUCTION

Neuropeptide Y (NPY) is a sympathetic neurotransmitter with wide-ranging effects in various organ systems, from the central nervous system to the cardiovascular (CV) system, the bone and the kidney [1]. The major systems expressing NPY include sympathetic neurons, enteric neurons and various brain pathways [2]. Like the main sympathetic transmitter, norepinephrine, NPY modulates renal transport mechanisms of sodium [3] and potassium [4] and has important effects on inflammation and immune function regulation [5, 6]. Circulating NPY mainly derives from the intestinal circulation [7] and this neuropeptide is emerging as a critical player in the interaction between the gut microbiota and brain acting as neuroendocrine transmitter orchestrating the microbiota gut–brain axis in health and disease [8]. NPY has been implicated in neurodegenerative diseases [9], gastrointestinal diseases [10], obesity [11] and CV diseases [12, 13].

Sympathetic overactivity was suspected to play a role in renal disease progression almost two decades ago [14, 15]. A sympathetic activity marker like heart rate predicts progression to kidney failure independently of other risk factors in elderly chronic kidney disease (CKD) patients [16]. Besides being an integral part of the sympathetic system, NPY has per se an established role in innate immunity and inflammation [5, 6], and inflammatory phenomena most likely play a relevant role in the progression of CKD towards kidney failure [17, 18]. The possibility that this neuropeptide is involved in CKD progression is suggested by sparse observations in various experimental models such as kidney inflammation in mice with systemic lupus [19] and kidney fibrosis and renal dysfunction in early postnatal hyperalimentation in the rat [20]. Furthermore, a genetic polymorphism in the precursor molecule to NPY, preproNPY, associates strongly with proteinuria and the risk of nephropathy in type 2 diabetics [21]. Of note, circulating NPY levels have already been associated with left ventricular hypertrophy [22] and incident CV events [23] as well as with bone disease [24] in advanced CKD.

With this background in mind, we tested the association of NPY with the progression renal function, proteinuria and incident renal events in a cohort of CKD patients in Southern Italy (the Multiple Intervention and AUdit in Renal Diseases to Optimize Care, MAURO) and in a parallel CKD cohort (the Multifactorial Approach and Superior Treatment Efficacy in Renal Patients, MASTERPLAN) in the Netherlands.

MATERIALS AND METHODS

The study protocol was in conformity with the ethical guidelines of our institutions and informed consent was obtained from each participant.

The first cohort (test cohort) included 753 patients with Stages 2–5 CKD (age: 62 ± 11 years; 60% males) consecutively recruited from 22 Nephrology Units in Southern Italy (Calabria, Sardinia and Sicily regions). All these patients were included in the MAURO study, a cluster randomized controlled trial in 22 renal clinics that aimed to assess the efficacy of a multimodal quality improvement intervention to increase compliance with guideline recommendations for prevention of CKD progression and CV complications. Patient enrolment was performed between October 2005 and 2008. The selection criteria and the detailed clinical characteristics of this cohort were described in previous papers [25, 26]. In the present study, we included the 753 patients with an available plasma sample out of 759 patients in the whole cohort. All patients were in stable clinical condition and none had inter-current infections or acute inflammatory processes. Inclusion criteria were: non-acute or rapidly evolving renal diseases; age ranging from 18 to 75 years; non-transplanted and non-pregnant, not affected by cancer or diseases in the terminal phase. The second cohort was composed of patients enrolled in the MASTERPLAN study. Briefly, the MASTERPLAN study is a randomized controlled trial that evaluated the added value of nurse practitioner care in reducing CV events and attenuating the decline of kidney function in patients with Stages 3–5 CKD [27, 28]. Patients were randomized to receive nurse practitioner support added to physician care (intervention group) or physician care alone (control group). For the purpose of this study, we considered 576 patients with Stages 3–5 CKD (age: 59 ± 13 years; 68% males) where a plasma sample for the measurement of NPY was available among the 788 patients included in the MASTERPLAN study. A detailed description of the two cohorts is given in Table 1. No patients had congestive heart failure at baseline either in the MAURO or in the MASTERPLAN cohort.

Laboratory measurements

In the MAURO study, the estimated glomerular filtration rate (eGFR) was estimated by the equation including isotope dilution mass spectrometry—creatinine and cystatin [29] and the MDRD175 four variables equation [30]. In the MASTERPLAN cohort, eGFR was calculated on the basis of serum creatinine by using the MDRD175 equation [30]. Serum lipids, glucose, albumin, calcium, phosphate, albumin, high-sensitivity C-reactive protein (CRP), 24-h urinary proteinuria and haemoglobin were measured by standard methods in routine clinical laboratories of participating centres. In both cohorts, NPY (intra-assay coefficient of variation: 4.5–5.0%; inter-assay coefficient of variation: 9.2–7.6%) was measured at a single laboratory (CNR Laboratories) by a commercially available RIA KIT (by Euro Diagnostica AB, Lundavagen, Malmo, Sweden). The antibody in this kit is specific for human NPY (100% cross-reaction) and does not cross-react with PPY (<0.01%), pancreatic polypeptide (<0.1%), PYY 3–36 (<0.1%) and NPY 22–36 (<0.1%).

Study plan

To test the hypothesis that NPY affects CKD progression, we first investigated the cross-sectional and longitudinal (by mixed linear models analysis, see the ‘Statistical analysis’ section) association between NPY levels, eGFR and 24-h urinary protein. To validate findings that emerged in the MAURO cohort, we then repeated the same analyses in the MASTERPLAN cohort. Finally, we performed a joint Cox regression analysis (i.e. an analysis in the combined cohort of MAURO and MASTERPLAN, n = 1329 stratified by study cohort) of the relationship between baseline NPY and the incidence rate of a composite endpoint of renal events (eGFR reduction >30% or dialysis/transplantation) and death.

Statistical analysis

Data are expressed as mean ± standard deviation (normally distributed data), median and interquartile range (non-normally distributed data) or as percentage frequency (binary data), as appropriate.

At baseline, the association between circulating levels of NPY and eGFR and 24-h urinary protein was investigated (both in the test and in the validation cohort) by Pearson product moment correlation coefficient (r) and by multiple linear regression analysis. In multiple linear regression models (with eGFR and 24-h urinary protein as dependent variables), we adjusted for a fixed series of well-established traditional risk factors [age, sex, diabetes, current smoking, systolic blood pressure (BP)] for CKD progression for a series of risk factors peculiar to CKD (albumin, phosphate and haemoglobin) as well as for serum CRP. The linear mixed model (LMM) accounts for all information including patients with only one renal function assessment, and can provide all estimates in a single-stage calculation and can deal with unequal number of measurements per patient [31]. In this analysis, the baseline eGFR is included in the analytical vector (the eGFR series) and therefore no adjustment for baseline eGFR is needed [32]. The relationships between baseline NPY and the evolution of eGFR and 24-h urinary protein over time were investigated in the test and in the validation cohort by crude and adjusted LMMs. Multiple LMMs included the key risk factor (baseline NPY), age, gender, diabetes and current smoking as well as repeated measurements over time of systolic BP, albumin, phosphate and haemoglobin. The cumulative incidence of the combined renal endpoint (30% GFR loss or dialysis start or renal transplantation), accounting for the competing risk of death, according to tertiles of baseline NPY was calculated as according to Coviello and Boggess [33]. Joint univariate and multiple Cox regression analyses in the two cohorts (n = 1329) were used to model time to the combined renal endpoint as a function of NPY levels. To account for the competitive risk of death [34, 35], we adopted the Fine and Gray [36] method. In this model, we included the same set of covariates considered for the cross-sectional multiple linear regression analyses (see Table 2) as well as baseline eGFR and 24-h urinary protein. In a sensitivity analysis, the type of cohort (MAURO versus MASTERPLAN) was also tested as a potential confounder or an effect modifier into the multivariate Fine and Gray model. The proportionality assumption of Cox models was tested by the analysis of Schoenfeld residuals and no violation was found. The homogeneity of the hazard ratio associated with a fixed increase in NPY levels (25 pmol/L) in the two cohorts was investigated by the effect modification analysis, i.e. by introducing into the same multiple Cox model the key risk factor (NPY, 25 pmol/L increase), the potential effect modifier and the interaction term ‘NPY× effect modifier’. In the Cox model accounting for the competing risks of death, data are expressed as sub-hazard ratios (SHR) and their 95% confidence interval (CI) and P-values. All calculations were made using standard statistical packages (SPSS for Windows Version 22, Chicago, IL, USA; STATA 13 for Windows, College Station, TX, USA).

RESULTS

The demographic and clinical characteristics of the MAURO and MASTERPLAN cohorts are summarized in Table 1. The two cohorts were quite similar for NPY levels as well as for age, gender, body mass index (BMI), background CV comorbidities, BP, lipids, calcium, phosphate, haemoglobin, albumin, CRP, eGFR and CKD stages (Table 1). The proportions of patients on renin–angiotensin system (RAS) inhibitors, diuretics and other antihypertensive drugs were quite homogenous in the two cohorts (Table 1). Furthermore, in the MAURO cohort, there was a lower proportion of smokers, a higher frequency of diabetics and a higher median level of proteinuria (0.6 g/day) when compared with the MASTERPLAN cohort (0.3 g/day) (Table 1).

NPY levels were normally distributed in both cohorts (Supplementary data, Figure S1). In cross-sectional analyses, circulating levels of NPY were related directly with 24-h proteinuria and inversely with eGFR (Supplementary data, Figure S2) and these relationships were largely independent of traditional risk factors and risk factors peculiar to CKD (Supplementary data, Table S1).

NPY, eGFR and proteinuria progression over time

In the MAURO study, NPY associated with the progression of the eGFR over follow-up, in both unadjusted and adjusted LMMs (Table 2). In the adjusted analysis in this cohort, a 25 pmol/L increase in NPY associated with a 3.60 mL/min/1.73 m² reduction in eGFR over a 3-year follow-up (95% CI: −4.46 to −2.74 mL/min/1.73 m²). In the MASTERPLAN cohort (Table 2), the adjusted rate of GFR loss overtime attributable to a 25 pmol/L increase in NPY was lower than in MAURO [−0.83 mL/min/1.73 m² (95% CI: −1.41 to −0.25 mL/min/1.73 m²) over a 4.75-year follow-up] but still highly significant (P = 0.005). Forcing 24-h urinary protein into the LMMs reported in Table 2, the NPY–GFR link did not substantially change both in the MAURO [effect estimation (25 pmol/L increase in NPY): −3.54 mL/min/1.73 m², 95% CI: −4.40 to −2.69 mL/min/1.73 m², P < 0.001] and in the MASTERPLAN cohort [effect estimation (25 pmol/L increase in NPY): −0.71 mL/min/1.73 m², 95% CI: −1.27 to −0.14, P = 0.01].

Similarly (Table 2), crude and adjusted NPY was directly and significantly related to the increase in 24-h proteinuria over time in the MAURO and in the MASTERPLAN cohort as well.

NPY and renal endpoints, combined analysis in the two cohorts

In the MAURO study, during the follow-up (median: 36 months; interquartile range: 21–36 months), 42 patients died and 244 had renal events (>30% decrease in the GFR, dialysis or kidney transplantation). In the MASTERPLAN (median follow-up: 57 months; interquartile range: 28–73 months), 109 patients died and 309 had renal events. In a joint analysis of the two cohorts accounting for the competitive risk of death (see the ‘Materials and methods’ section), NPY predicted the combined renal endpoint (30% GFR loss or starting dialysis or transplantation) in both unadjusted and adjusted analyses including baseline eGFR and 24 h urinary protein and other potential confounders (Table 3). In the adjusted Fine and Gray model (Table 3), the global NPY SHR (for a 25 pmol/L increase) for the combined endpoint was 1.09 (95% CI: 1.03–1.16; P = 0.005), being of similar magnitude in the two cohorts [MAURO: SHR 1.14, 95% CI: 1.04–1.25 versus MASTERPLAN: SHR 1.06, 95% CI: 0.98–1.15 (P = 0.25)]. In close similarity to the LMM analysis of the eGFR, adjustment for 24-h urinary protein did not affect the relationship between NPY and the combined endpoint in the Cox regression analysis. Indeed, the risk by NPY (SHR: 1.10, 95% CI: 1.03–1.16) in a model not including 24-h urinary protein was virtually identical to that of the model including this risk factor (SHR: 1.09, 95% CI: 1.03–1.16; P = 0.005). Of note, the study cohorts per se (MAURO versus MASTERPLAN) did not represent neither a confounding factor (P = 0.65) nor an effect modifier (P = 0.25) for the relationship between NPY and the combined renal endpoint. In a complementary categorical analysis stratifying patients into NPY tertiles there was a dose–response relationship between NPY and the cumulative risk for progression to kidney failure (Figure 1).

FIGURE 1

Cumulative incidence of 30% GFR loss or dialysis/transplantation (accounting for the competing risks of death) across NPY tertiles.

Open in new tabDownload slide

The interaction of NPY with major risk factors for the combined renal endpoint is presented in Figure 2. This analysis shows that the risk associated with increased NPY was unmodified by gender, CV disease and proteinuria. However, age was a strong modifier of the same relationship (P = 0.007), the risk being confined to patients older than 65 years while patients younger than this age had no risk excess by NPY. The CKD stage was an additional effect modifier of the NPY–combined renal endpoint relationship (P = 0.035) in that the risk excess was present in patients with mild-to-moderate CKD while it was just marginal and not significant in Stage 4 CKD and null in Stage 5 CKD patients.

FIGURE 2

Stratified effect modification analyses of the NPY–combined renal endpoint relationship (30% GFR loss or dialysis/transplantation) accounting for the competing risk of death.

Open in new tabDownload slide

DISCUSSION

In this study, the plasma concentration of NPY associated with the progression of the GFR and proteinuria over time as well as with time to a combined renal endpoint (GFR decline >30% starting chronic dialysis or transplantation) in two separate cohorts in Southern and Northern Europe. The associations of NPY were largely independent of established risk factors for CKD progression.

Even though less sensitive than norepinephrine, circulating NPY is considered as a reliable index for quantifying sympathetic neural responses regulating the systemic circulation [7]. High sympathetic activity has effects that may engender and/or amplify renal damage, including proliferative effects mediated by beta-adrenoceptors [37]. In an experimental model of chronic volume overload, i.e. a model characterized by high sympathetic activity and RAS activation, albuminuria and podocyte injury are prevented by sympathetic denervation [38]. Directly measured sympathetic activity is closely related with the GFR and with proteinuria in CKD patients [39] and heart rate, a functional correlate and a surrogate of sympathetic activity, predicts CKD progression in elderly patients [16]. Pilot studies in CKD patients with resistant hypertension show that sympathetic denervation associates with hypertension control and GFR stabilization [40]. Thus, our findings are in keeping with the emerging experimental evidence that high sympathetic activity plays a role in CKD progression. In the present study, NPY associated with GFR loss also in models adjusting for BP, suggesting that this neurotransmitter may affect CKD progression also independently of hypertensive mechanism(s). Congestive heart failure has been associated with high NPY [41, 42]. However, in both cohorts no patient had congestive heart failure and adjustment for other CV comorbidities did not meaningfully alter the hazard ratio for renal outcomes in the present study.

On the other hand, NPY per se plays a wide-ranging role in innate immunity modulation because it impinges on activation of antigen-presenting cells, T helper cell differentiation and lymphocyte proliferation, and on chemotaxis and oxidative burst and nitric oxide synthesis in neutrophils [6]. The crucial role of NPY in inflammation via its Y1 receptor is epitomized by the demonstration that experimental inflammation in the colon is markedly attenuated in the Y1-receptor knockout mice [43] and a tumour necrosis factor–NPY cross-talk modulates inflammation, barrier functions and colonic motility during inflammation in experimental colitis [44]. As alluded to before, circulating NPY mainly derives from the intestinal circulation [7] and it is considered as a relevant player in neural circuits whereby the gut–brain axis modulates inflammation. Indeed, NPY and other neuroendocrine factors generated in the gut, vagal and spinal afferent neurons, inflammatory cytokines and signalling molecules generated by the intestinal microbiota connect the gut to the brain [8]. The potential relevance of this pathway for human health is suggested by the observation that chronic constipation—a possible consequence of persistent inflammation triggered by altered gut microbiota secondary to laxative use [45]—is suspected to be implicated in CV disease [46] and in CKD [47] as well.

Even though in both cohorts of this study, NPY levels went along with the progression of proteinuria over time, the risk for faster GFR loss and renal events was virtually unaffected by statistical adjustment for proteinuria. This finding suggests that NPY may lead to kidney failure also by non-proteinuric pathways. Along with this possibility, in African Americans, genetic variants of a biomarker of sympathetic activity like the chromograning gene associate with a substantially non-proteinuric disease like nephrosclerosis [48]. Of note, age was a strong modifier of the NPY–combined renal endpoint relationship, which is in line with the observation that a biomarker of sympathetic activity like heart rate predicts an excess risk for progression to kidney failure in old but not in young CKD patients [16]. Furthermore, the risk for the combined renal endpoint by NPY was modified by the CKD stage being robust in patients with mild and moderate CKD but absent in Stages 4 and 5 CKD patients. This finding generates the hypothesis that interventions aimed at modifying NPY may produce beneficial effects on the kidney only if applied at a mild-to-moderate stage of disease.

Our study is notable for being based on two well-characterized European CKD cohorts progressing at differing rates towards kidney failure, the higher progression rate in MAURO being likely explained by the higher prevalence of diabetes and proteinuria. However, the several limitations of our hypothesis generating observations should be frankly exposed. Even though our data suggest a possible a etiological role of high NPY in CKD progression, our study cannot prove causality. In general, chronically elevated NPY levels exert noxious effects in experimental models including atherogenesis [49] and left ventricular hypertrophy [50] and associate with left ventricular concentric remodelling [22] and incident CV events [23] in advanced CKD. However, high NPY may have also favourable, protective effects for the CV system such as the Y2 receptor-mediated neo-angiogenesis in experimental ischaemia [51] and the anti-fibrotic effect in the myocardium via its molecular fragment 3-36 [52]. By the same token, the same neuropeptide may help to preserve renal function after acute exposure to cisplatin in rats [53]. Detailed studies in experimental models are still needed to understand if NPY is causally involved in CKD or if it is a mere disease marker.

In conclusion, in this study in two European CKD cohorts, NPY associated with proteinuria and faster CKD progression as well as with a higher risk of kidney failure. These hypothesis-generating findings suggest that the sympathetic system and/or properties intrinsic to the NPY molecule, including interference with innate immunity and perhaps mechanism(s) related with the gut microbiota, may play a relevant role in CKD progression. Antagonists to NPY are being developed and the identification of small molecule compounds with high degrees of NPY receptor subtype selectivity [54] opens interesting possibilities for furthering the exploration of the role in NPY in various human diseases and in CKD.

AUTHORS’ CONTRIBUTIONS

Investigators of the MAURO and the MASTERPLAN cohorts contributed equally to this study. C.Z. and P.B. designed the study. G.D., G.T., J.vd.B. and M.B. performed the statistical analysis and gave a relevant contribution to data interpretation. P.P. performed centralized NPY measurements at the CNR Reggio Cal Unit. D.L. and M.P. were responsible for clinical data collection in MAURO as were J.vd.B. and A.v.Z. for MASTERPLAN. J.W. and F.M. are principal investigators in MASTERPLAN and MAURO, respectively, and provided relevant intellectual contribution to data analysis and interpretation. C.Z. produced the first draft of the manuscript, which was then thoroughly revised along the suggestions by the other authors. All authors approved the submitted version of this study.

CONFLICT OF INTEREST STATEMENT

F.M. reports personal fees from Bayer during the conduct of the study. M.B. reports grants from Astra-Zeneca, grants from Dutch Heart Foundation, grants from Netherlands Organization for Health Research, grants from UMC Utrecht, during the conduct of the study. J.vd.B. reports grants from Dutch Kidney Foundation during the conduct of the study. Other authors declared no conflicts of interest.

REFERENCES