Circulating Proneurotensin Concentrations and Cardiovascular Disease Events in the CommunityHighlights
The Framingham Heart Study
Objective—Neurotensin is a peptide whose receptor (sortilin receptor 1) is linked to cardiovascular disease (CVD) development. We hypothesized concentrations of proneurotensin (stable profragment of neurotensin) would predict incident cardiovascular events in community-based subjects.
Approach and Results—Blood samples from 3439 participants in the Framingham Heart Study (FHS) Offspring cohort (mean age 59.2 years, 47.1% male) were tested for proneurotensin. Primary outcome of interest was incident hard CVD (myocardial infarction, stroke, and cardiovascular death); interaction between proneurotensin concentration with sex, low-density lipoprotein concentrations, and sortilin receptor 1 single-nucleotide polymorphisms was sought. At baseline, those in the highest log-proneurotensin quartile were younger and heavier (P<0.001); across proneurotensin quartiles, more prevalent hard CVD (from 3% to 7%; P<0.001) and diabetes mellitus (from 6% to 14%; P<0.001) were present. In age- and sex-adjusted models, log-proneurotensin concentrations predicted incident hard CVD (hazard ratio [HR], 1.24 per SD change in log-proneurotensin; 95% confidence intervals [CIs], 1.11–1.39; P<0.001), a finding that remained on adjustment for standard CVD risk factors (HR, 1.13; 95% CI, 1.01–1.27; P=0.03). Elevated log-proneurotensin concentrations were associated with shorter time to first event (P=0.02). We found no effect modification by sex, low-density lipoprotein concentration, or sortilin receptor 1 single-nucleotide polymorphisms. Concentrations of proneurotensin were modestly associated with left ventricular mass and coronary artery calcium in these subjects.
Conclusions—Higher concentrations of proneurotensin are associated with a greater risk of incident cardiovascular events in the community. This association did not vary according to sex, baseline low-density lipoprotein, or sortilin receptor 1 genotype.
Efforts at reducing cardiovascular events rely on the accurate identification of individuals at risk. Unfortunately, patients who experience cardiovascular events often have a paucity of traditional factors predictive of cardiovascular disease (CVD) to facilitate recognition of risk. In this regard, measurement of circulating biomarkers has been examined as an option for assessing risk for cardiovascular events beyond standard risk factors. Although potentially useful to predict and reclassify risk in some cohorts, studies of such testing for predicting cardiovascular events in low to intermediate risk populations have returned mixed results,1,2 suggesting further efforts are needed to better understand the role of testing of circulating substances for risk prediction in such patients. Beyond potential clinical application, biomarker measurement has also been leveraged as a tool to understand mechanism of CVD onset. More studies of novel and established circulating markers of disease are thus needed, both to advance understanding of optimal means for risk stratification and supplement knowledge about mechanism of disease.
Neurotensin is a 13-amino acid peptide originally isolated from bovine hypothalamic3 and later from intestinal tissue.4 Neurotensin has a wide range of biological roles in the body,5 notably including a broad range of effects on the cardiovascular system; these include regulation of heart rate, myocardial contractility, and vascular tone.6 Effects of neurotensin are transduced primarily through 3 receptors: the G-protein–coupled NTS1 and NTS2 receptors and the non–G-protein–coupled NTS3, otherwise known as sortilin receptor 1 (SORT1), a member of the Vps10p-domain receptor family. SORT1 (also known as sortilin) is involved in the binding of several unrelated ligands, and it plays an important role in hepatic secretion of very low–density lipoprotein cholesterol and regulation of circulating LDL cholesterol concentrations. In addition, genetic variation in the 1p13 locus containing the SORT1 gene is also linked to coronary artery disease development.7 Although SORT1 seems linked to lipid metabolism and CVD risk, it remains unclear if neurotensin plays a role in this association.
Measurement of neurotensin in blood is challenging because of its instability and rapid clearance from the circulation.5 To overcome this issue, immunoassays have been developed for the detection of the propeptide fragment of the peptide, which is released in equimolar amounts to mature neurotensin. Recent data from the Malmö Diet and Cancer Study suggested that concentrations of proneurotensin were independently predictive of diabetes mellitus and CVD, particularly in women.8 However, beyond these preliminary findings, no other data exist on association of circulating proneurotensin concentrations with the incidence of CVD events, nor are mechanistic analyses available. Accordingly, we sought to examine links between proneurotensin and CVD in a cohort of patients from the Framingham Heart Study (FHS) Offspring study. Our hypothesis was that PNT concentrations would independently and positively predict CVD events, and it would do so in a manner mediated via either blood LDL cholesterol values or genetic variation in the SORT1 receptor.
Materials and Methods
Materials and methods are available in the online-only Data Supplement.
Characteristics of the study sample as a function of proneurotensin quartiles are shown in Table 1. The mean age of the study sample was 59.19±10 years, and 53% of participants were women. Compared with subjects in log-proneurotensin quartiles 1 through 3, those in the highest quartile were more likely to be younger (P=0.006), heavier (P<0.001), and more likely to smoke (P<0.001). There was no difference in LDL cholesterol concentrations across log-proneurotensin quartiles; similarly, across quartiles of LDL cholesterol, there was no difference in log-proneurotensin concentrations (P=0.71).
In multivariable linear regression analyses, variables independently correlated with log-proneurotensin concentrations included waist girth (β=0.0044; P=0.005), smoking (β=0.0818; P<0.001), and prevalent diabetes mellitus (β=0.1438; P<0.001); concentrations of LDL cholesterol were not predictive of log-proneurotensin concentrations (β=−0.0002; P=0.39). Furthermore, neither age (β=−0.0011; P=0.10) nor male sex (β=−0.0007; P=0.96) significantly correlated/predicted concentrations of log-proneurotensin.
As detailed in Table 2, at baseline, across proneurotensin quartiles study participants with higher concentrations were more likely to have prevalent diabetes mellitus (from 6% to 14%; P<0.001), prevalent hard CVD (from 3% to 7%; P<0.001), or prevalent hard coronary heart disease (CHD; from 3% to 5%; P=0.06). No association between log-proneurotensin and prevalent cancer (including breast cancer) was observed.
Biomarker Concentrations and Outcomes
During a mean follow-up of 14.0 years, 342 (10.5%) individuals had a hard CVD event, with 166 myocardial infarctions, 148 strokes, 27 CHD deaths, and 1 stroke death. During similar follow-up time, 209 (6.3%) individuals had a hard CHD event.
Table 3 details predictive value of log-proneurotensin for hard CVD events. In age- and sex-adjusted Cox proportional hazards models (Table 3), log-proneurotensin concentrations were positively associated with incident hard CVD (HR, 1.242 per 1 SD change in log-proneurotensin; 95% CI, 1.11–1.39; P<0.001). In models adjusted for standard risk factors (including body mass index), log-proneurotensin remained significantly associated with incident hard CVD (HR, 1.13 per 1 SD change in log-proneurotensin; 95% CI, 1.01–1.27; P=0.03). The HR for log-proneurotensin remained significant in models forcing concentrations of LDL (HR, 1.121 per 1 SD change in log-proneurotensin; 95% CI, 1.002–1.254; P=0.05), the interaction factor of male×LDL (HR, 1.121 per 1 SD change in log-proneurotensin; 95% CI, 1.002–1.254; P=0.05), or an LDL cholesterol above the median for the group (HR, 1.122 per 1 SD change in log-proneurotensin; 95% CI, 1.003–1.215; P=0.05).
Examining risk across log-proneurotensin quartiles in age- and sex-adjusted models (Table I in the online-only Data Supplement), greatest risk of incident hard CVD was observed in log-proneurotensin quartile 4 (HR, 1.53 versus quartile 1; P=0.005). Using stepwise selection for prediction of hard CVD, comparable results were found, with higher concentrations of log-proneurotensin predicting risk (Table 3).
Considering hard CHD, similar results were found, with higher concentrations of log-proneurotensin predicting hard CHD in adjusted models that also contained prevalent CVD (Tables I and II in the online-only Data Supplement). In fully adjusted models for hard CHD, we found an HR of 1.156 per 1 SD change in log-proneurotensin for predicting hard CHD (95% CI, 1.005–1.330; P=0.04).
Addition of further biomarker results for highly sensitive troponin I, growth differentiation factor-15, or soluble ST2 (previously reported to be predictive of cardiovascular events in this cohort9) in a stepwise model resulted in retention of growth differentiation factor-15 as a predictor of hard CVD (HR, 1.25 per 1 SD change in log-transform; 95% CI, 1.11–1.40; P<0.001), whereas log-proneurotensin became marginally nonsignificant (HR, 1.11 per 1 SD change in log-proneurotensin; 95% CI, 0.99–1.24; P=0.07).
In interaction testing with fully adjusted models, we did not observe a proneurotensin×LDL interaction (P=0.97) for prediction of hard CVD events. Similarly, we found no proneurotensin×SORT1 single-nucleotide polymorphism (SNP) interactions for prognostication of hard CVD (proneurotensin×rs629301, P=0.76; proneurotensin×rs646776, P=0.56; and proneurotensin×rs12740374, P=0.65). Similarly, negative results were found in proneurotensin×LDL or proneurotensin×SNP interaction analyses for hard CHD.
In contrast to previous data,8 we found no interaction term with respect to sex- and proneurotensin-based prognostication (Table 4). Although the HR for log-proneurotensin predicting hard CVD was numerically higher in women (HR, 1.175 [95% CI, 0.993–1.390]) compared with men (HR, 1.118 [95% CI, 0.962–1.300]), these differences did not approach statistical significance.
Notably, log-transformed concentrations of proneurotensin were not related to incident change in body mass index or waist girth. Log-proneurotensin did not predict incident diabetes mellitus; during follow up, there were only 32 incident cases. Over a mean follow-up of 8.7 years, the HR for log-proneurotensin to predict incident diabetes mellitus was 0.942 (95% CI, 0.662–1.340; P=0.74).
Finally, in Kaplan–Meier analyses, shorter time to first event was seen in higher log-proneurotensin values (Figure; log rank P=0.02). Similar results were found for hard CHD.
Cardiac Structure and Function
In age-, sex-, and height-adjusted regression, proneurotensin concentrations were associated with LVM (parameter estimate 0.016 per SD of log-transformed proneurotensin; P=0.0002) and the presence of LVSD (P=0.05). Across proneurotensin quartiles (Q), significantly higher mean LVM (P=0.001) was observed; LVSD was least likely in proneurotensin Q1 versus Q4 (P=0.05). In multivariable-adjusted models, these findings were attenuated (P=0.10 for LVM; P=0.22 for LVSD). Proneurotensin concentrations were associated with extent of CAC in age- and sex-adjusted regression analyses (parameter estimate 0.145 per SD of log-transformed proneurotensin; P=0.02); however, in multivariable-adjusted analyses, this finding was no longer significant (P=0.10). Findings were similar when excluding subjects with prevalent CVD.
The principal findings of our analysis were that concentrations of PNT were associated cross-sectionally with a more deleterious cardiometabolic phenotype, and they were prospectively associated with incident cardiovascular events in the population, independently of LDL concentrations or SORT1 genotypes relevant to the development of atherosclerosis. The association of PNT with hard CVD and hard CHD remained robust after adjustment for traditional cardiovascular risk factors, and it provided modest risk reclassification for predicting events. Concentrations of proneurotensin were modestly associated with cardiac structure and function in echocardiographic and CAC imaging, but in rigorously adjusted models, these associations were less obvious. These results support a possible role for the neurotensin system in the development of clinical CVD, but further evaluation is needed.
Neurotensin is thought to work as a local hormone on peripheral organs, including the heart; it has a broad range of cardiovascular effects, regulating heart rate, myocardial contractility, and blood pressure.6 Although measurement of neurotensin had been challenging because of analytic instability, recent development of a propeptide assay to quantify neurotensin has overcome this challenge. We, therefore, measured proneurotensin in the FHS Offspring Study in an effort to examine the prognostic role of the neurotensin system for CVD incidence.
Our results are notable because little information exists presently about neurotensin and risk for heart disease. Data from Melander et al8 from the Malmö Diet and Cancer Study suggest that concentrations of proneurotensin predicted adverse cardiometabolic outcome, including CVD and incident diabetes mellitus. An intriguing interaction between proneurotensin-based prognostication and sex was found by Melander et al,8 suggesting proneurotensin provided unique prognostic information particularly in women, correctly reclassifying as much as 40% of women for risk of cardiovascular death. In addition, in Malmö, proneurotensin predicted incident breast cancer. Our results are reasonably comparable with those from Malmö with respect to the ability of proneurotensin to predict hard CVD and hard CHD, although such ability to predict is much more modest in our more rigorously adjusted analyses. In addition, we could not confirm a sex-specific value of proneurotensin. Furthermore, although proneurotensin was associated with more deleterious cardiometabolic state at enrollment (with higher values of proneurotensin associated with more prevalent obesity and diabetes mellitus), we could not confirm proneurotensin predicted incident obesity or diabetes mellitus. Finally, no predictive ability to prognosticate incident cancers was seen in our analysis. The differences in results may be explained on the basis of the fact that our study sample was smaller, and participants differ considerably in terms of baseline cardiometabolic risk compared with those in the Malmö analysis. In addition, our statistical models were more rigorously adjusted. Despite more modest results, our findings are important because they clarify those from Malmö.
We initially hypothesized that the deleterious effect(s) of neurotensin might be explainable on the basis of binding to SORT1. This receptor is intracellular and non–G-protein–coupled and plays a role in endocytosis and trafficking of several molecules (including various cholesterol particles). SORT1 has been implicated in LDL cholesterol metabolism and in very low–density lipoprotein and proprotein convertase subtililisin/kexin type 9 secretion.10 Genetic variation of SORT1 is pivotally linked to coronary artery disease development in humans, in part, through its effects on lipoprotein metabolism.7 In theory, therefore, higher values of proneurotensin could influence cardiac risk through interactions at the level of the SORT1 receptor via interference with normal lipid processing. In our analysis, however, we found no association between proneurotensin concentrations and LDL cholesterol values. In addition, in evaluating various SORT1 SNPs important to CVD development, we could not detect any proneurotensin×SORT1 interaction for prognosis; this lack of significant interaction is not because of solely lack of power as the outcome HR per 1 SD increase of log-PNT for participants above and below median rs629301 were 1.23 and 1.21, respectively, indicating consistent log-PNT effect across the values of this SNP. As another example, a similar trend was seen for rs1274074 (HR was 1.20 for participants above the median and 1.24 for participants below it). Indeed, with the current sample size/event rates, we have ≈90% power to detect an interaction with SNP if HR for participants below median SNP was approximately twice that for participants above the median SNP or vice-versa. Finally, although we did not perform interaction testing between proneurotensin and other variables influenced by the SORT1 receptor (such as proprotein convertase subtililisin/kexin type 9), our results suggest that the deleterious link(s) between proneurotensin and CVD may be mediated via effects of neurotensin on receptors other than SORT1.
Although neurotensin binds both NTS1 and NTS2, substantial differences exist between the 2 receptors. Although both are G-protein coupled, NTS1 has considerably higher affinity for neurotensin. Furthermore, although both receptors are found in cardiovascular tissue, NTS1 is currently suspected to be more involved in regulation of cardiovascular effects of neurotensin. Manipulation of NTS1 receptor function with SR48692 (a selective nonpeptide NTS1 inhibitor) resulted in dose-dependent effects on blood pressure, heart rate, myocardial contractility, vascular tone, permeability, and endothelial cell survival.11–14 Hypothetically, therefore, higher concentrations of neurotensin may result in cardiac stimulation, increased vascular tone, and accelerated atherogenesis as a consequence of binding to the NTS1 receptor. In contrast, NTS2—a low affinity receptor for neurotensin—is not currently thought to play a role in cardiovascular responses. Of importance, it remains unclear if circulating proneurotensin concentrations reflect tissue-based concentrations of neurotensin. More studies are needed to better understand the role of both NTS1 and tissue-based neurotensin as participants in the development of CVD.
Limitations of our analysis include the fact that our sample size is relatively small and our event rates are modest. Nonetheless, our results support the primary hypothesis that proneurotensin is predictive of cardiovascular events in the community, although in a manner somewhat more modest than that demonstrated by Melander et al.8 Our results provide balance to the literature. Although the more modest prognostic implication of proneurotensin in our study may be because of limited event rates, for both hard CVD and hard CHD outcomes, we had >85% power to detect a difference between each of the upper and the reference quartiles of log-proneurotensin. The highly skewed nature of proneurotensin makes association between graded variables (such as age or body mass index) somewhat challenging. We have log-transformed proneurotensin to best address this fact. Although binding of neurotensin to NTS1 remains a potential mechanistic explanation for the increased cardiovascular risk associated with higher values of proneurotensin, we cannot directly inform mechanism of how the neurotensin system predicts CVD or CHD. Finally, unfortunately, we lack SNP data on NTS1 polymorphisms.
In summary, concentrations of proneurotensin predicted onset of hard CVD and hard CHD events in a community-based cohort. The ability of proneurotensin to prognosticate such events seems to be independent of LDL cholesterol values or SORT1 genotypes, suggesting deleterious cardiovascular effects of neurotensin are mediated via another mechanism, such as binding to the NTS1 receptor. Further data are needed on the role(s) played by the neurotensin system in cardiovascular risk.
We thank SphingoTec for proneurotensin analysis.
Sources of Funding
Dr Januzzi is supported, in part, by the Hutter Family Professorship in the Field of Cardiology, as well as the DeSanctis Endowed Clinical Scholar in Medicine Fund. Dr Gaggin is supported. in part, by the Clark Fund for Cardiac Research Innovation. Dr Vasan is supported, in part, by National Heart Lung and Blood Institute contracts N01-HC-25195 and HHSN268201500001I.
Dr Januzzi has received grant support from Siemens, Singulex, and Prevencio; consulting income from Roche Diagnostics, Critical Diagnostics, Sphingotec, Phillips, and Novartis; and participates in clinical end point committees for Novartis, Amgen, Janssen, and Boehringer Ingelheim. Dr Gaggin has received consulting income from Roche Diagnostics, American Regent, EchoSense, Boston Heart Diagnostics, and Critical Diagnostics. Dr Maisel has received grant support from Abbott, Roche Diagnostics, Alere, and he has received consulting income from Critical Diagnostics and Sphingotec. Dr Wang reports research support from Diasorin and consulting income from Takeda. The other authors report no conflicts.
The online-only Data Supplement is available with this article at http://atvb.ahajournals.org/lookup/suppl/doi:10.1161/ATVBAHA.116.307847/-/DC1.
- Nonstandard Abbreviations and Acronyms
- coronary heart disease
- cardiovascular disease
- Framingham Heart Study
- low-density lipoprotein
- single-nucleotide polymorphism
- Sortilin receptor 1
- Received January 27, 2016.
- Accepted May 24, 2016.
- © 2016 American Heart Association, Inc.
- Wang TJ,
- Gona P,
- Larson MG,
- Tofler GH,
- Levy D,
- Newton-Cheh C,
- Jacques PF,
- Rifai N,
- Selhub J,
- Robins SJ,
- Benjamin EJ,
- D’Agostino RB,
- Vasan RS
- Carraway R,
- Leeman SE
- Kitabgi P,
- Carraway R,
- Leeman SE
- Melander O,
- Maisel AS,
- Almgren P,
- Manjer J,
- Belting M,
- Hedblad B,
- Engström G,
- Kilger U,
- Nilsson P,
- Bergmann A,
- Orho-Melander M
- Wang TJ,
- Wollert KC,
- Larson MG,
- Coglianese E,
- McCabe EL,
- Cheng S,
- Ho JE,
- Fradley MG,
- Ghorbani A,
- Xanthakis V,
- Kempf T,
- Benjamin EJ,
- Levy D,
- Vasan RS,
- Januzzi JL
- Donelan J,
- Boucher W,
- Papadopoulou N,
- Lytinas M,
- Papaliodis D,
- Dobner P,
- Theoharides TC
- Gully D,
- Canton M,
- Boigegrain R,
- Jeanjean F,
- Molimard JC,
- Poncelet M,
- Gueudet C,
- Heaulme M,
- Leyris R,
- Brouard A
- Schaeffer P,
- Laplace MC,
- Savi P,
- Pflieger AM,
- Gully D,
- Herbert JM
Neurotensin is a neuropeptide with a broad range of effects in the body, including feeding behavior. The heart neurotensin regulates heart rate, myocardial contractility, and blood pressure.
There are 3 receptors for neurotensin, including the sortilin receptor that is also involved in lipid trafficking.
Measurement of neurotensin is challenging because of instability. We measured concentrations of proneurotensin, which is a stable profragment equivalent of neurotensin.
Higher concentrations of proneurotensin were found to be cross-sectionally associated with a greater risk of incident cardiovascular events in the community. This association did not vary according to sex, baseline cholesterol values, or sortilin genotype.
Concentrations of proneurotensin were modestly associated with cardiac structure and function and coronary calcium.
More data are needed to better understand the link(s) between neurotensin and cardiovascular disease in the community.