Phthalate-associated hypertension in premature infants: a prospective mechanistic cohort study

The only phthalate approved by the U.S. Food and Drug Administration for use in medical supplies is di-(2-ethylhexyl) phthalate (DEHP), a compound commonly added to polyvinyl chloride (PVC) to soften plastic. DEHP has been identified in a variety of hospital devices including a selection of intravenous (IV) fluid bags and some respiratory equipment [1‐5]. Inherent use of such devices in a neonatal intensive care unit (NICU) may expose small premature infants to relatively high levels of DEHP, exceeding typical adult exposures by several orders of magnitude [1, 3, 6].

In 2013, with a report from Trasande et al. and later corroborated by others, phthalates surfaced as a cause of increased blood pressure in children and adults [7‐10]. At present, the mechanism of this increase remains unclear. However, only a few years earlier, Zhao et al. pointed toward a possible mechanism by showing that certain DEHP metabolites cause sodium retention due to a licorice-like action inhibiting the enzyme11β-hydroxysteroid dehydrogenase type 2 (11β-HSD2) [11].

In 2017, our group reported on a large cohort of 97 premature infants with idiopathic low-renin hypertension, some of whom had chronic lung disease [12]. Al Awad et al. recently reported on a homogenous group of premature infants with low-renin hypertension, all of whom had chronic lung disease [13]. Hypertensive patients in these two studies all shared certain similarities including near-universal low or undetectable plasma renin activity (PRA), presentation of high blood pressure around 35–40 weeks postmenstrual age (PMA), and an excellent response to treatment with spironolactone. These similarities suggest the possibility of a unifying mechanism for the observed hypertension involving sodium transport in the distal nephron.

When considered in concert, the three concepts outlined above—large exposure to DEHP, a potential mechanism of sodium retention tied to DEHP, and cohorts of premature infants with low-renin hypertension—provide a basis for the hypothesis that DEHP increases blood pressure and may cause hypertension in premature infants. This current report tests this hypothesis both by comparing DEHP exposure between hypertensive and normotensive premature infants, as well as by examining the linear relationship between DEHP exposure and blood pressure in the overall cohort. Further, we investigate the infants’ absorption of DEHP based on the detection of toxic urine metabolites, including the principal metabolite, mono-(2-ethylhexyl) phthalate (MEHP) [14‐16]. Lastly, we explore a plausible mechanism associated with idiopathic low-renin hypertension in premature infants by evaluating for MEHP-inhibited 11β-HSD2 and consequential activation of mineralocorticoid receptor pathways.

A prospective observational study was conducted at two medical centers with approvals from the Institutional Review Boards and informed consent given for all participants. We recorded DEHP exposures and blood pressures in premature infants from birth for at least 6 months or until resolution of hypertension off medication occurred, whichever came last. Based on our prior study, we would expect no new cases of idiopathic hypertension after 6 months PMA, nor would we anticipate the ongoing need for anti-hypertensive medications beyond 8 months PMA [12]. Inclusion and exclusion criteria (Table 1) were designed to avoid bias related to clinical factors that might affect blood pressure. Study funding limited initial enrollment to 20 subjects, after which an analysis of the data was performed and presented herein. Two subjects were recruited monthly into the study until enrollment was complete. Although there were perhaps as many as 600 total potential premature infants to enroll, only 2–4 met the study criteria at each of our bi-monthly recruiting days. The infants with the lowest gestational age were approached first to minimize the number of outpatient visits required.

The principal statistical comparisons between normotensive and hypertensive groups for DEHP exposures, blood pressure, and diagnostics were planned to be performed at the time of diagnosis of hypertension (and a corresponding PMA for normotensive infants). The diagnosis of hypertension was anticipated to be near 40 weeks PMA based on our prior work [12].

Of the 26 patients approached for participation, 20 enrolled in the study. One patient unexpectedly died in the first month of life for reasons unrelated to blood pressure and another failed to return for study visits prior to the 40-week PMA visit. This left 18 patients available for analysis. No secondary hypertension occurred in this cohort, and no other patients met exclusion criteria.

Figure 1 provides a visual depiction of SBP index, intravenous, and respiratory-related DEHP exposures, and sodium transporter expression throughout the study period. Figure 1a depicts the average SBP index in hypertensive versus normotensive patients in relation to the type of DEHP exposure, whether respiratory or intravenous. On average, SBP for the hypertensive group did not increase until after 34 weeks PMA, with the SBP index peaking at 40 weeks PMA. Average group SBP normalized as infants began treatment with spironolactone. Normotensive patients showed a blunted peak of SBP index at 40 weeks PMA, but did not achieve levels consistent with systolic hypertension. Most exposures to either intravenous or respiratory DEHP in hypertensive patients occurred early in life, beginning before any rise in SBP index, mostly complete by 36 weeks PMA. The median time between the subjects’ last DEHP exposure and diagnosis of hypertension was 3.57 weeks. Figure 1 b and c show the urinary exosomal expression of ENaC and pNCC for hypertensive and normotensive groups reported in 2-week intervals of postmenstrual age. Increased ENaC expression in the hypertensive group coincided with the peak SBP index around 40 weeks PMA. More interestingly, pNCC urinary exosomal expression in the hypertensive group was similar to that in the normotensive group through 34 weeks PMA. After 34 weeks PMA, pNCC expression was consistently higher in the hypertensive group compared to the normotensive group for the time when SBP index was higher in the hypertensive group.

Next, we move from the overview to our principal statistical comparison between the hypertensive group (at the diagnosis of hypertension), and the normotensive group (at 40 weeks PMA) for demographics, DEHP exposures, and risk factors for hypertension, which are shown in Table 3. Both intravenous and respiratory cumulative DEHP exposures were greater in the hypertensive as compared to the normotensive group (p = 0.029, p = 0.041 respectively). Individual patient data for exposures and onset of hypertension are available in the Appendix. Gender, race, birthweight, receipt of antenatal steroids, and diagnosis of chronic lung disease did not differ significantly between the groups. Gestational age was lower in the hypertensive group, (p = 0.012). In addition, the length of stay for the initial NICU hospitalization was longer in the hypertensive group (p = 0.027).

All hypertensive patients had PRA levels that were below 2.0 ng/mL/h, compared to two of nine normotensive infants (p = 0.003). The common lower limit of normal for PRA in children is 2.0 ng/mL/h. At the 40-week PMA principal comparison point, urinary cortisol-to-cortisone ratios were significantly higher in hypertensive infants as compared with normotensive infants (p = 0.02). Both ENaC and pNCC, markers of MR activation, were significantly higher in hypertensive patients (p = 0.042, p = 0.031).

Correlations between exposures and previously reported risk factors of hypertension in premature infants are shown in Table 4. Gestational age, birthweight, and length of stay are all correlated with cumulative respiratory DEHP exposure but not with IV DEHP exposure. Linear regression analysis demonstrated that SBP index at the principal comparison (onset of hypertension) was associated with cumulative IV DEHP exposure, but not respiratory DEHP exposure (Table 5). The urinary cortisol-to-cortisone ratio was predicted by intravenous DEHP exposure, and was highly predictive of SBP index. Figure 2 shows the close relationship between urinary cortisol-to-cortisone ratio and SBP index at onset of hypertension (or 40 weeks PMA for normotensive infants). In a multivariable model including IV DEHP exposure and the cortisol-to-cortisone ratio, the magnitude of the IV DEHP exposures effect on SBP index attenuates and approaches zero—evidence that the urinary cortisol-to-cortisone ratio mediates the relationship between DEHP exposure and SBP index (Table 6). Finally, both pNCC and ENaC were also predictive of SBP index at the diagnosis of hypertension (or 40 weeks PMA for normotensive infants) (Table 7).

The results for the DEHP urine metabolite testing are shown in Table 8. One sample was obtained in 14 of 20 enrolled infants—all at the study initiation visit (when blood pressure was normal in all). DEHP metabolites were present for six of ten infants with DEHP exposures documented during this 48 h window pre-sample. By N-1 Chi-square analysis, the presence of DEHP metabolites was related to the prior 48-h IV DEHP exposure (p = 0.030), as well as prior respiratory DEHP exposure (p = 0.033). In contrast, when no DEHP exposure occurred during this 2-day period, no patient was found with detectable urinary DEHP metabolites.

This study presents novel evidence that DEHP exposure is associated with systolic blood pressure increases in premature infants, including hypertension with larger exposures. The case for intravenous DEHP exposure increasing blood pressure is particularly strong given the following four findings: the linear relationship with SBP index, mediation of this relationship by the cortisol-to-cortisone ratio, and mechanistic links to MR activation.

The relationship between respiratory DEHP exposure and SBP is more complex. Although associated with hypertension in the studied cohort, no linear relationship was observed between respiratory DEHP exposure and SBP index, at least at the time of the diagnosis of hypertension. The inability to quantify respiratory DEHP exposure with precision confounded the ability to determine such a relationship. In addition, by virtue of the presence of DEHP in most respiratory equipment, respiratory DEHP exposure remains integrally linked with chronic lung disease and other markers of prematurity. This observation raises the question whether respiratory DEHP has a causal role in hypertension, or is merely just another marker of chronic lung disease and other risk factors related to extreme prematurity. A recent report of Stroustrup et al. showing premature infants exposed to respiratory support had significantly higher levels of DEHP urine metabolites compared to unexposed infants suggests the need for an in-depth look at the impact of respiratory DEHP on neonatal blood pressure [5].

The delay between DEHP exposures and peak blood pressure is notable, especially for intravenous DEHP exposures. One possible explanation for the apparent delayed effect was proposed by Martinez-Arguelles who offered that DEHP may act through a “second hit” phenomenon with initial latency [29]. Immaturity of sodium retention mechanisms of the very premature kidney is another explanation for this delay [30, 31]. The distal nephron may be unresponsive to MR stimuli until maturation of sodium transport function has occurred. Two observations support this explanation: ENaC expression in urinary exosomes, as an indirect measure of channel expression and activity, was not consistently elevated in hypertensive patients relative to controls until 40 weeks PMA; and, perhaps more significantly, pNCC was not elevated over the normotensive cohort prior to 36 weeks PMA, but manifested a higher expression thereafter until resolution of hypertension. This presumed tubular immaturity in the sodium channel and transporter expression may be analogous to that seen for neonatal potassium excretion, with a paucity of flow-dependent conducting potassium channels early in life [32].

The urine metabolite results suggest that both IV and respiratory DEHP exposures yield systemic absorption and formation of urinary MEHP. The data is not robust enough to estimate the magnitude or variability of DEHP that might leach from the various devices, nor can we predict how different respiratory DEHP exposures compare with each other or with IV DEHP exposures concerning the effect on blood pressure. Previous studies have shown a systemic transfer from DEHP-containing IV bags and PVC tubes into the bloodstream or other tissues [33, 34]. DEHP does not form covalent bonds to the PVC polymer, allowing DEHP to leach out of the PVC polymer matrix. The rate of DEHP leaching is influenced by lipophilic qualities of the fluid in contact, temperature, contact time, contact surface area, and fluid flow rate [33, 35‐39]. There are estimates of DEHP leaching rates for endotracheal tubes [4], blood transfusions [40], lipid infusions [33], but not for specific PVC respiratory circuitry or IV fluid contained in bags with DEHP. One NICU unit, which labeled itself as “phthalate-free” signifying attempts to use DEHP-free equipment when available, still found detectable levels of MEHP in all of the 149 urine specimens tested [5].

Studies on mechanisms and effects of phthalate toxicity performed mostly on rats have examined both maternal and developing infant exposure [14‐16, 29, 41]. Endocrine disruption effects, described as anti-androgenic, have been extensively reported from prenatal phthalate exposure and are summarized in a recent review [42]. Martinez-Arguelles et al. recently reviewed current knowledge of mechanisms mediating endocrine disruption from in-utero DEHP exposure that include alterations in adrenal steroidogenesis as well as evidence of epigenetic changes in rats leading to disease later in life [29, 41]. Studies of direct DEHP exposure in children have shown evidence of both endocrine and non-endocrine toxicity and have been reviewed elsewhere [14, 43, 44].

Aside from Zhao’s work, there are few other studies suggesting mechanisms that might explain how phthalate exposure could increase blood pressure. There is evidence that phthalates may cause oxidative stress [45, 46]. Kambia et al. treated infants and children with hyperalimentation given through PVC tubing containing DEHP and found that plasma DEHP levels correlated with malondialdehyde, a marker of free radical production [38]. Such oxidative stress may produce temporary changes in arterial tone, which increases blood pressure [47, 48]. Meanwhile, Zhao’s work, which exposed human tissue to various phthalates, convincingly demonstrated that MEHP (but not DEHP) was a potent inhibitor of human11β-HSD2 at clinically relevant concentrations [11]. When 11β-HSD2 is inhibited by black licorice, or when it is deficient as a result of genetic mutations (apparent mineralocorticoid excess), the result is sodium retention and hypertension [49].

Analysis of our data strongly suggests the mechanism of hypertension in our study cohort includes activation of the mineralocorticoid receptor. Our demonstration of increased sodium channel and transporter expression predicting higher SBP index aligns with activation of the MR, for ENaC at the height of the SBP index difference, and for pNCC throughout the entire time frame of the difference in SBP index between the two groups (PMA weeks 36 to 42). The evidence showing that urinary cortisol-to-cortisone ratio mediates the effect of IV DEHP on the SBP index suggests that inhibition of 11β-HSD2 by DEHP metabolites leads to MR activation in our cohort, in keeping with the observation by Zhao on the effect of DEHP metabolites on 11β-HSD2 [11].

The same clinical characteristics of low PRA, presentation around 35–40 weeks PMA, and an excellent response to spironolactone therapy—all seen in both our prior study [12] and in that of Al Awad et al. [13]—were also seen in this current study cohort. These similarities suggest that the mechanism of action found in this cohort may be generalizable to other premature infants with idiopathic or unexplained low-renin hypertension, with or without chronic lung disease.

Strengths of this study include its prospective design as well as urine metabolite data demonstrating DEHP transfer from medical devices to patients. Further, at least two analyses lend credence to the study’s hypotheses: (1) the linear relationship between IV exposures and SBP difference and (2) the analysis showing this relationship was mediated by urine cortisol-to-cortisone ratio. Weaknesses of the study include the lack of precise quantitation of DEHP exposures from respiratory devices and the small sample size. With only 18 subjects for analysis, we were limited in the complexity of our model building and therefore tested only the main variables of interest, which were IV and respiratory exposures to DEHP, and the effect that exposures had on SBP index and the cortisol-to-cortisone ratio. Ideally, we would have been able to control for important variables such as gestational age or similar markers of extreme prematurity previously linked to hypertension in premature infants.

In conclusion, both IV and respiratory DEHP exposures are greater in hypertensive patients as compared to normotensive patients. Cumulative IV DEHP exposure is predictive of SBP index, a relationship mediated via the cortisol-to-cortisone ratio. As DEHP is metabolized to MEHP, a known inhibitor of 11β-HSD2, a plausible mechanism of MR-mediated sodium retention, increased blood pressure, and low PRA is proposed. These results are provocative as to the potential role played by phthalates in idiopathic hypertension in premature infants. To determine the significance and safety of DEHP in respiratory tubing in regard to blood pressure effects, further studies are warranted.