Drug-Induced Renal Damage in Preterm Neonates: State of the Art and Methods for Early Detection

2.1 Regulatory Aspects

The general lack of appropriate pharmaceutical formulations for children promoted legal obligation to perform studies to support use in the pediatric population [5]. In particular, from 1997 the US FDA with the ‘Modernization Act’ and from 2002 the European Medicines Agency (EMA) with the consultation paper ‘Better Medicines for Children’ undertook actions to promote clinical trials in children for both authorized and investigational medicinal products [6, 7]. These documents were then implemented in the ‘Pediatric Research Equity Act’ (FDA, 2003) [8] and the ‘Regulation on Medicinal Products for Paediatric Use’ (EMA, 2006) [9].

The main objectives of the regulatory actions were to ensure high-quality research into the development of medicines for children, to make sure that the majority of them are specifically authorized for such use, and to ensure the availability of high-quality information [10]. To support companies in new development strategies, a recent FDA regulatory initiative (FDA Safety and Innovation Act [11]) included the neonatal population among its main topics; specific expertise in neonatology is requested within the Office of Pediatric Therapeutics and the Pediatric Review Committee. In Europe, the EU Regulation of 2006 established the Paediatric Committee (PDCO), within the EMA, which was responsible for the evaluation of Pediatric Investigation Plans (PIPs) presented by companies as mandatory programs for the pediatric indication [9]. Other health organizations have also faced the problem of appropriate drug use in the pediatric population; the World Health Organization (WHO) developed a program on medicines for children, resulting in the WHO Model Formulary on how to use essential medicines for children [12]. Several National Health Institutes published guidebooks on the use of widely administered drugs in children (e.g. Italian Institute of Health [13]).

Some results of the legislative actions undertaken so far have been provided by the approval of the 500th pediatric label change in 2013 by the FDA [14], the 221 changes regarding safety and efficacy, for submission of old or new studies, and 89 additions of dosing information for children in the EU [15]. Despite this, neonates (especially preterm newborns) are neglected in clinical trials and a worldwide gap of knowledge on the use of medicinal products in neonates still persists [16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27].

2.2 Unlicensed and Off-Label Drug Use Among Preterm Neonates

Definitions for ‘unauthorized’ (or ‘unlicensed’) and ‘off-label’ use of drugs may differ among authors but, in general, a medicinal product is considered unauthorized if it has no indication for the specific population, and is off-label when not included in the approved labeling; for example, a change in the formulation (i.e. to obtain a lower dose) or in the method of administration [16, 17, 23, 25, 26, 28].

Haslund-Krog et al. [25] took into account the 100 most used drugs in children from the national Danish database and found that 13 are used off-label (defined as one of the following: absence of pediatric information, lack of clinical data, contraindication for use in children, drug substances being an extemporaneous preparation or unlicensed). Twenty-one percent of drug substances reviewed had a PIP, and only six (28.6 %) of these included preterms and newborns in their PIPs. A study in Finland, carried out to verify the impact of the European Pediatric Regulation, actually found that off-label use in newborns was higher in 2011 than in 2001 (51 vs. 22 %), whereas prescriptions of unauthorized medicines decreased from 31 % in 2001 to 25 % in 2011 [26]. The prevalence of off-label use in NICUs was reported to be 46.5 % in Italy [16] and 39 % in Ireland [17].

In the US, less than 0.5 % of all drug exposures in neonates was approved by federal legislation. Revision of the FDA database for studies submitted between 1997 and 2010 involving the pediatric population, showed 406 pediatric labeling changes, of which only 6 % included new neonatal information [20].

Several reasons explain why only few clinical trials are performed in neonates, i.e. ethics problems, parent distrust, low economic interest by drug companies, and low funds for specific age-related studies [29]. Moreover, feasibility problems occur in studying drugs in newborns according to good clinical practice, i.e. the amount of blood required for measuring safety laboratory values when added to the volume required for pharmacokinetic measurements quickly exceeds safe limits; additionally, organ function is difficult to measure, and even harder is the evaluation of efficacy without gold standards [18]. Other strong obstacles to research are the limited number of neonates with a given disease in each individual center, highly variable practices (drugs and duration of treatment, prophylaxis vs. treatment, monitoring, medical care, including type of ventilation) and lack of support for networks.

Although off-label or unlicensed use of medicines does not necessarily imply inappropriate use (i.e. benefits outweigh potential risks) and is often necessary, it has been associated with increased adverse drug reactions (ADRs) in children [30] and neonates [19]. Risks factors in newborns are the immaturity of most organ functions and the developmental changes occurring in the postnatal period, which could significantly influence plasma concentration of medicines, their safety, and their efficacy [31]. Since neonates frequently suffer from multiple concomitant conditions, a combination of medicinal products may result in a higher risk of interactions [19].

4.1 Drug-Induced Kidney Injury

Since the kidney is the main organ for excretion of a drug and its metabolites, preterm neonates are at higher risk of overdosing due to immature renal function [32]. The renal contribution to excretion occurs at several levels, i.e. glomerular filtration, tubular secretion and reabsorption, or intracellular enzymatic processing, mainly in proximal tubular cells [46, 47]. Preterm babies show specific glomerotubular imbalance because GFR sharply increases after birth while tubular secretion matures slower, but more steadily, to result in a proportional higher capacity for postnatal maturation of renal tubular function [42]. Pharmacokinetic and pharmacodynamic parameters differ not only between preterm and term neonates but also among preterms. As a matter of fact, the cut-off for achieving nephrogenesis, and thus the capacity of drug renal elimination, is assumed at the 33rd week of gestational age [32].

4.1.3 Classification of Drug-Induced Kidney Injury

The clinical classification of renal damage is challenging because symptoms are non-specific and simply consist of urine flow interruption. The pathophysiological classification of kidney injuries is based on the mechanism of renal dysfunction: (a) hemodynamically mediated, (b) glomerulonephritis, (c) acute tubular necrosis, (d) interstitial nephritis, (e) papillary necrosis, and (f) obstructive nephropathy (Fig. 2).

Open image in new window

1. (a)

   Hemodynamically-mediated renal failure arises when mechanisms of maintenance of intra-glomerular pressure are impaired, e.g. decrease in blood flow to the kidney, vasoconstriction of glomerular afferent arteriolar, or vasodilatation of glomerular efferent arterioles. The autoregulation of intraglomerular pressure is essential to maintain GFR and urine output, and the main influencing factors are circulating prostaglandins (vasodilatation of afferent arterioles) and circulating angiotensin II (vasoconstriction of efferent arteriole). Thus, non-steroidal anti-inflammatory drugs (NSAIDs), angiotensin II-converting enzymes (ACEIs), angiotensin II-receptor blockers (ARBs), and calcineurin inhibitors are among the drugs inducing this type of injury [68].

    
2. (b)

   Renal damage in glomerulonephritis syndrome is due to inflammation reactions, protein infiltration, and cell-mediated immunity processes. Decreased serum albumin levels induce over-production of proteins by the liver, which contributes to the development of nephrotic syndrome, i.e. proteinuria, hypoalbuminemia, hyperlipidemia, and edema. Treatment with NSAIDs, ampicillin, rifampin, foscarnet, and lithium causes renal failure through this mechanism [68, 69].

    
3. (c)

   Acute tubular necrosis is characterized by tubular cell death as a result of ischemic effects of toxic exposure to agents such as aminoglycosides and amphotericin B. Tubular cells have the highest exposure to circulating toxins because they concentrate and reabsorb the glomerular filtrate; tubular damage is mediated by the alteration of mitochondrial function, interfering with tubular transport, increasing oxidative stress, or forming free radicals [70, 71].

    
4. (d)

   Administration of β-lactam antibiotics, vancomycin, NSAIDs, and calcineurin inhibitors correlates with acute allergic interstitial nephritis, which is mediated by inflammatory reactions in the renal tubules and interstitium [68, 70]. Most of those cases occur as a hypersensitivity reaction to a substance, thus heterogeneous classes of drugs may be the cause of this type of renal injury.

    
5. (e)

   Papillary necrosis is associated with the use of analgesic agents, such as acetaminophen and NSAIDs, but also kidney infections are reported to cause this damage [72, 73].

    
6. (f)

   Another type of injury is the mechanical obstruction of renal function; some drugs (acyclovir, ciprofloxacin, sulfonamides, methotrexate) or metabolites could precipitate, after glomerular filtration, into tubules and act as mechanical obstruction of urine flow, leading to renal insufficiency [70, 74].

    

4.2 Treatment-Related Renal Damage by Drug Class

Few and heterogeneous data are available on drug use in NICUs. In most cases, only unauthorized and off-label drug use was investigated, in order to understand the impact of the European Pediatric Regulation [16, 17, 22, 25, 26, 27], whereas Neubert et al. [28] and Clark et al. [75] reported overall medication use in German and US NICUs, respectively (Table 2). In these national reports, antimicrobial and cardiovascular agents, analgesics and respiratory drugs are the most commonly used in preterm neonates [16, 17, 25, 26, 28, 75]; antibiotics have the highest exposure rates and also the highest differences between nations in terms of drug substances. In the US, ampicillin and gentamicin have been reported as the most frequently used antibiotics, followed by cefotaxime and vancomycin [75]; the most used are piperacillin and vancomycin in Germany [28], cefotaxime in Finland [26], ampicillin and amikacin in Italy [16], and gentamicin in Ireland [17]. This heterogeneity in the choice of antibiotics is related to the lack of guidelines and comparative clinical studies, as documented by a Cochrane review comparing different antibiotic regimens for suspected late-onset sepsis in newborns [76].

Aminoglycosides, such as amikacin and gentamicin, are widely prescribed in NICUs for the prevention and treatment of infections. Aminoglycosides have well-known nephrotoxic effects; they are partially reabsorbed into proximal tubular cells and their cationic charge facilitates their binding to membrane phospholipids of tubular epithelial cells. This leads to the activation of intracellular transport into lysosomes, where they bind to phospholipids and inhibit phospholipase activity. This process precipitates in acute tubular necrosis development, as the result of accumulation of phospholipids, interruption of cellular functions, and cell death [77]. Aminoglycoside-induced AKI is characterized by elevation of serum creatinine, blood urea nitrogen, and reduction of urine output, but is reversible with drug discontinuation. Age, coadministration of other nephrotoxic drugs, and duration of therapy are among the factors increasing the risk of aminoglycoside-induced nephrotoxicity [68]. It has also been investigated whether different regimens of aminoglycoside treatment (gentamicin ‘once a day’ vs. ‘multiple doses a day’) may reduce nephrotoxicity, but no changes were found [78]. Recent studies on aminoglycoside use in preterm neonates evidenced that GFR levels were, surprisingly, almost unchanged, at least after short-term administration in the early postnatal period at therapeutic dosages [79].

Another class of antibacterial agents used for both the prevention and treatment of recurring infections in the NICU are β-lactam antibiotics, particularly ampicillin, piperacillin, and cefotaxime. The administration of these molecules has been correlated with nephrotoxicity, particularly with acute interstitial nephritis, affecting the renal tubules and interstitium with consequent acute renal failure. β-Lactam-induced nephrotoxicity is mediated by hypersensitivity reactions, believed to be cell mediated as T cells are the main pathological interstitial finding; it is also possible to find lymphocytes, monocytes, and eosinophils in the interstitial infiltrate. Renal effects come along with typical systemic manifestations, such as fever, eosinophilia, and skin rash, which are not commonly seen in acute interstitial nephritis caused by other drug classes. Kidney injury is reversible after discontinuation of the drug, and it may take weeks to months to be complete. In this case, glucocorticoid therapy may improve recovery [70, 80]. Acute interstitial nephritis can also result from vancomycin treatment, as a consequence of antigen precipitation into the interstitium, which triggers an immune reaction [70, 81].

Amphotericin B is the most nephrotoxic antimicrobial drug and remains the drug of choice to treat severe systemic fungal infections (also in neonatal wards). Tubular cell toxicity due to amphotericin B depends on its binding to sterols of cell membranes and consequent alteration of membrane permeability to sodium; the increase of intracellular sodium concentration and the activation of vasoconstrictor prostaglandins lead to vasoconstriction of renal vessels. As a consequence, RBF and GFR decrease and ischemic injury occurs [82]. Amphotericin B-induced distal tubulopathy is evidenced by increased serum urea and creatinine concentrations, as a result of decreased GFR, and hypokalemia, metabolic acidosis, hypomagnesemia, and loss of urine concentration ability [83]. Risk factors for amphotericin B nephrotoxicity include renal insufficiency, volume depletion, large individual and cumulative doses, and coadministration of other nephrotoxic drugs [84, 85]. Because of the clinical usefulness of amphotericin B, strategies were sought to decrease its adverse effects, e.g. its lipid-associated formulation, which showed similar efficacy but lower nephrotoxicity. However, the cost of these new formulations limited their use to patients at highest risk of renal failure [86]. To prevent amphotericin B nephrotoxicity in children, Goldman and Koren [83] suggested introducing, as standard care, sodium supplementation pre- and post-infusion to reduce vasoconstriction, as well as close monitoring of renal function over a lipid-associated formulation, unless significant nephrotoxicity develops [86].

Among diuretics, furosemide is the preferred drug in the NICUs [16, 28, 75] and, along with the other ‘loop’ diuretics, it increases the rate of urine flow and sodium excretion by affecting the delivery of solutes out of the loop of Henle. Furosemide intake is associated with water and electrolyte imbalances, particularly hyponatremia, hypokalemia, and hypochloremic alkalosis. Moreover, hypercalciuria occurs with long-term furosemide therapy, and the risk of developing nephrocalcinosis is higher in low birth-weight neonates (underdevelopment of renal functions, such as decreased GRF, low citrate excretion and alkaline urine). It is also reported that, in most cases, the discontinuation of treatment is associated with resolution of the renal calcification. The use of furosemide in neonates may also cause ototoxicity, and strategies have been studied in order to avoid hearing loss, including slow continuous infusion and short-term administration, and it is strongly recommended to avoid the association of furosemide with aminoglycosides. Additionally, furosemide enhances prostaglandin E₂ synthesis, which acts as a potent dilator of the PDA, thus inhibiting the effect of drugs used to close it. Careful monitoring is needed in babies with respiratory distress syndrome treated with furosemide because they are at higher risk of precipitation of a symptomatic PDA [87].

In neonates, short-term treatment with NSAIDs (especially ibuprofen and indomethacin) is mostly for the management of PDA. Although the overall risk-benefit profile of this drug class is positive, NSAIDs are implicated in AKI development in patients with pre-existing renal diseases, poor RBF, or when other nephrotoxic agents are coadministered (i.e. aminoglycosides, diuretics). In these cases, activation of the renin-angiotensin system and the rise of circulating catecholamines stimulate renal prostaglandin production, which protects renal perfusion and glomerular filtration. The inhibition of prostaglandin synthesis by NSAIDs may lead to a rapid decline of renal function after only a few doses. In addition, NSAIDs are also responsible for acute interstitial nephritis and glomerulonephritis [77]. Renal failure is reversible after discontinuation.

A new therapeutic approach to close PDA in preterm neonates consists of the use of acetaminophen instead of classic NSAID treatment. Acetaminophen is widely prescribed among neonates to treat moderate pain or fever, and it seems to have a better safety profile than ibuprofen and indomethacin; however, the different therapeutic scheme used for the management of PDA has not been investigated in terms of respiratory and renal long-term outcomes [88].

In this scenario, we mention the use of caffeine citrate (caffeine), largely prescribed as a first-line drug for the treatment or prevention of apnea in preterm neonates [90]. It has been hypothesized that the hemodynamic effects of caffeine might have long-term consequences on the health status of previously exposed children. However, recent studies show that even if caffeine administration in NICU patients has short-term adverse effects, such as increased heart rate, jitteriness, irritability, seizures [91], and delayed weight gain [92], the survival rate of children up to 5 years of age is improved for caffeine treatment compared with placebo, without neurodevelopmental impairment [93].

Specific features of the neonatal kidney, which complicate renal drug handling, may influence drug-induced toxicity. Neonates seem to be more sensitive to vasoactive aspects (e.g. ibuprofen-induced reduction in GFR), and the immaturity of allergic reaction capacity may facilitate the onset of interstitial nephritis. The elimination of many drugs is mediated by tubular kidney transporters, which have lower activity in neonates [94]. This may cause intracellular accumulation of active substances; however, the implication of these factors on drug elimination and the development of nephrotoxicity in early infancy is a matter of investigation.

5.1 Biomarkers to Monitor Renal Safety of Drugs

Serum creatinine levels (sCr) are widely used in the diagnosis of AKI, but sCr-based definitions are not highly predictive because (1) sCr measures function, not injury; (2) sCr may not change until 25–50 % of kidney function is compromised; (3) sCr overestimates renal function due to tubular secretion of creatinine at lower GFRs; (4) sCr varies by muscle mass, hydration status, sex, and age, and its measurement can be affected by endogenous substances; (5) sCr is not specific to different types of AKI; (6) in some conditions, such as pre-renal azotemia, sCr might be non-specific for AKI; and (7) sCr cannot be applied in the early postnatal period because serum Cr reflects maternal kidney function. In addition, sCr levels show significant variation, depending also on the degree of prematurity [95, 96].

The identification of novel biomarkers will allow early detection of AKI and could improve patient outcomes [51]. Figure 3 provides a general overview on promising biomarkers for renal damage in the entire population (especially in adults). In the neonatal population, the only biomarkers under investigation are serum and urinary neutrophil gelatinase-associated lipocalin (NGAL), urinary interleukin (IL-18), osteopontin (OPN), kidney injury molecule-1 (KIM-1), β-trace protein (BTP), and serum cystatin C (CysC).

Open image in new window

Studies on AKI biomarkers in neonates have been mainly performed in specific populations at risk, such as VLBW, asphyxiated newborns, and children undergoing cardiac surgery with cardiopulmonary bypass (CPB). Although several new candidate biomarkers have been studied, some turned out to depend on both gestational age and birth weight. Moreover, the results of these studies are often difficult to evaluate because sCr has been inappropriately used as the gold standard marker for AKI [97].

5.2 Pharmacokinetic Aspects

The real impact of immaturity on drug disposition in this population has not been fully defined, especially because of the great heterogeneity in disease characteristics and comedications, in addition to age (postnatal, postmenstrual), and, moreover, because performing pharmacokinetic studies in the neonatal population remains challenging. A novel and promising approach to improve drug therapy and to facilitate studies in neonates and young infants is knowledge integration through pharmacokinetic modeling, which is either driven by mechanism-based pharmacokinetics or by physiology-based pharmacokinetics. Mechanism-based models use a ‘from compound to model’ approach; based on compound-specific observations, covariates are described, resulting in mechanism-based models that reflect a semiphysiological maturational function [118, 119, 120]. In contrast, physiology-based pharmacokinetics use a ‘from physiology to clinical observations’ approach; driven by characteristics of neonatal physiology (e.g. weight, cardiac output, renal function) and anatomy, as well as its age-related changes, a physiology-based pharmacokinetic model is developed [118, 119, 120]. Population pharmacokinetic approaches hereby provide a statistical tool to study and explore variability in pediatric drug responses among and within individuals representative of those in whom the drug will be used clinically [121, 122, 123]. Such models can drive study design and convert neonatal pharmacological studies from explorative to confirmatory [121, 122, 123]. Once a model is developed, validation is needed in a prospective trial. This methodology will improve dosing guidelines at the individual level [121, 122, 123].

The feasibility of these concepts for GFR-driven clearance in neonates has been described using a stepwise approach. In the first step, amikacin clearance was predicted by weight at birth (maturation until delivery), postnatal age (maturation since delivery), and disease characteristics (ibuprofen exposure, peripartum asphyxia) [124]. Since amikacin clearance reflects GFR ontogeny, it is reasonable to assume that the same pharmacokinetic model and its covariates will also accurately predict clearance maturation of other compounds that are cleared by the same route. Using datasets of netilmicin, vancomycin, tobramycin, and gentamicin in neonates, it was confirmed that the performance of the amikacin GFR model was similar to the independent, compound-specific models [125]. This provides evidence that the neonatal amikacin covariate model indeed contains physiological information since the information extracted from the first dataset on a given compound can be extrapolated to another compound that undergoes elimination by the same route as long as the clinical characteristics remain similar. Consequently, the same approach has been used to describe GFR maturation throughout life [125].

A very specific covariate that also needs further exploration in combination with all the aforementioned aspects on drug-related renal toxicity in preterm neonates is the relation between pharmacogenomics, drug disposition, and drug-induced renal damage. The preterm newborn is deficient in drug-handling capacity, including drug metabolizing enzymes (e.g. cytochromes P450), drug transporters (e.g. renal tubular organic anion transporters), or receptors (e.g. renin-angiotensin-aldosterone system, dopamine) [4]. This immature drug-handling capacity may potentially be superimposed to genetic polymorphisms determining drug metabolism and transport, thereby further increasing the interpatient variability in drug dose, response, and renal side effects [126]. To illustrate the potential relevance of pharmacogenetic, polymorphism-driven renal drug toxicity, we refer to the work on ifosfamide in infants and toddlers [127].