Developmental changes of fluconazole clearance in neonates and infants in relation to ontogeny of glomerular filtration rate: literature review and data analysis

Background Fluconazole is frequently prescribed for the treatment of systemic fungal infection in neonates and infants. At present, prediction of fluconazole doses according to developmental changes in fluconazole clearance is not being done in these patients. We aimed to formulate a developmental model of fluconazole clearance taking into account the ontogeny of renal function, since the drug is largely eliminated renally. Methods We systematically retrieved the data of fluconazole pharmacokinetics and renal function in children and adults from databases (MEDLINE and Japan Medical Abstracts Society). Datasets were retrieved from individual children or groups from 9 studies comprising 55 neonates or infants at postmenstrual age (PMA) 27–58 weeks. Datasets were retrieved from 5 studies comprising 60 children and from 13 studies comprising 152 adults. Datasets of glomerular filtration rate (GFR) for individual pediatric subjects were retrieved from 4 studies comprising 187 neonates or infants. Results Fluconazole clearance normalized to body surface area (BSA) (CLBSA) in neonates was 1/3 to 1/4 of adult values, but CLBSA increased rapidly during the neonatal and infantile periods and attained near adult values at PMA 60 weeks. A significant correlation between CLBSA and PMA was observed in neonates and infants: CLBSA (mL/min/m2) = 0.26・ PMA (weeks) – 4.9 (r = 0.68, p < 0.001). In addition, the developmental time course of GFR normalized to BSA (GFRBSA) was fitted well to a sigmoidal model with the maximum GFRBSA of 149 mL/min/1.73m2, PMA associated with 50% of GFRBSA,max (PMA50) of 54 weeks, and the Hill coefficient of 3.7. A significant correlation between fluconazole clearance and GFR was found in neonates and infants: CL (mL/min) = 0.34・GFR (mL/min) – 0.53 (r = 0.84, p < 0.001). Assuming that plasma drug concentrations required for treating fungal infection are comparable between children and adults, fluconazole doses for pediatric patients with given PMAs may be predicted from adult doses (such as 100 mg/day) using size-normalized clearance as a scaling factor. The predicted doses for neonates and infants were largely within the ranges recommended in the prescribing information. Conclusions The present study indicates that fluconazole doses for neonates and infants may be predicted from developmental change of systemic clearance, the ontogeny of which parallels the maturation of nephron function.


Background
Numerous studies have been conducted to establish a comprehensive dosing formula for neonates, infants, and children who undergo tremendous developmental changes in pharmacokinetics and pharmacodynamics [1]. Recently, Anderson and Holford [2] reported that adult doses may be scaled down to pediatric doses using a child/adult body surface area (BSA) ratio or an allometric scaling factor (exponent of 3/4 of the child/adult body weight ratio) for children older than 2 or 3 years, but additional factor of maturation should also be taken into account for neonates and infants. However, whether the above approaches work well in predicting fluconazole doses for neonates and infants remains largely unknown.
Fluconazole is a triazole antifungal agent frequently used for the treatment of systemic candidiasis or cryptococcosis in pediatric patients including neonates and infants, at doses of 3 to 6 mg/kg [3]. However, these dosing schemes are largely empirical and are not supported by pharmacokinetic data. At present, no dosing formula has been established for estimating fluconazole doses considering developmental changes of systemic clearance (CL) of fluconazole.
With this background, we aimed to study the developmental time courses of fluconazole CL and glomerular filtration rate (GFR) as a function of postmenstrual age (PMA) in pediatric patients, using datasets retrieved systematically from relevant patient populations and adults using databases in English (MEDLINE) and Japanese (Japan Medical Abstract Society). Furthermore, we aimed to formulate an equation to estimate fluconazole doses for pediatric patients as a function of PMA by combining the relationship between CL and GFR and that between GFR and PMA in children. Here, we present the data indicating that fluconazole doses for neonates and infants may be predicted by our formula by comparing with the doses predicted using empirical formulas of Augsberger [4] and Crawford [5] as well as those recommended in the current prescribing information [3].

Search for pharmacokinetic data of fluconazole
We searched for studies on fluconazole pharmacokinetics performed in premature and full-term neonates, infants, children, and adults in the MEDLINE database using at a combination of the following keywords: "UK49858 OR fluconazole" AND "child* OR pediatric* OR infant* OR neonate* OR premature*" AND "pharmacokinetic*". Additionally, we searched for relevant studies performed in adults using keywords of "UK49858 OR fluconazole" AND adult* AND pharmacokinetic*. We also searched for relevant studies reported in Japanese in the database operated by Japan Medical Abstracts Society, using the same set of keywords in Japanese. We also searched for the articles cited in the retrieved literature. We collected CL of fluconazole in studies in which the drug was administered intravenously. In articles that did not report CL but included data of area under the curve from time 0 to infinity (AUC 0-inf ) after intravenous injection, we calculated CL by dividing AUC 0-inf by the corresponding doses. When plasma concentration-time data were available after oral administration, systemic clearance was estimated by assuming that oral bioavailability of the drug was 94%, as reported elsewhere [6]. When only AUC from time 0 to 24 h (AUC 0-24 ) were available, AUC were estimated by adding AUC 24-∞ estimated by extrapolation from concentration measured at 24 h and the elimination constant.
Developmental changes of fluconazole clearance and GFR as a function of PMA Fluconazole CL in pediatric patients retrieved from the literature were size-corrected or normalized to body weight (BW) and body surface area (BSA) (CL BW and CL BSA , respectively) as scaling factors for each patient. In the literature, individual body weights were available for most of the full-term neonates, but heights were available from approximately 30% of them. For non-Japanese neonates or infants at PMA 40 weeks or older without individual height data, the 50 percentile height for the corresponding chronological age according to the WHO growth chart for boys was used [7]. For Japanese infants, the growth chart for Japanese boys was used [8]. For most premature neonates, however, individual body weights and heights were not available in the literature. Their body weights and heights were estimated using Ahn's data table that lists the average values based on the records of 5014 premature infants [9], irrespective of ethnicity.
BSA for full-term infants was estimated according to the formula of Haycock et al. [10] as follows: For children and adults whose individual body weights and heights were available, BSA was estimated according to the formula of DuBois and DuBois [11].
In 13 articles on adult patients, the data of heights and weights were available in three articles comprising 39 individual patients. BSA of the respective patients were calculated using Eq. 2. In four articles, only mean heights and weights for the study populations were available. In these cases, group BSA was calculated using Eq. 2. In the remaining 6 articles, no height and weight data were available. In these cases, we assumed that the patients had standard body weight (63 kg) and height (170 cm), and hence BSA of 1.73 m 2 .
Developmental changes of fluconazole CL in premature and full-term neonates at PMA 60 weeks or younger were analyzed by a linear regression model as a function of PMA. For children aged above 5 years and adults, a linear regression model was used for the analysis using chronological age.

Developmental model for GFR
We searched for articles that reported individual GFR and demographic data (body weight and height) of neonates or infants. While individual body weights were available for all subjects, heights were unavailable in 3 of 4 articles. Therefore, for the analysis on the development of GFR, we estimated BSA according to the Neo-BSA w formula reported by Ahn [9].
Retrieved GFR (unadjusted for size) and GFR normalized to BSA (GFR BSA ) were fitted to a sigmoidal hyperbolic model as a function of PMA using the non-linear least squares regression method.
where GFR (t) is GFR at a given PMA, GFR max is the maximum GFR, PMA 50 is the PMA associated with 50% of GFR max , and n is the Hill coefficient indicating the steepness of the curve. We used this equation to estimate GFR of pediatric patients with fluconazole CL retrieved from the literature, since a previous study [12] showed that the developmental changes of GFR as a function of PMA were fitted well to the sigmoidal model.

Comparisons of fluconazole doses for neonates and infants estimated by empirical formulas with those predicted by the present CL-based formulas and those in prescribing information
We assumed that fluconazole CL in a representative adult is 0.23 mL/min/kg [13] and the corresponding standard daily dose of the drug is 100 mg daily for most systemic fungal infections [3]. Also assuming that average plasma concentrations required for eliminating susceptible pathogens are largely similar between adults and children, we estimated fluconazole doses for neonates and infants with PMA 28-60 weeks by multiplying the standard adult dose (100 mg/day) by a ratio of fluconazole CL estimated for a given child to that of the representative adult. Fluconazole CL for a child was estimated by multiplying fluconazole CL BW or CL BSA at a given PMA by the corresponding mean BW or BSA. Fluconazole CL BW or CL BSA at a given PMA was estimated using the linear regression equation for the corresponding CL as a function of PMA. Heights and body weights of infants at PMA older than 40 weeks (full-term neonates) were substituted by the corresponding 50 percentile values obtained from the WHO growth chart [7]. Heights and body weights of infants at PMA younger than 40 weeks (premature neonates) were estimated according to Ahn's table [9] containing mean values for the respective PMA. We estimated pediatric doses of fluconazole at PMA 28-60 weeks using two empirical formulas [4,5] as follows: We also calculated the recommended doses of the drug according to the descriptions in the current prescribing information [3]. Specifically, daily doses of 3 mg/kg and 3 to 6 mg/kg are recommended for the treatment of systemic candidiasis and cryptococcosis, respectively, for infants and children older than 1 month. However, neonates younger than 2 weeks and those aged 3-4 weeks are recommended to receive the same daily doses of the drug as in older children, but every 72 and 48 h, respectively. The current prescribing information for fluconazole does not contain dose recommendation for premature neonates [3]. We calculated the daily doses of fluconazole according to these recommendations for neonates and infants younger than PMA 60 weeks.

Statistical analysis
All statistical analyses were conducted using JMP Pro (ver. 12.2.0, SAS Institute Inc. USA). Correlation between fluconazole CL and PMA in neonates and infants, and correlation between drug CL and chronological age in children and adults were analyzed by the least-squares linear regression. Non-linear least square regression method was used to analyze the sigmoidal hyperbolic model between GFR and PMA for neonates and infants. A p value less than 0.05 was considered significant.

Developmental changes of pharmacokinetic parameters of fluconazole in pediatric and adult patients
Forty-two pharmacokinetic datasets of 55 neonates or infants at PMA 27-58 weeks were obtained from the literature [14][15][16][17][18][19][20][21][22] (Table 1). While the information about the countries where the studies were conducted is available, no information was given in literature regarding the ethnicity of the patients. In some studies [14,18,20], only means or ranges for the group were available. Saxen et al. [14] conducted pharmacokinetic studies in 7 patients: 3 times for 4 patients and twice for 3 patients at 6 mg/kg. Nahata et al. [21], conducted pharmacokinetic studies in 2 patients by intravenous and oral administrations 1 week apart and in 4 patients only by an oral administration. Wenzl et al. [22] studied pharmacokinetics of the drug in one patient twice with different doses. Collectively, the doses of fluconazole studied were 3-6 mg/kg for oral administration and 2-25 mg/kg for intravenous administration. There were large interindividual differences in elimination half-life (t 1/2 ) and CL: t 1/2 ranged from 10.7 to 88.6 h and CL normalized to body weight (CL BW ) ranged from 0.16 to 1.18 mL/ min/kg.
Sixteen pharmacokinetic datasets were retrieved from 60 children aged from 0.3 to 18.3 years (Table 2) [20,[23][24][25][26]. Specifically, Fujii et al. [20] reported the results of a multicenter clinical trial of fluconazole in which the drug was administered to 72 children and antimicrobial effects were studied. In their report, the pharmacokinetics of the drug intravenously or orally administered at different doses (3, 6, and 12 mg/kg) was studied in 7 neonates and 17 children, but only mean pharmacokinetic parameters for patients who received the same dose via the same route were available. In the study of Seay et al. [25], each child received the drug by intravenous and oral administrations and the data obtained from both routes were fitted to a pharmacokinetic model simultaneously. No data were available from children aged from 2 to 4 years. There was large inter-individual variability in the PK data: t 1/2 ranged from 11.9 to 42.3 h.
For adults, 60 pharmacokinetic datasets were retrieved from those aged from 18 to 65 years (103 healthy volunteers, 32 patients with AIDS, 9 patients with vaginal candidiasis, and 8 patients with burn) [27][28][29][30][31][32][33][34][35][36][37][38][39]. In the reports of Shiba et al. [27] and DeMuria et al. [36], all patients received the drug by intravenous and oral routes on different occasions, For these patients, only the pharmacokinetic data obtained after an intravenous administration were included in analysis. In the report of Ripa et al. [28], 18 patients were allocated to 3 groups (n = 6 each) and received the drug at different doses: 50 or 150 mg orally, or 100 mg intravenously. We calculated fluconazole CL from plasma concentration-time data obtained after oral administration assuming that the oral bioavailability of fluconazole was 94% [6]. Doses employed for oral administration were from 25 to 200 mg and those for intravenous administration were from 25 to 400 mg. T 1/2 ranged from 14.4 to 75.2 h; CL ranged from 8.8 to 26.2 mL/min; and CL BW ranged from 0.19 to 0.63 mL/min/kg ( Table 2).  [14], data were given as ranges of means for patients who received fluconazole at different PMA. In the studies of Fujii et al. [20] and Piper et al. [18], data were given as mean for individual groups. In the studies of Krzeska et al. [15], Kondo et al. [16], Seki et al. [17], Nahata et al. [21] and Wenzl et al. [22], data were given as ranges for individual patients' values a n = 4, b mean, c interquartile range, d the same patients were also studied in PO route

Relationship between fluconazole CL and PMA in neonates and infants compared to that in children and adults
While fluconazole CL in neonates and infants was much lower than that in children older than 5 years and adults due mainly to smaller body size, there was a significant positive correlation between fluconazole CL and PMA in neonates and infants (Fig. 1a): Y = 0.093X -2.45, r = 0.81 (p < 0.001). A significant correlation between fluconazole CL and age was also observed in young children and adults (Fig. 1b): Y = 0.154X + 14.74, r = 0.36 (p < 0.005).
To evaluate the development of fluconazole CL in size-adjusted values in subjects ranging from neonates to adults having different body sizes, we normalized fluconazole CL to BSA (CL BSA ) and body weight (CL BW ). CL BSA in neonates was 1/3 to 1/4 of that in adults, but it increased rapidly after birth and reached largely the adult values at PMA 60 weeks (approximately 5 months postpartum) (Fig. 2a). There was a significant positive correlation between CL BSA and PMA in neonates and infants: CL BSA : Y = 0.263X -4.92, r = 0.68 (p < 0.001). In contrast, a significant negative correlation between CL BSA and chronological age was found in children older than 5 years and adults (Fig. 2b): Y = − 0.097X + 14.22, r = − 0.36 (p < 0.005).
Fluconazole CL BW in neonates was comparable to that in adults (Fig. 3a). There was also a significant correlation between CL BW and PMA (Y = 0.011X -0.04, r = 0.46, p < 0.005), and CL BW at PMA 60 weeks apparently increased to a level exceeding the adult values (Fig. 3a). In contrast, there was a significant negative correlation between CL BW and age in children older than 5 years and adults (Fig. 3b): Y = − 0.006X + 0.50, r = − 0.60 (p < 0.001).

Ontogeny of GFR in neonates to young children
A total of 187 relevant datasets of GFR and age in children, spanning premature neonates through children aged 12 years, were retrieved from the articles of Rubin et al. [40], Fawer et al. [41], Coulthard [42] and van der Heijden et al. [43]. Briefly, Rubin et al. [40] reported individual data of GFR and age for 63 pediatric patients with chronological ages ranging from 2 days to 12 years and body weights ranging from 2.4 to 35.5 kg. They measured GFR by mannitol clearance. We assumed that no premature neonates were studied in their study, since there were no descriptions in the article suggesting premature labor and all neonates had gestational ages of 40 weeks or longer. Fawer et al. [41] studied GFR in 44 pediatric patients having gestational ages ranging from 28 to 43 weeks (postpartum ages from 0.5 to 19 days) and body weights at birth ranging from 1.1 to 5.38 kg. They measured GFR by inulin clearance. Coulthard [42] studied GFR in 39 pediatric patients with gestational ages ranging from 27 to 40 weeks (chronological ages from 2 to 33 days) and body weights ranging from 0.85 to 3.85 kg. They measured GFR by inulin clearance. Van der Heijden et al. [43] studied GFR in 41 pediatric patients with gestational ages ranging from 27 to 36 weeks (chronological ages from 3 to 11 days) and body weights at birth ranging from 0.81 to 2.74 kg. They measured GFR by inulin clearance. Individual body weights and heights of full-term infants were available in the article of Rubin et al. [40], whereas only individual body heights were available in the other articles [41][42][43]. Therefore, we decided to estimate BSA for these pediatric subjects using Ahn's equation (Eq. 3).
Developmental changes of GFR and GFR BSA analyzed as a function of PMA are shown in Fig. 4a and b, respectively. By fitting the retrieved datasets to a sigmoidal hyperbolic function model, we obtained the following parameters. For the GFR-PMA relationship, GFR max was 70 mL/min; PMA 50 was 113 weeks; and Hill coefficient was 2.5 (Fig. 4a). For the GFR BSA -PMA relationship, GFR BSAmax was 149 mL/min/1.73m 2 ; PMA 50 was 54 weeks; and Hill coefficient was 3.7 (Fig. 4b).

Relationship between fluconazole CL and GFR in neonates and infants
A significant correlation was observed between fluconazole CL and GFR in neonates and infants: Y = 0.34X -0.53, r = 0.84, p < 0.001 (Fig. 5). Since we have shown that GFR in a neonate or infant at a given PMA can be estimated using the sigmoidal hyperbolic equation (Fig. 4), fluconazole CL for the corresponding neonate or infant may be estimated by substituting his or her GFR into the regression equation shown in Fig. 5.

Comparisons between fluconazole doses for neonates and infants predicted using empirical formulas and using CL-based formulas in reference to recommended doses in prescribing information
Compared to the daily doses of fluconazole recommended in the prescribing information for neonates and infants [3], Augsberger's and Crawford's formulas appeared to overestimate the doses for neonates at PMA 40-42 weeks, and appeared to underestimate the doses for infants at PMA 52 weeks and older (Fig. 6). In contrast, daily doses of fluconazole for neonates and infants predicted by our CL-based method were essentially within the doses recommended by the prescribing information for infants at PMA 40-60 weeks. For premature neonates, the doses predicted by our method was approximately 50% smaller than those predicted by the 2 empirical formulas.

Discussion
To the best of our knowledge, the present study is the first to demonstrate significant increases in size-  (Figs. 1, 2 and 3). Particularly, CL BSA in neonates was substantially lower than adult value but increased rapidly after birth reaching the level of adults at approximately PMA 60 weeks (Fig. 2a). These data suggest that the development of fluconazole CL may be attributed not only to an increase in size of body (and organs associated with the elimination of fluconazole) but also to functional maturation of the organ(s) which is associated with elimination of fluconazole. Since fluconazole is eliminated mainly by the kidneys, our data imply rapid and substantial maturation in nephron function in these pediatric subjects. Our notion may be supported by the findings that GFR BSA also increased rapidly after birth, reaching a plateau that is largely comparable to adult levels by PMA 100 weeks (approximately age 1.5 years) (Fig. 4). Our data thus indicate that a b should not be estimated simply from usual adult doses using BSA or BW as size-scaling factors, but should be estimated taking into account the development of fluconazole CL in association with functional maturation of nephrons (i.e., GFR BSA ). We found that CL BSA obtained from neonates at PMA younger than 40 weeks was approximately 1/4 to 1/3 of adult values (Fig. 2a), whereas the corresponding CL BW appeared largely comparable to adult values (Fig. 3a). A previous study has shown that the developmental time courses of organ size or mass (such as kidneys) and BSA are almost identical [2]. In this context, we surmise that neonates may have substantially immature fluconazole CL per unit kidney mass (i.e., CL normalized to BSA), compared to adults ( Fig. 2a and b). Similar findings have been reported for other drugs [44][45][46]. These findings may be explained by a difference in the developmental time course of body weight and that of organs. The development of body weight has been shown to lag behind that of organ size and BSA [2]. As a result, the ratios of organ mass (such as liver and kidney mass) per body weight in children are greater, maximally two-fold, than those in adults. Consequently, fluconazole CL BW in neonates might have been overestimated and thereby appeared similar to that in adults, and fluconazole CL BW in infants around PMA 60 weeks and children aged 5 to 10 years might have been overestimated and appear to be greater than that in adults ( Fig. 3a and b).
We found that while unadjusted GFR (mL/min) in neonates were substantially lower than those in adults, GFR increased rapidly as shown by a PMA 50 of 113 weeks and appeared to reach a plateau around PMA 600 weeks or age 10 years (Fig. 4a). Rodin et al. [12] reported data being similar to ours and analyzed the curvilinear data with the sigmoidal model. We also adopted the model in the present study. Interestingly, the plateau level (149 mL/min/1.73m 2 ) for children was comparable to that of a representative adult male (130 mL/min/1.73m 2 ) [47] (Fig. 4b). Since GFR BSA represents the nephron function per standardized mass of kidney, this finding indicates that neonatal nephron function is very immature at birth but develops rapidly in early childhood reaching almost adult level by around age 2 years. Our findings are supported by previous studies on the ontogeny of renal morphology and physiology. A study has shown that although nephrogenesis is complete by 36 weeks of gestation, neonatal GFR is only approximately 5% of the adult value [48]. Neonatal nephron function is immature due mainly to vasoconstriction of the renal microvasculature, whereas the postnatal increase in renal mass is due almost entirely to tubular growth [48].
Assuming that plasma fluconazole concentrations required for the treatment of systemic fungal infection are comparable between children and adults, and that the  Fig. 6 Comparisons of fluconazole doses for neonates and infants estimated using various formulas for children at PMA ranging from 28 to 60 weeks. Open (○) and closed circles (•) connected with solid lines represent doses predicted by fluconazole clearance normalized to body weight (CL BW ) and fluconazole clearance normalized to body surface area (CL BSA ), respectively in the present study. Open (□) and closed squares (■) represent those predicted by the Crawford's and Augsberger's formulas, respectively. The shaded area represents doses calculated according to the recommended doses for treating systemic candidiasis for children (3 to 6 mg/kg/ day) in the latest prescribing information [3] typical adult dose of the drug is 100 mg/day, it is possible to estimate individual fluconazole doses for neonates or infants from adult doses using fluconazole CL of children as a scaling factor. Since there is good relationships between GFR of neonates and infants and PMA (Fig. 4) and between fluconazole CL and GFR (Fig. 5), fluconazole CM for a neonates and infants may be estimated by PMA. The fluconazole doses for neonates at PMA 40-44 weeks predicted by the two empirical dosing formulas are higher than those estimated using CL BSA or CL BW and those recommended in the prescribing information. In contrast, the doses estimated using CL BSA or CL BW were similar to those recommended in the prescribing information for systemic candidiasis (3 to 6 mg/kg/day). For infants at PMA 44 weeks and older, fluconazole doses predicted by our CL-based methods and the 2 empirical formulas are essentially within the range of doses recommended by the prescribing information.
The present study has some limitations. Since pharmacokinetic studies for children are difficult to conduct, only sparse data and datasets with incomplete clinical information are available from the literature. As a result, we often had to utilize average demographic data from subjects with comparable PMA or mean pharmacokinetic parameters for a group in estimating parameters for individual subjects. In addition, since no pharmacokinetic data were available for children from PMA 60 weeks to age 5 years, we cannot estimate the development of fluconazole CL during this period. Finally, CL-based prediction of fluconazole doses for pediatric patients with fungal infection would only be valid if therapeutic plasma drug concentrations are similar between children and adults. Since neonates and infants are immunologically less competent than older children, whether they need higher fluconazole concentrations than adults for treatment remains to be studied. The reason why CL BW and CL BSA showed weak, but significant, negative correlations with age (Figs. 2b and 3b) are most likely associated with an age-dependent decrease in GFR [49].

Conclusions
This study reveals that size-normalized fluconazole CL in neonates and infants is substantially lower than that in adults, but fluconazole CL develops rapidly during early childhood. Immature nephron function in neonates and infants may contribute to immature renal clearance for fluconazole. Fluconazole doses for neonates and infants may be estimated by taking into consideration developmental change of CL BSA that is associated with the development of GFR.
Abbreviations BSA: Body surface area; CL: Systemic clearance; CL BSA : Clearance normalized to BSA; CL BW : Clearance normalized to body weight; GFR: Glomerular filtration rate; GFR BSA : Glomerular filtration ratio normalized to BSA; GFR BSA,max : The maximum value for GFR; PMA: Postmenopausal age; PMA 50 : Postmenopausal age associated with 50% of the maximum GFR