Relationship of Caffeine Dosing With Serum Alkaline Phosphatase Levels in Extremely Low-Birth-Weight Infants

Abstract

OBJECTIVE

To determine whether patients receiving higher doses of caffeine have increased alkaline phosphatase (ALP) levels, as a biomarker for osteopenia.

METHODS

This descriptive, retrospective study included 152 extremely low-birth-weight infants (ie, <1 kg) admitted from January 1, 2007, to September 30, 2009, who received caffeine for >2 weeks. Patients were divided into a low-dose (<7.5 mg/kg/day) and high-dose (≥7.5 mg/kg/day) group based on maximum caffeine dose received. The primary objective was to compare peak ALP levels between groups. Secondary objectives included a comparison of caffeine regimens and risk factors for osteopenia between groups and identification of factors significantly related to increase in ALP. Between-group analysis was performed using the chi-squared or Fisher exact test and Wilcoxon Mann-Whitney median test or t-tests where appropriate. A linear regression model was used, with peak ALP as the dependent variable.

RESULTS

A majority of the patients (n=122) were included in the high-dose caffeine group. No significant difference in maximum ALP level between groups (median, 599.5, versus 602.5 units/L, p=0.72). Gestational age and birth weight were inversely related to ALP, whereas parenteral nutrition duration was directly related. No significant relationship between caffeine dose and ALP was identified.

CONCLUSIONS

In this preliminary study, using ALP as a biochemical marker for bone turnover, there does not appear to be a dose-related effect between ALP and caffeine dose.

INTRODUCTION

Caffeine is one of the most frequently prescribed medications in the neonatal intensive care unit (NICU).¹ It is commonly prescribed for the treatment of apnea of prematurity and to facilitate extubation. For the treatment of apnea, the recommended dosage for caffeine citrate is a 20 mg/kg loading dose (10 mg/kg caffeine base), followed by a maintenance dose of 5 to 10 mg/kg/day (2.5-5 mg/kg/day caffeine base).² In addition to increasing the central inspiratory drive, caffeine can also produce renal effects such as diuresis and natriuresis.^3,4 Additionally, Zanardo and colleagues⁵ noted that renal excretion of calcium was increased in premature neonates following 5 days of treatment with caffeine base at a maintenance dose of 2.5 mg/kg/day (equivalent to caffeine citrate, 5 mg/kg/day). Patients in that study who received caffeine had a 10- to 15-fold increase in urine calcium excretion compared to controls.

In the neonatal population, additional loss of calcium can promote the development of osteopenia of prematurity, also known as metabolic bone disease of prematurity. This disease is characterized by decreased bone mineral density in comparison to intrauterine bone density at a comparable gestational age. Development of osteopenia is inversely related to gestational age and birth weight, with approximately 55% of extremely low-birth-weight infants (ELBW; i.e., <1000 grams) presenting with radiographic changes.⁶ In addition to loss of calcium, development of osteopenia can occur secondary to inadequate supplementation or depletion of any of the essential vitamins or minerals for bone development (e.g., calcium, phosphorus, or vitamin D). Several risk factors for development of osteopenia have been identified and include prematurity, prolonged use of parenteral nutrition (PN) for >3 to 4 weeks, and immobilization (ie, paralysis).^7,8 Additionally, medication use (e.g., corticosteroids, loop diuretics, methylxanthines, and antiepileptics) has been associated with development of osteopenia.^7–9 Diagnosis of osteopenia is often confirmed with radiographic evidence; however, milder forms or early onset of the disease may not be detected because these radiographic changes are not apparent until bone mineralization is reduced by at least 20%.⁸ In the clinical setting, alkaline phosphatase (ALP) is frequently used as a biochemical marker for osteopenia due to its ease of measurement. Because changes in ALP often precede radiologic changes, an elevation in ALP may prompt additional radiologic work-up for diagnosis of osteopenia.

The increased urinary loss of calcium associated with use of caffeine could potentiate the development of osteopenia in this population. It is unknown whether calcium depletion is a dose-related effect, with those receiving higher doses of caffeine being at greater risk for development of osteopenia. The objective of this study was to determine whether higher doses of caffeine are associated with development of osteopenia, using ALP as a biomarker.

METHODS

Study Design

This was a descriptive, retrospective study of ELBW (ie, ≤1000 g) neonates admitted at ≤48 hours of life to the NICU from January 1, 2007, through September 30, 2009, who received caffeine for >2 weeks duration. This study was conducted in a tertiary care, academic hospital with an 88-bed level III-C NICU. Following Institutional Review Board approval, patients were identified from the NeoData electronic database (Isoprime Corp., Lisle, IL). Patients were excluded from analysis if they had congenital defects of bone formation/mineralization or congenital defects of the renal or endocrine system. Additional exclusion criteria were prenatal exposure to anticonvulsants, requirement for dialysis, use of enteral calcitriol, and death at <4 weeks of life. Those patients who met inclusion criteria were further subdivided into “low” (ie, <7.5 mg/kg/day) and “high” (≥7.5 mg/kg/day) dose caffeine groups, based on the maximum daily dose of caffeine citrate received.

Data Collection

For each patient included in the study, demographic data were collected, which included gender, completed weeks of gestation, birth-weight, and NICU length of stay. Data collection for the caffeine regimen included completed gestational age at caffeine initiation, duration of caffeine use, maximum caffeine citrate dose (mg/kg/day), and cumulative caffeine citrate dose (mg). Data collection of concomitant risk factors for osteopenia included use of loop diuretics (cumulative dose, mg/kg), corticosteroids (cumulative dose, mg/kg), paralytic agents (duration), or phenobarbital (duration); duration of PN; and weekly average of calcium (mg/kg), phosphorus (mg/kg), and vitamin D (IU) intake from both enteral and parenteral nutrition. Due to the retrospective nature of the study and the fact that radiographic studies were not performed on all patients, ALP was used as a surrogate marker for bone breakdown related to development of osteopenia. The maximum serum ALP level during the time between initiation and 1 week following discontinuation of caffeine was recorded. For those patients discharged on caffeine, peak ALP levels were collected within the time frame of admission. In addition to these variables, radiology reports were also reviewed for documentation of nephrocalcinosis, pathological fractures, or osteopenia. Development of necrotizing enterocolitis (NEC) and conjugated hyperbilirubinemia were also noted.

The primary objective of this study was to compare the peak ALP level between low and high-dose caffeine groups. Secondary objectives included a comparison of caffeine regimen between groups, including cumulative dose and duration of exposure. Additionally, we compared risk factors for osteopenia that included exposure to medications associated with osteopenia, duration of PN use, and supplementation of essential vitamins and minerals.

Statistical Analysis

Descriptive statistics were performed for all study variables stratified by caffeine dose group. Between-group analysis was performed using the chi-squared or Fisher exact test, as appropriate, for categorical variables, and Wilcoxon Mann-Whitney median tests or Student t-test, as appropriate, for continuous variables. Additionally, a general linear regression variable selection procedure was employed to determine the variables that are significantly related to changes in ALP. Independent variables considered in the variable selection procedure included gender, gestational age, birth weight, length of stay, gestational age at which caffeine was initiated, duration of caffeine use, maximum caffeine dose, cumulative caffeine dose, use of concomitant agents (e.g., loop diuretics, steroids, paralytics, antiepileptic agents), and duration of PN. Data were analyzed using SAS software version 9.2 (Cary, NC).¹⁰ A p value of <0.05 was considered statistically significant.

RESULTS

Patient Demographics and Maximum ALP

A total of 172 ELBW infants were identified within the designated study period. Twenty patients were excluded for reasons including admission at >48 hours (n=5), enteral calcitriol (n=6), duration of caffeine <2 weeks (n=4), prenatal antiepileptic exposure (n=1), and congenital defects of bone, kidney, or endocrine system (n=4). A total of 152 patients were included in the study, with 30 patients in the low-dose caffeine group and 122 patients in the high-dose caffeine group. Baseline demographics of the low- and high-dose caffeine groups are shown in Table 1. Gestational age, NICU length of stay, and gestational age at caffeine initiation were statistically significant between groups, with the high-dose group being more premature and having a longer duration of stay.

Table 1.

Patient Demographics

In reference to the primary objective of this study, there was no significant difference between low- and high-dose groups in median maximum ALP level, 599.5 units/L (range 203-1490) versus 602.5 (range 217-1752), respectively (p=0.718). It should be noted that 7 patients (23.3%) in the low-dose group and 22 patients (18%) in the high-dose group had serum ALP levels ≥900 units/L. Additionally, there were no significant differences between groups in documentation of radiological diagnosis of osteopenia (10% versus 4.1%, low- versus high-dose group, respectively, p=0.19) or nephrocalcinosis (0% versus 3.3%, low- versus high-dose group, respectively, p=0.58). Additionally, there were no significant differences between low- and high-dose groups in documentation of fractures on radiography (0 versus 4.9%, respectively (p=0.60). Diagnoses of NEC and cholestasis were not statistically significant between groups (p=0.69 and 0.51, respectively).

Caffeine Regimen

Table 2 shows data for the caffeine regimen. Overall, patients in the high-dose group had significantly greater exposure to caffeine, with nearly 3 weeks longer duration of use and approximately two times the cumulative caffeine dose (mg). The median dosages were 6.8 and 10 mg/kg/day for the low- and high-dose groups, respectively.

Table 2.

Caffeine Regimen of Low- and High-Dose Caffeine Groups

Concomitant Risk Factors for Development of Osteopenia

Comparison of groups revealed no statistically significant difference in exposure to other medications associated with development of osteopenia (Table 3). Most patients in both groups received a loop diuretic (ie, furosemide). There were no significant differences between groups in the duration of use of paralytics or phenobarbital as well as cumulative dose of furosemide. Likewise, there were no statistically significant differences in use or cumulative exposure of steroids between groups. Patients received hydrocortisone or dexamethasone or both agents. For the purpose of comparison of steroid exposure between groups, the dexamethasone dose was converted to hydrocortisone by using steroid equivalents (i.e., 20 mg hydrocortisone = 0.75 mg dexamethasone).

Table 3.

Concomitant Risk Factors for Development of Osteopenia

The requirement for PN was not statistically significant between groups, with nearly equivalent duration of use (p=0.46). Further evaluation of the intake of the essential minerals and vitamins for bone development via both parenteral and enteral nutrition revealed statistically significant differences in intake of calcium (mg/kg) at week 10 (p=0.0499) and phosphorus (mg/kg) at week 2 (p=0.0231). Additionally, statistically significant differences in vitamin D (IU) intake occurred between groups from weeks 9 to 12 (p=0.0124, 0.0008, 0.008, and 0.002, respectively) (Table 4).

Table 4.

Comparison of Calcium, Phosphorus, and Vitamin D Intake

Regression Analysis

A linear regression analysis was conducted, with the dependent variable being maximum ALP. We used a stepwise variable selection procedure, gestational age, birth weight, gestational age at caffeine initiation, and duration of PN and noted a significant relationship with increased maximum ALP level. Gestational age and birth-weight were inversely related to increase in ALP. For each additional week in gestation ALP is decreased by 49.4 units/L (p=0.012), and for each kilogram increase in birth weight the ALP is decreased by 400.3 units/L (p=0.0202). The gestational age at which caffeine was initiated and duration of PN were directly related to increase in ALP. With each 1-week increase in gestational age at initiation of caffeine, the ALP increased by 32.5 units/L (p=0.0385). Last, for each additional day of PN, the ALP is increased by 3.2 units/L (p=0.0001). It should be noted that the maximum caffeine dose was not significantly related to increase in ALP.

DISCUSSION

Consumption of caffeine has been identified as a risk factor for low bone mineral density and fractures in adolescent females and postmenopausal women.^11–13 This is thought to be secondary to increased urinary and intestinal calcium excretion with caffeine consumption.¹³ The renal effects of caffeine, including diuresis and loss of sodium and calcium in the urine, have also been reported in the neonate model.^3–5 The association of caffeine with low bone mineral density and increased renal excretion of calcium prompted this retrospective evaluation of the association of caffeine dose with osteopenia, using ALP as a surrogate marker.

Zanardo and colleagues⁵ demonstrated increased urinary calcium excretion in 10 premature neonates following a 5-day course of caffeine. Patients receiving caffeine had a 10-fold increase in calcium excretion compared to that of the control group. The dose-related renal effects of caffeine have only been evaluated in the newborn rabbit model following administration of caffeine sodium benzoate.⁴ The newborn rabbits that received a one-time dose of 10 mg/kg had significant increase in diuresis and naturesis versus no change in the 5 mg/kg group. To date, the dose-related renal effects of caffeine have not been demonstrated in the human neonate. Additionally, it has not been determined if the urinary loss of calcium is a dose-related effect, with greater calcium excretion at higher doses. Theoretically, greater loss of urinary calcium could equate to a greater predisposition to development of osteopenia. The loss of calcium associated with use of caffeine can result in a decrease in serum calcium levels. This in turn will stimulate release of parathyroid hormone, which will ultimately increase osteoclastic activity to release calcium from the bone matrix.

Eighty percent of patients in our study received high-dose caffeine, with a maximum dosage of 15 mg/kg/day. Approximately 42% of patients (51/122) in the high-dose group received a maximum caffeine dose >10 mg/kg/day. These higher doses were used in patients who continued to have apnea, and the dose was titrated upward, with tachycardia as the dose-limiting effect.

Maximum ALP level was not found to be statistically significant between caffeine dose groups. It should be noted that a wide variation in ALP levels was present in both groups. Additionally, the regression analysis also confirmed that there is no significant relationship between maximum caffeine dose (mg/kg) and increase in ALP. Maximum ALP levels ≥900 units/L were noted in 7 patients (23.3%) in the low-dose group and 22 patients (18%) in the high-dose group. It has been proposed that an ALP level of ≥900 units/L, along with a serum phosphorus concentration of <5.5 mg/dL, can be used as a screening method for osteopenia in premature neonates.¹⁴

There were no statistical differences noted between groups in regard to concomitant risk factors for development of osteopenia, other than intake of calcium, phosphorus, and vitamin D. Differences noted between groups were significant for 1 week for phosphorus and calcium and 4 weeks for vitamin D. It was difficult to determine why these differences occurred at those specific time intervals. Differences in vitamin D intake occurred in weeks 9 to 12, in which the high-dose group received more than two times the amount of vitamin D (Table 4). When interpreting these results, it is important to note that only 5 patients (16.7%) in the low-dose group were still receiving caffeine by week 9. The small number of patients in the low-dose group and the unequal distribution between groups may have affected the results. The evaluation of steroid use was complicated by the fact that different agents were used (i.e., hydrocortisone, dexamethasone) within each group. Glucocorticoid activity is responsible for the osteopenic effects on the bone, with dexamethasone having 25-times the glucocorticoid activity of hydrocortisone.¹⁵ In an attempt to compare cumulative exposure between groups, dexamethasone doses were converted to hydrocortisone equivalents. It should be noted that the range for cumulative hydrocortisone exposure in the high dose group is very wide; however, only three patients received a cumulative dose over 40 mg/kg. Approximately 74% (14/19) of the group received cumulative doses <20 mg/kg. Additionally, when included in the stepwise regression variable selection procedure, no significant relationship was noted between cumulative steroid dose and ALP.

Step-wise linear regression identified only four variables with a significant relationship to ALP level. Prematurity and low birth weight had an inverse relationship with ALP level, which has also been noted in previous studies.^7,16,17 Based on the results of the regression for each additional week of gestation the ALP would decrease by 49.4 units/L, assuming all other factors were consistent. Likewise, for each 1-kg increase in birth weight, the ALP would decrease by 400 units/L. Duration of PN had a direct relationship with increase in ALP, with each additional day of PN the ALP level increases by 3.2 units/L. There was also a direct relationship noted with the gestational age (ie, postmenstrual age) at which caffeine was initiated. For each 1-week increase in postmenstrual age, there was a 32.5 units/L increase in ALP. This finding might be explained by the fact that neonates who are more unstable may require longer ventilatory support and not require use of caffeine until a later date. Those infants are also more likely to require PN for a longer duration secondary to their instability.

Due to the retrospective nature, there are inherent limitations with this study. ALP was used as a biochemical marker for the development of osteopenia, for which a positive correlation has been reported in previous studies.^16,17 However, the usefulness of alkaline phosphatase as a biochemical marker of osteopenia has been debated.¹⁸ Radiographic evidence is typically used to confirm diagnosis of osteopenia; however, due to the retrospective design, this was not performed in all patients. In fact, only 8 patients in the study (low-dose, n=3; high-dose, n=5) had a radiographic diagnosis of osteopenia and six patients in the high-dose had a documented fracture. An additional limitation in the study is that there are multiple factors that contribute to the development of osteopenia. An attempt was made to collect information on all known factors in an effort to determine if groups were similar in these regards and only caffeine dose differed.

The unequal distribution of patients between groups and the small number of patients in the low-dose group is an additional limitation identified. With the small number of patients, there is a potential for making a type-II error in reporting no significant difference between groups. Selection of a different cut-off point for classification of low- and high-dose could have resulted in more even distribution between groups and different results. However, a linear regression was employed as a means to identify factors with significant relationships to change in ALP level that may have been misclassified due to the small sample size. The regression analysis did not identify caffeine dose to be significantly related to elevation in ALP level. Last, we classified patients based on their maximum caffeine dose; however, we did not collect how long they received that maximum dose. Therefore, there is a possibility that some patients classified in the high-dose group received that dose for only a short duration.

CONCLUSIONS

This study found no relationship between caffeine dosage and maximum ALP level. Using ALP as a marker for osteopenia, use of higher caffeine doses does not appear to put the infant at greater risk for osteopenia of prematurity. The only factors demonstrating a significant relationship were lower gestational age and birth weight, prolonged PN, and gestational age at caffeine initiation. Future prospective studies are needed to confirm these results. These studies can be designed to control for some of the contributing factors of osteopenia and can use radiographic evaluation for diagnosis.

ABBREVIATIONS

ALP

alkaline phosphatase

ELBW

extremely low birth weight infants

IU

international units

NEC

necrotizing enterocolitis

NICU

Neonatal Intensive Care Unit

PN

parenteral nutrition