Alkaline phosphatase (ALP) is regularly measured in clinical practice. Changes in serum levels are observed in a number of clinical conditions. In neonatology, it has been proposed as a useful marker for both a diagnosis and an indication of the severity of metabolic bone disease (MBD) in infants born preterm. Nutritional practices, aimed at reducing the occurrence or severity of MBD, have led to ALP being proposed as a stand-alone means of monitoring treatment. The current evidence does not support this use: ALP only achieves usefulness in a diagnostic and monitoring capacity when combined with other serum and imaging techniques.
Statistics from Altmetric.com
Serum alkaline phosphatase (ALP) levels are routinely included in liver function testing (LFTs) and when measuring ‘bone biochemistry’. This paper discusses the evidence base for its measurement in neonatal medicine (box 1), and explores whether this can usefully guide intervention.
Box 1 Search strategy and selection criteria
We searched for studies from 1980-2011 using Medline and combining the search terms ‘alkaline phosphatase’ AND (preterm OR premature OR low birthweight infant). We identified studies with measures of ALP and outcome measures published in English. There were no studies where treatment randomisation was made on the basis of ALP alone. Table 1 details key studies that included measurement of ALP and comparative measure(s) of bone density or growth, specifically relating to preterm or low birth-weight infants. We focused on the role that measurement of ALP might play in decision making in the routine nutritional management of MBD in neonatal units.
ALP was first identified in the 1920s1 from numerous body tissues. The highest activity was found in those from bone and ossifying cartilages of young animals suggesting a role in ossification. The osteoblast was identified as the likely source of the enzyme, with optimal in vitro function occurring at an alkaline pH (8.4–9.4). Alkaline phosphatase (ortho-phosphoric-monoester phospho-hydrolase) removes phosphate-containing groups from many different molecules.2 Human ALP also has phospho-transferase3 and protein-phosphatase activity4 in vitro.
In humans, three distinct ALPs (coded on chromosome 2) are produced in a ‘tissue-specific’ manner: placental, placental-like (or ‘germ-cell’) and intestinal. A fourth ALP (coded on chromosome 1) is also produced throughout the body: tissue ‘non-specific’ alkaline phosphatase (TNSALP).5 These four ‘major’ ALP types are ‘true’ isoenzymes. Differential post-translational modification of TNSALP (in bone, liver and kidney)6 produces further isoforms. For simplicity, we will refer to all ALP forms as iso-enzymes.
Figure 1 shows the role of ALP in providing phosphate to the process by which a nidus is formed to promote bone mineralisation at the growing front. During high bone turnover or deficiency of vitamin D/mineral substrate, increased serum concentrations of bone-ALP reflect increased turnover and rupture of matrix vesicles at the bone front as the body attempts to promote bone mineralisation.
Liver-ALP is expressed on the cannalicular membrane of hepatocytes and in the endothelial cells around the central and portal veins. A rise in liver-ALP is usually attributed to cholestasis, most likely from increased hepatic production and subsequent release into the bloodstream rather than ‘overflow’ due to obstruction of the bile ducts.8 ,9 A rise in serum ALP is then usually a delayed finding in the acute phase of biliary obstruction. There is no strong evidence for a physiological role of ALP in the human liver.10
Technological background: how is ALP measured?
The International Federation for Clinical Chemistry (IFCC) recommended method quantifies ALP by measuring enzymatic action on chromogenic (‘self-indicating’) solutions using bichromatic spectrophotometry.11 The rate of production of the visible products of ALP-mediated test-compound hydrolysis at 37°C is proportional to the enzymatic activity (hence quantity) of ALP. ALP is reported in U/l, with an assay-dependent reference range. This range varies depending on age and sex: it is generally about 150–300 U/l in neonates but as wide as 60–450 U/l in pubertal males. A fasting blood sample in a plain, clotted-serum or lithium-heparin tube should be used. Anticoagulant substances complex Mg2+ and Zn2+ both required for in vitro ALP activity, and should be avoided. Marked haemolysis, hyper-bilirubinaemia or hyper-lipidemia can also interfere with assay accuracy. ALP isoenzymes can be separated using gel electrophoresis and other techniques.12 Essentially qualitative, these tests do not give a finite, isoenzyme quantification. Isoenzyme analysis is technically difficult as many isoforms show similar gel-migration characteristics.
Metabolic bone disease of prematurity
In utero accretion of calcium and phosphate in the fetus far outstrips that which can be supplied by ex utero parenteral nutrition or unfortified milk feeding.13 Metabolic bone disease of prematurity, MBD (‘osteopenia of prematurity’ or ‘rickets of prematurity’) is therefore almost universal, especially in extremely low birth weight (ELBW) infants. Figure 2 shows the characteristic X-ray appearances of osteopenic bones, splaying and cupping of long-bone ends and a humeral fracture.High metabolic demands for calcium and phosphorus occur over the first few postnatal days, resulting in obligatory bone-reabsorption. Despite this there is also increased ALP activity attempting to promote bone mineralisation in the preterm skeleton (figure 1). A rise in ALP is therefore seen in most preterm infants over the first 3 weeks, achieving a plateau by 5–6 weeks.13 Some infants reach ALP levels 5–10 times the upper limit of the adult normal range. This may take 2–3 months to peak14 and is often associated with prolonged parenteral nutrition.
In preterm neonates, does serum ALP predict current mineral status?
Studies in preterm infants have shown associations between high ALP and both low plasma phosphorus and high calcium levels.15 ,16 Both are used as serum markers of ‘current mineral status’, with phosphorus also being important for lean mass accretion.15 ,17 ALP has also been shown retrospectively to have a linear association with serum phosphorus levels.15 A study in preterm infants ≤28 weeks gestation showed there is marked urinary conservation of phosphate, in the face of hypophosphataemia, when fed unsupplemented breast milk compared to a preterm formula.18 The infants fed preterm formula had greater absolute losses of phosphate, but an overall greater net absorption and higher phosphate levels due to higher intakes. In the unsupplemented breastfed babies, urinary calcium losses were higher due to phosphate deficiency and a subsequent inability to utilise absorbed calcium. The effect that variation in renal handling of calcium and phosphorus has on serum levels of calcium and phosphorus reduces their reliability as markers of current mineral status in reflecting underlying bone stores. Similar issues apply to the use of ALP as a marker of current mineral status.
If the ‘seeding’ role of ALP in bone mineralisation is considered (figure 1), it is plausible that ALP levels will normalise in the face of adequate mineral substrate due to satisfactory mineralisation rate at the growing front. However, the bone may remain osteopenic for a longer period of time while it re-mineralises. Thus, even where overall mineral status remains poor, serum phosphorus, calcium and ALP may well be normal. To assess overall mineral status, bone imaging is preferable to serum markers alone. The efficacy of ALP in reflecting directly measured bone mineral status is examined below.
In preterm infants, does ALP alone predict the need to start treatment for MBD?
Measurement of bone mineral content (BMC) and, more informatively, bone mineral density (BMD, which takes bone volume into account) is most accurately assessed by dual-energy absorptiometry (DEXA) scanning. This gives an indication of bone structure and could be used to diagnose and monitor metabolic bone disease. However, it requires exposure to ionising radiation (albeit minimal), and is not routinely available. Though it is theoretically the ‘gold standard’ for detecting and monitoring MBD, it is currently only a research tool in neonatology. Changes on plain radiographs have also been used to diagnose MBD, but cannot quantify the mineralisation or density of the bone and are not reliably temporally related to the underlying disease progression.
ALP has been proposed as a proxy measure which might be used to define a threshold predicting the need to start treatment for MBD. Studies have shown a correlation between ALP measurements in preterm infants of ≥1200 U/l and reduced stature at both 9 and 18 months, and this level has therefore been proposed as a threshold beyond which treatment might be started for MBD.15 However, this study included no measurement of bone-density in relation to the ALP levels in the neonatal period. Faerk et al16 also failed to find any association between ALP (mean, peak and time to peak) and BMC as measured by DEXA in preterm infants. They showed that there was no significant difference in BMC between infants with the highest (>1200 U/l) and lowest (<600 U/l) measurements suggesting ALP was not useful as a guide to starting treatment. Backström et al19 demonstrated that stand-alone ALP measurements >900 U/l only had 88% sensitivity and 71% specificity for predicting apparent BMD <95 mg/cm3 (their cut-off for metabolic bone disease) using DEXA in a cohort of ex-preterm infants at 3 months corrected gestational age. By comparison in the same study, serum inorganic phosphate levels <1.8 mmol/l had a 96% specificity for predicting the degree of bone mineralisation. Hung et al20 prospectively measured ALP (total) and Bone ALP in preterm (≤34 weeks) fortnightly from 3 weeks of age until term equivalent. Two groups were identified from the cohort based on the presence of osteopenia on radiographs at term equivalent age. They found an ALP cut-off of >700 IU/l at 3 weeks of age gave only 73% sensitivity and 74% specificity for predicting the presence of osteopenia (Bone ALP gave 71% and 80% respectively).
Plain film appearances diagnostic of MBD have a lag time of 2–6 weeks from occurrence of peak ALP levels,14 which renders both a threshold level of ALP and serial use of plain films unhelpful in prospectively identifying MBD. In one series21 only a single infant who previously passed the ‘high’ ALP cut-off had radiological evidence of bone disease with formal scoring of the films. Direct comparison of radiographic appearances of MBD and measured bone mineralisation show very weak correlation,22 with between 20 and 40% demineralisation needing to occur before radiological changes of MBD become apparent.23 ,24 ALP cannot, therefore, determine when to start treatment for MBD if used as a stand-alone measure: no fixed threshold appears clinically useful.
A pragmatic approach to starting treatment for MBD has been proposed25 based on a survey of UK neonatal unit policies for treatment of MBD in preterm infants. The authors suggest that in infants <1500 g, or <28 weeks gestation, or in those who have had parenteral nutrition for ≥4 weeks, ALP should be measured. If ALP is more than 500 U/l, then weekly urinary phosphate, ALP, calcium and serum inorganic phosphorus should be measured. Substrate deficiencies (especially phosphate <1.8 mmol/l, but also calcium, Vitamin D etc) should then be corrected. Backström et al showed that a combination of serum ALP >900 U/l and serum inorganic phosphate <1.8 mmol/l (in ex-preterm infants, at a corrected age of 3 months) was 100% sensitive and 71% specific for reduced BMC19 measured by DEXA. While Hung et al20 did not find measurement of phosphate increased their prognostic accuracy it is a striking feature of their work that in their osteopenic group, even at only 3 weeks old, the serum phosphate was significantly lower than in the group who did not develop MBD. Their infants also received calcium and phosphorus supplementation as standard.
In neonates who have MBD, does measurement of ALP guide treatment with supplements to improve mineralisation and growth?
In MBD, there are two main therapies: supplementation with vitamin D (to optimise plasma 1,25-dihydroxyvitamin D (1,25(OH)2D) levels) and dietary Ca/PO4 supplementation to optimise mineralisation substrate availability. The outcomes of interest are improved mineralisation of bone (ie, prevention or resolution of MBD) and subsequent improvement of linear growth.
Calcium and phosphate supplementation
Although the association of low phosphorus levels and high ALP appears linear,15 ,20 it is not possible to recommend supplementing dietary phosphate beyond any specific value for ALP. The variation seen between babies in the degree of urinary conservation of phosphate18 means that ‘flat rate’ supplementation with phosphate beyond a fixed ALP threshold cannot be assumed to guarantee an adequate concentration of phosphorus at the bone growing front in an individual baby. In addition, variation in ALP levels once adequate mineralisation has started (a drop in ALP would be expected) might lead to inadequate duration of supplementation with phosphorus. There are no data to show that changes in ALP are consistently proportional to serum phosphate levels across a population in order to make it useful in monitoring mineral supplementation.
The role of phosphorus in preserving linear growth,15 ,17 however, is such that it would be prudent to check serum phosphate whenever ALP is found to be raised and supplements given, where there is high risk of MBD. Direct measurement of serum phosphate will help to maintain normal serum concentrations and optimise mineralisation and linear growth regardless of the ALP level. No consistent association between serum calcium and ALP has been demonstrated. However, measurement of serum calcium whenever serum phosphate is measured will aid decision making, as inappropriate management may result in either a high or low calcium level.
There is no evidence to support any specific ALP level as a threshold for stopping treatment primarily because of the lack of reliable temporal associations between ALP peak/trough level and MBD measured directly or by imaging.14 ,16 If treatment is started with supplemental phosphate or calcium, repeat measures along with linear growth assessment may help determine efficacy, progress and when to stop, but quantitative measurement of bone mineralisation is rarely possible in routine practice.25
Vitamin D supplementation
Vitamin D is essential to optimise calcium uptake but the association between ALP and circulating vitamin D levels in preterm infants is not well established. Postnatal vitamin D supplementation (to the infant, or high dose to breastfeeding mothers) increases circulating 1,25(OH)2D levels, but does not affect ALP levels.26 In ELBW infants with adequate endogenous 1,25(OH)2D, detection of high levels of ALP showed no correlation with 1,25(OH)2D levels.27 Higher dose vitamin D supplementation produces higher serum 1,25(OH)2D levels but appears to have no effect on ALP levels.28
Independent serum vitamin D levels do not correlate well with radiographic appearances of MBD,27 directly measured BMC28 or phosphate levels.29 When differential levels of supplementation were trialled, there was a trend towards higher BMC only when there was a higher level of serum phosphate and calcium.28 Spontaneous resolution of MBD has also been seen without vitamin D supplementation in preterm infants who had adequate circulating 1,25(OH)2D levels at the time of radiographic diagnosis of MBD.27 ALP levels are, therefore, of no clear use as a parameter by which to judge or monitor vitamin D supplementation in MBD.
Do ALP measurements in infants receiving neonatal intensive or special care predict MBD at discharge or in the immediate postdischarge period?
None of the studies examined looked for MBD at the time of discharge, more commonly referring to a specific postmenstrual age. Preterm infants have higher ALP levels14 and lower BMC at term corrected age than term-born infants.14 ,22 Only Faerk et al16 specifically looked at BMC in ex-preterm infants at term corrected gestational age. They found no association between ALP levels and were unable to predict bone mineralisation based on longitudinal ALP measurements. Backström et al28 gave high- or low-dose vitamin D supplements, for 3 months after delivery, to infants born at <33 weeks gestation. By 3 months corrected gestational age, there was no difference in BMC between the high- and low-supplement groups and no relationship to the measured peak ALP. No measurement was performed at term corrected gestation. ALP levels in preterm infants on neonatal units measured at discharge cannot reliably predict the presence, or assess the severity, of MBD.
Do ALP levels in premature infants predict the presence of MBD or growth restriction in later childhood?
The discrepancy between BMC in ex-preterm infants and term infants does not disappear until the age of at least a year of age.30 Changes in BMC between 3 and 6 months of age related to mineral supplementation, were no longer apparent when the children were assessed at 9–11 years old.31 The authors felt that this reflected that complete remineralisation required an extended period of breastfeeding independent of early mineral supplementation. ALP did not correlate with subsequent linear growth in this cohort. In another cohort15, high ALP was shown to be related to linear growth restriction at 18 months, but not specifically to MBD. At 5 years of age, a subgroup of this cohort was examined and BMC was found to be mostly related to increased feeding with breast milk.32 ALP measurements during the initial hospital stay do not predict MBD later in childhood.
Quiz: True or false?
ALP is linearly associated with serum calcium and phosphate levels in preterm infants
ALP is a more specific predictor of BMC than serum phosphate levels
Measurement of serum phosphate is helpful in optimising mineralisation especially in infants receiving breast milk
ALP levels are a poor predictor of later BMD or linear growth
See the end of this paper for the answers.
Is ALP isoenzyme analysis useful in preterm infants with raised serum ALP levels?
ALP interpretation requires age- and sex-specific reference ranges. Differentiating between bone or hepatic causes is usually straightforward by additional serum measures and imaging. ALP is a sensitive marker, where cholestasis occurs because bile acids cause membrane lysis by detergent effect and release membrane-bound ALP into serum. A combination of γ-glutamyltransferase (GGT), ALP, bilirubin and transaminases is the most specific way to distinguish cholestasis with hepatocyte damage from other causes of raised ALP such as MBD. Measurement of bone-specific isoenzymes does not appear to be cost-effective . When compared prospectively in preterm infants, bone-specific ALP has an equivalent sensitivity (73%) and only marginally better specificity (80%)20 than total serum ALP in predicting MBD. Bone specific ALP offers no clear advantage over the (cheaper) combined measure of serum ALP and phosphate in combination.19
If the raised ALP is not obviously bone or liver related, then other differential diagnoses (box 2) should be considered. Most of these conditions, however, have clear diagnostic presentations which prompt more relevant clinical tests than serum ALP measurement. ALP isoenzyme analysis is generally not useful in differentiating them.
Box 2 Additional differential diagnoses in presence of raised ALP
▶ Benign transient hyperphosphatasemia of infancy and childhood: is most common diagnosis if only ALP raised; usually occurs during/after intercurrent illness, often viral; resolves spontaneously
▶ Benign familial hyperphosphatasaemia: is rare; one form associated with mental retardation
▶ Gut perforation and necrosis
▶ Endocrine disturbance for example, acromegaly, hyperthyroidism
▶ Bacterial infection
▶ Post-operative recovery
ALP, alkaline phosphatase.
Can a low level of alkaline phosphatase activity be ignored?
Low ALP levels attract little attention. Congenital hypophosphatasia (HPP) is a rare but important cause of low ALP. Characterised by age of onset (perinatal, infantile, childhood and adult), HPP is generally less severe the older the age of presentation. Perinatal HPP is usually fatal. Infantile HPP causes death in around 50% of sufferers by respiratory compromise. Diagnosis of HPP is made on radiographic appearance. Serum calcium and phosphorus are usually normal or raised. Inheritance of TNSALP defects is autosomal recessive in the most severe forms (perinatal and infantile) and either autosomal dominant or recessive in milder forms. There is no established medical treatment. Magnesium and Zinc are both important cofactors for ALP function. In the neonate, therefore, wider consideration of on-going nutrition and measurement of serum Mg2+/Zn2+ should be considered in the face of a persistent, inappropriately low ALP.
Answers to the quiz
ALP is a widely available and used serum marker. However, its in vivo role beyond bone metabolism has not been well characterised and the complexity of the process of metabolic bone disease means it is a test of limited usefulness as a stand-alone measurement. It is, however, potentially useful in testing as part of a panel of tests. Future research to clarify the in vivo role of ALP outside bone mineralisation would allow more informed interpretation of ALP measurements. It would also be useful for broader quantitative work to be done in assessment with multiple investigative modalities in the neonatal unit, to try to improve identification of preterm infants at risk of metabolic bone disease and their treatment. Metabolic bone disease remains a common problem. Evidence linking bone mineral accretion with later life diseases such as osteoporosis suggests nutritional management of preterm infants before and after hospital discharge remains a research priority.
Clinical ‘bottom line’
■ The in vivo effects of ALP remain under investigation and reflect the current evidence that it is not a useful indicator of disease when taken as a stand-alone marker.
■ Independent measures of ALP are of little use in predicting and determining risk, or treatment thresholds, for MBD in preterm infants.
■ ALP is not useful in monitoring on-going treatment for metabolic bone disease.
■ High ALP levels in preterm infants are strongly associated with phosphorus deficiency and later growth, but do not predict outcome.
■ ALP isoenzyme analysis is rarely helpful in determining causes of raised ALP.
Competing interests None.
Provenance and peer review Commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.