Archives of Disease in Childhood, 1989, 64, 1403-1409

Regular review

Bone disease in preterm infantsN BISHOP

Dunn Nutritional Laboratory, Cambridge

New work in bone physiology and cell biologyduring the last decade has made it possible toconstruct a model for the bone disease of preterminfants variously labelled 'rickets', 'osteopenia', and'metabolic bone disease of prematurity'. The modelproposed here explains its pathogenesis and itsoutcomes, and suggests a sequence of appropriateinvestigations as well as a scheme of management.The figure illustrates the basic processes of bonemineral metabolism in the perinatal period, andprovides a framework against which the derange-ment in mineral homoeostasis leading to bonedisease can be more clearly visualised.

Fetal bone mineral homoeostasis

Mineral accretion rates for both calcium andphosphorus increase throughout pregnancy, reachinga maxiumum during the third trimester of 3-0-3*7mmol/kg/day for calcium and 2-4-2-7 mmol/kg/dayfor phosphorus.'

Fetal plasma calcitonin concentrations are alsoraised in utero2; though this peptide hormone isknown principally for its hypocalcaemic action,there is substantial evidence that it has appreciableanabolic, mineral accreting effects on bone.

In contrast, the principal hormones mediatingbone resorption in later life, parathormone, and 1,25 dihydroxycholecalciferol, are found in low con-centrations in the fetus.2 3 Interestingly, however,prolonged low dose administration of these hor-mones in animals results in increased bone mass andthus both may be actively concerned in the accretionrather than resorption of mineral in utero.4

The rwst 48 hours

After ligation of the cord, the supply of calcium,phosphorus, and all other nutrients ceases abruptly.The continuing demand by bone for calcium en-trains a rapid fall in blood calcium concentrations;the nadir for ionised calcium is usually at about 18hours of age, slightly before that for total calcium.5

In well preterm infants, hypocalcaemia will usuallybegin to improve by 24-30 hours of age, with valuesin the adult normal range attained by 48-60 hours.Factors unrelated to bone metabolism, particularlytissue hypoxia with subsequent calcium influx, maycontribute to the low calcium concentrations seen inthe sickest infants, in whom hypocalcaemia is morelikely to persist and be more pronounced.

Previous workers have suggested, however, thatthe plasma parathormone concentration does notrise after birth, and that parathormone 'resistance' islikely to occur in preterm infants. Much of the earlywork on parathormone in the perinatal period wascarried out using radioimmunoassays for either thecarbon or the nitrogen terminal moiety of themolecule. Parathormone is an 84 amino acid peptidethat is rapidly and continuously synthesised andalmost immediately subjected to intracellulardegradation.6 Inactive fragments and intact mole-cules are stored together and then released,principally in response to hypocalcaemia. Anincrease in the amount of active hormone secretedcould remain undetected, as the total terminalspecific assay activity might not change appreciably.More recent studies, however, carried out withintact-molecule and active fragment (residues 1-34)assays, have shown a two fold to five fold increase inactive parathormone secretion during the first 48hours after birth.5 The principal target organs forthe parathormone molecules thus released are boneand kidney. In response to parathormone the kidneyreabsorbs calcium and actively excretes phosphorus.A reasonable indication of the response to para-thormone would therefore be to monitor urinaryoutput of phosphorus over the first days after birth.The longitudinal changes in whole blood ionised

calcium up to the age of 72 hours, and in urinaryphosphorus excretion up to the age of 5 days werestudied in 18 preterm infants. High urinary concen-trations of phosphorus on days 2 and 3 after birthwere observed; subsequently, the phosphorus lossdiminished rapidly and by day 5 phosphorus ex-cretion had ceased in most of the infants studied

1403

1404 Bishop

In the uterus

First 48 hours

Bone effects of parat

(i) High blood calcium, phosphate, and calcitonin concentrations with high bone mineral accretion rates(ii) Low circulating parathormone, and 1,25 dihydroxycholecalciferol concentrations

(i) Transplacental calcium infusion ceases, bone accretion continues, and blood calcium concentration falls(ii) Plasma calcitonin surge exacerbates fall in blood calcium concentration(iii) Hypocalcaemia induces parathormone rise in plasma with effects on bone and kidney

Renal effects of parathormone

Increasedproduction of

1,25 dihydroxycholecalciferol Adaptive reabsorption ofcalcium with increased phosphate

excretion: body phosphorusstores fall

Bone effects ofhormone 1,25 dihydroxycholecalciferol

Bone resorption

Matrix degradation Release ofproducts-some act Calcium, Phosphate

as local humoral factors

Nutrient supply

Production of bone matrix Bone mi

New bone formation:remodelling, growth,and mineralisation

Potential for further lossof phosphorus in urine

ineral accretion +Influence of favourable

local factors,matrix vesicles

Figure Outline of processes of bone mineral metabolism in the perinatal period.

(unpublished observations). These observationsparallel results from animal studies of the effects ofexogenously administered parathormone on phos-phate metabolism in states of phosphorus repletionand depletion.4Another important consequence of the action of

parathormone on the kidney is the enhancement of1,25 dihydrocholecalciferol synthesis. 1,25 di-hydroxycholecalciferol is the most active metaboliteof vitamin D3 affecting not only the gastrointestinal

absorption of calcium and phosphorus, but also themobilisation of calcium from bone. 1,25 dihydroxy-cholecalciferol has a central role in the maintenanceof calcium homoeostasis, which is discussed in detailbelow.The release of parathormone is probably poten-

tiated by the apparently paradoxical release ofcalcitonin almost immediately after birth.7 In post-natal life calcitonin is secreted postprandially inresponse to gastrin production, and the initial surge

I

of gastrin after the first feed may be responsible forthe increase in plasma calcitonin concentrationsseen at this time.' Calcitonin inhibits the resorptiveresponse of osteoclasts and so delays the supply ofcalcium from bone to the circulation. The postnatalregulation of calcium homoeostasis is principallyachieved by the interlocking actions of para-thormone and 1,25 dihydroxycholecalciferol onbone, and by their separate effects on absorptionand retention of mineral substrate.

Bone resorption

The isolation and culture of pure cell lines ofosteoblasts and osteoclasts, have enabled rapidadvances in our understanding of the underlyingprocesses in bone.9`11 It is now clear that osteoblastsand osteoclasts act. together to undertake boneresorption, and that osteoclasts, having no para-thormone receptors, exert resorptive activity inresponse to signals from osteoblasts. The specificeffects of parathormone on bone are4: increasedosteoblast permeability to calcium; release ofcollagenase from osteoblasts; and release of osteo-clast activating factor(s) from osteoblasts, as a resultof which osteoclasts increase in number and activity.

In addition, 1,25 dihydroxycholecalciferol,produced in response to the increased concentra-tions of parathormone, exerts synergistic effects onbone.9 1 These are: activation of an osteoblasticcalcium pump; increased activation and fusion of themonocyte/macrophage precursors of osteoclasts;production by osteoblasts of osteocalcin (Glaprotein), which is chemotactic for osteoclasts;effects on immune cell function, particularlylymphocytes, with reduced interleukin 2, and in-creased interleukin 1 production, which enhancesosteoclast formation and activity; and possibly areduced response of osteoblasts to parathormone(reduced cyclic adenosine monophosphate response).Thus parathormone and 1,25 dihydroxychole-

calciferol have complementary actions; the increasein osteoblast permeability to calcium flux with theactivation of a specific calcium pump provides anacute response to falling ionised calcium concen-

trations. The recruitment, activation, and fusion ofosteoclast precursors, and their subsequent activityin response to osteoblast derived humoral factors,provides a longer term source of calcium.

New bone formation

The result of the resorptive process is to producecalcium, phosphorus, and breakdown products ofbone matrix. These breakdown products are

thought to act locally to promote new bone forma-

Bone disease in preterm infants 1405

tion by osteoblasts. The increase in resorptiveactivity initiated by the surge of parathormone afterbirth is therefore matched by a concurrent increasein new bone formation. Though matrix volume isnot reduced during this period of intense activity,net loss of bone mineral will occur if the exogenoussupply of mineral substrate is inadequate.

In addition to the supply of adequate mineralsubstrate to normally functioning osteoblasts, afavourable local environment for bone mineralisa-tion is also crucial to the remodelling and growth ofbone; many factors have been identified in labora-tory studies as having a role. In particular, there is agrowing body of evidence to support the part playedby matrix vesicles in the initiation and propagationof crystallisation. 2

Matrix vesicles are discrete sacs that are derivedfrom the osteoblast cell membrane. They arecomposed of a phospholipid bilayer rich in phos-phatase enzymes including alkaline phosphatase,and they accumulate at the growing front of bone.At the pH of the mineralisation front, alkalinephosphatase functions principally as a phospho-transferase, transporting phosphate residues thathave been cleaved by other phosphatase enzymesinto the vesicle's sap.Calcium enters the vesicle by diffusion, and is

trapped by phosphatidyl serine. The additionalaccumulation of phosphate raises the saturation ofthe vesicle sap to the point where the calcium/phosphate solubility product is exceeded andcrystallisation begins. Electron microscopic pictureshave shown the growth of crystals on the innerleaflet of the vesicle that leads to its subsequentdisruption as the ends of the crystal pierce thebilayered membrane.These crystals then seed into the fluid at the

mineralisation front and, given adequate mineralsubstrate there, act as foci for further crystallisation.The rate of turnover of matrix vesicles with therelease of their membrane constituents, therefore,reflects the rate of initiation of crystallisation.Laboratory studies have shown that there aregreatly increased numbers of matrix vesicles inrachitic growth plates'3; this lends support to theconcept that increased alkaline phosphatase activityin plasma may represent increased vesicle turnoverin substrate or vitamin D deficient states.During the early neonatal period the main deter-

minants of bone remodelling, mineralisation, andgrowth are those that have been discussed in detailabove. There are, however, many other factorsaffecting the fine control of bone homoeostasis,9 11

of which two are of particular relevance to thepreterm infant.Aluminium is a potent inhibitor of bone minerali-

1406 Bishop

sation, and is present as a contaminant in parenteralnutrition solutions. 13 Up to 80% of the intravenouslyadministered load may be retained, and significantdeposition was found in bone after three weeks ofparenteral feeding. It is possible that aluminiummay exacerbate bone disease in preterm infants fedintravenously.

Immobility also causes loss of bone mass. Stressgenerated electrical potentials have been implicatedin the osteogenic process, and prolonged periods ofsedation or paralysis during mechanical ventilationincrease the possibility of the loss of bone mass.

Mineral substrate insufficiency

Given an adequate nutrient supply, remodelling,mineralisation, and growth of bone should proceednormally in most infants. For bone disease todevelop, depletion of mineral substrate must occur.Phosphorus depletion is likely to develop morerapidly as it may initially be lost in the urine, andprotoplasmic metabolic requirements for phos-phorus are greater than for calcium (extrapolatingfrom data on fetal accretion rates and body com-position studies, 0-6-0-7 mmol/kg/day comparedwith 0-2-03 mmol/kg/day).

Inadequate dietary provision of phosphorus-forinstance, the exclusive use of unsupplementedhuman milk-compounded by the initial urinaryphosphorus losses will result in low tissue phos-phorus stores, and low circulating concentrations ofphosphorus.The reduced delivery of phosphorus to the kidney

prevents further appreciable urinary losses andenhances renal production of 1,25 dihydroxy-cholecalciferol. The increase in circulating 1,25dihydroxycholecalciferol in turn increases gastro-intestinal absorption of both calcium and phos-phorus. In addition, the release of parathormone isinhibited, further reducing the risk of phosphorusloss in the urine. As a corollary, however, renalreabsorption of calcium is reduced, with consequenthypercalcuria. The inhibition of parathormonerelease may also slow the process of bone re-absorption; nevertheless, the potent bone resorbingactivity of 1,25 dihydroxycholecalciferol willcontinue to remove some calcium and phosphorusfrom bone.

In addition to the phosphorus absorbing andretaining processes detailed above, it is possible thathypophosphataemia is a key factor in acceleratingdirectly or indirectly the turnover of matrix vesiclesand hence increasing plasma alkaline phosphataseactivity.

If mineral substrate provision continues to beinadequate, further substrate will be lost from bone

in order to supply the needs of other tissues. Thebiochemical outcomes of these processes are inter-linked; reduced concentrations of phosphate inurine and plasma precede the increasing urinary lossof calcium, and in extreme cases, hypercalcaemia.Raised plasma alkaline phosphatase activity isseen principally after 6 weeks of age.

Radiological and anthropometric changes occurslowly, being seldom evident before 6 weeks ofage. l 15 In the long term the principle outcomes forbone are linear growth, mineral content, andstructural integrity. In a large study of preterminfants receiving different diets during the neonatalperiod, we found a significant association betweenthe increase in plasma alkaline phosphatase activityand a reduction in height achieved at both 9 and 18months implying that bone disease, reflected byincreased remodelling activity during the first weeksof life, had a lasting effect on the infants' growthpotential up to the age of 18 months.16

If these differences persist, then it is likely that thenutritional deprivation sustained by bone during thisapparently critical phase of development has 'pro-grammed' the bone to grow less slowly, as catch upgrowth would otherwise be observed when dietarysufficiency was achieved.'6The regulatory mechanisms for this adaptive

change remain to be elucidated, but could involvechanges in cell number, type, or function, eitherlocally or systemically.

Investigation of early bone disease

Plasma phosphate concentrations fall gradually from2 mmol/l to 10-O1-5 mmol/l over the first week, andoften reach a nadir during the second week afterbirth (unpublished observations). In infantsdepleted of phosphorus as a result of urinary lossesand poor intake a further reduction to <1 mmol/lmay occur, and this has been reported to beassociated with the later development of bio-chemical and radiological evidence of bone disease.Urinary phosphate excretion initially may be in-creased but by day 5 is usually negligible. Bycontrast, urinary calcium losses increase and persistduring the period that tissue phosphorus storesremain depleted. A prolonged absence of phosphatefrom the urine with persisting calciuria would implycontinued tissue phosphorus depletion, and mightbe a useful marker to follow sequentially in anindividual infant.The natural history of plasma alkaline phos-

phatase activity is to rise over the first 3 weeks andplateau until the age of 5-6 weeks. Rises that occurafter this are seen principally in infants withpersistently low plasma phosphorus concentrations

receiving low phosphate diets-for instance, un-

supplemented human milk. Increased plasmaalkaline phosphatase activity is widely quoted asbeing indicative of bone disease; difficulties arise inthe interpretation of results and comparison withother centres because of the use of different assay

systems with widely varying ranges and differentunits of measurement. Peak concentrations ofgreater than 7-5 times the maximum adult normalvalue for that particular alkaline phosphate assay

have been associated with reduced linear growthvelocity in the short term.14 In the work previouslyreferred to we found an area of demarcation at fivetimes the maximum normal adult value for plasmaalkaline phosphatase activity, with appreciablereductions in growth potential for infants with peakconcentrations exceeding this limit.16

Radiological changes are usually not seen untilthe age of 6 weeks; reduced bone density, andabnormal bone remodelling in the form of cuppin ,

splaying, and fraying of epiphyses may occur, -

and-in extreme cases-there may be fractures ofboth ribs and long bones. The interpretation ofradiographs is, however, subjective and the use ofscoring systems has not improved their predictivevalue for minor to moderate degrees of deminerali-sation.

Photonabsorptiometry is a quick and accuratemethod of assessing sequentially the changes inbone mineral content at a specific site, usually thedistal radius. 17 Photonabsorptiometry has shownthat in infants receiving diets containing littlemineral substrate, bone mineral content may remainunchanged, or even decrease initially, and thenincrease at a rate much less that attained in theuterus. By contrast, infants supplied with mineral inamounts approaching the intrauterine rate canmaintain the fetal rate of mineral accretion.18 Theuse of photonabsorptiometry is restricted to a fewselected centres, however, and its principle use atpresent is for research rather than as an aid todiagnosis.

Radiographic densitometry is a low dose wholebody technique that provides accurate informationabout the overall mineral state of the skeleton. Asyet it has not been applied to preterm infants, but itcould provide valuable data for body compositionand mineral metabolism studies.

It has been often noted that peak alkalinephosphatase activity rarely occurs at the sametime as radiological evidence of abnormal boneremodelling, or the degree of bone demineralisationas measured by photonabsorptiometry. This isessentially a reflection of the intrinsic properties ofeach investigation-plasma alkaline phosphataseactivity is a measure of bone activity, possibly of the

Bone disease in preterm infants 1407

rate of mineral crystallisation; photonabsorptio-metry gives an estimation of the amount of mineralactually in bone; and radiographs best show theabnormal remodelling re'sulting from an inadequateprovision of mineral substrate for bones that arecontinuing to increase their matrix volume.

Short term anthropometry is useful as a non-specific adjunct to the radiological and biochemicalinvestigations in that a reduced linear growth velo-city at the age of 6 weeks would provide furtherevidence to support the diagnosis of bone disease. 16For practical purposes, sequential analysis of

urinary calcium and phosphorus losses is likely toprovide the earliest evidence of incipient metabolicbone disease. If by the age of 3 weeks calciumexcretion is continuing, with no phosphorus appear-ing in the urine despite adopting the prophylacticmeasures outlined below, further mineral supple-mentation should be instituted.

Management

The management of this condition should essentiallyfollow the dictum 'prevention is better than cure'.The degree and duration of mineral, and in parti-cular phosphorus, depletion that will result in bonedisease and the amount of supplementation that willprevent it are unknown. It is nevertheless possible tolook at the provision of substrate by current feedingpractices, formulate estimates of comparative bonemineral accretion rates, and so assess the minimum'preventative' amounts of substrate intake required.Unsupplemented human milk contains 0-5 mmol/

100 ml of phosphorus. For infants receiving 200ml/kg/day, and assuming 90-95% retention, 0-9-0-95 mmol/kg/day of phosphorus will be delivered.After allowing for basal protoplasmic requirements,approximately 0-3 mmol of phosphorus will beavailable for deposition in bone mineral. Calciumand phosphorus accrete at a ratio of 5:3 in bone; upto 0-5 mmol/kg of calcium might therefore bedeposited-about 15% of the intrauterine accretionrate.By contrast, a preterm formula containing 1

mmol/100 ml of phosphorus supplied at 180 ml/kg/day should result in a phosphorus retention of1-61-7 mmollkg/day. After allowing for proto-plasmic requirements, about 1 mmol/kg/day ofphosphorus is available for bone mineralisation,complexing with 1-6 mmol/kg/day of calcium-approximately 50% of the intrauterine accretionrate. Formulas with higher calcium and phosphoruscontents are available in some countries, and havebeen used widely without adverse effects. Reportedmineral accretion rates for infants fed these milksapproach those achieved in the uterus,'8 but the

1408 Bishop

reports are usually of well infants, fed fully by theenteral route by the age of 1 week. In addition,concern has been expressed generally that not all ofthe calcium and phosphate in these milks is availablefor absorption, possibly as a result of precipitationbefore feeding.Given that protoplasmic requirements may be

increased because of pre-existing tissue phosphorusdepletion, particularly in infants who have pre-viously been intravenously fed, the provision of 1mmol/100 ml of phosphorus in milk given to preterminfants should be regarded as an absolute minimum.The addition of phosphorus to expressed human

milk is already common. Buffered neutral phos-phate can safely be admixed with human milk over a24 hour period to raise the phosphorus content from0-5 mmolV100 ml to 1-0 mmol/100 ml.

Larger quantities of both calcium and phosphoruscan be added to human milk. It is important to addthe phosphorus salt (usually disodium phosphate)first so that it can enter the fat micelles. There is anappreciable risk of precipitation if calcium (usuallygiven as calcium gluconate/glubionate) is addedbefore the phosphorus.As much as 1-6 mmol of phosphorus and 1-35

mmol of calcium can be added in this way. 19Although some precipitation does occur, balancestudies indicate that the absorption and retention ofboth calcium and phosphorus from such solutions isgood.The multinutrient fortifiers added to breast milk

in some countries rely upon 'stabilising agents' tohold large quantities of mineral substrate in solu-tion. Problems with precipitation and reducedabsorption and retention have been reported,however, and in one study20 it was found thatretention of substrate from a preterm formulaexceeded that from a fortified human milk solution.The use of such fortifiers is still the subject ofintense research, and they cannot be unreservedlyrecommended for general use.The addition of phosphorus to preterm formula is

unlikely to be beneficial as the total amount ofcalcium retained and available for bone mineralisa-tion is already completely utilised. Further additionof phosphate may in fact precipitate calcium fromthe milk solution, further reducing availability of thesubstrate.Vitamin D3 should be given routinely to all

preterm infants.2' For infants born in the UnitedKingdom whose mothers have received vitamin Dsupplementation during pregnancy the currentrecommendations are for 400-1000 IU/day,although it should be noted that with adequatedietary provision of calcium and phosphorus as littleas 100 IU/day has been given without obvious

adverse outcome for bone. For infants born tomothers not receiving vitamin D supplementation,as much as 1200 IU/day may be required. Increasingthe supply of vitamin D3 further may increase theshort term retention of calcium, particularly in thoseinfants in whom the supply of substrate is poor, butmay also have longer term consequences for boneresorption and remodelling activities. In additionthere may be as yet undefined consequences oncellular differentiation in other tissues; vitamin D3 isa steroid type molecule, and a potent mitogenicagent.Most preterm infants will not benefit from being

given the active metabolites of vitamin D3; theseshould be confined to those patients in whom avitamin D3 resistant state, for example, X linkedhypophosphataemic rickets, has been confirmed.

Intravenously fed preterm infants are most at riskfrom poor supplies of mineral substrate. The solu-bility of calcium and phosphorus in parenteralnutrition solutions depends on a number of factors:the pH and the amino acid composition of thesolution, the calcium salt used, and the temperatureto which the solution is exposed.The more acidic the solution, the greater the

quantity of mineral substrate it can hold, and formost solutions calcium gluconate is more solublethan calcium chloride. Recently, amino acid solu-tions 'tailored' to the requirements of infants havebeen produced, with claims of improved capacity forcalcium and phosphate. The testing of thesesolutions has not, however, taken account of theprolonged exposure to high temperatures (37°C formore than two hours) that may obtain in clinicalpractice.The risk of precipitation with line blockage, or

even microembolisation of crystalline particles,should not be ignored when using solutions withcalcium and phosphate contents near the limit ofsolubility. The use of microporous filters within theline would reduce these risks appreciably. Usingcurrently available amino acid solutions infused at150 mllkg/day, 1-1 mmol/kg/day of phosphorus and1-5 mmol/kg/day of calcium can be administered atmost. Claims of improved solubility and mineraldelivery using glucose-i-phosphate have been made,but the use of an intermediary metabolite for thispurpose is still under investigation.

In addition to problems with calcium and phos-phorus delivery, the presence of high concentrationsof aluminium in some solutions has given rise toconcern about possible inhibition of osteoblasticfunction in infants receiving prolonged intravenousfeeding.

In clinical practice, the infants likely to requiremineral supplementation are those of less than 33

Bone disease in preterm infants 1409

weeks' gestation. A reasonable 'prophylactic'measure would be the addition of 0-5 mmol ofneutral phosphate solution to each 100 ml ofexpressed breast milk, or the use of a pretermformula containing at least 1 mmol/100 ml ofphosphorus, and 1-75 mmol/100 ml of calcium.Supplementation should start as soon as enteralfeeds are started, and continue until the infantachieves a weight of 2 kg or leaves the nursery. Aurinary calcium: phosphorus ratio of >1 at the ageof 3 weeks is an indication for further supple-mentation. For infants receiving preterm formulas,milks with higher mineral densities are availableoverseas, and may be introduced in the UnitedKingdom in the near future. Breast milk can besupplemented with both calcium and phorphorususing the previously described method.

It is hoped that early mineral supplementationwill result in a reduction in the incidence ofmetabolic bone disease during the neonatal period,and maximise the potential for subsequent bonegrowth.

References

Ziegler E, O'Donnel A, Nelson S, Fomon S. Body compositionof the reference fetus. Growth 1976;40:329-41.

2 Pitkin RM, Cruikshank DP, Schauberger CW, Reynolds WA,Williams GA, Horgis GK. Fetal calciotropic hormones andneonatal calcium homeostasis. Pediatrics 1980;66:77-82.

3 Delvin E, Glorieux F, Salle B, David L, Varenne J. Control ofvitamin D metabolism in preterm infants: fetomaternal relation-ships. Arch Dis Child 1982;57:754-7.

4 Wong GL. Skeletal effects of parathyroid hormone. In: PeckWA, ed. Bone and mineral research. Vol 4. Amsterdam:Elsevier, 1986:103-30.

5 Cooper U, Anast CS. Circulating immunoreactive parathyroidhormone levels in premature infants and the response to calciumtherapy. Acta Paediatr Scand 1985;74:669-73.

6 Cohn VD, Elting JJ. Synthesis and secretion of parathormoneand secretory I protein by the parathyroid gland. In: Peck WA,ed. Bone and mineral research. Vol 2. Amsterdam: Elsevier,1984:1-64.

7 Romagnoli C, Zecca E, Tortorolo G, Diodato A, Fazzini G,

Sorcini-Carta M. Plasma thyrocalcitonin and parathyroidhormone concentrations in early neonatal hypocalcaemia. ArchDis Child 1987;62:580-4.

8Talmage RV, Cooper CW, Toverud SU. The physiologicalsignificance of calcitonin. In: Peck WA, ed. Bone and mineralresearch. Vol 1. Amsterdam: Excerpta Medica, 1983:74-143.

9Sakamoto S, Sakamoto M. Bone collagenase, osteoblasts, andcell mediated bone resorption. In: Peck WA, ed. Bone andmineral research. Vol 4. Amsterdam: Elsevier, 1986:49-102.

0 Mundy GR, Roodman GD. Osteoclast ontogeny and function.In: Peck WA, ed. Bone and mineral research. Vol 5. Amster-dam: Elsevier, 1987:209-80.Baron R, Vignery A, Horowitz M. Lymphocytes, macrophagesand the regulation of bone remodelling. In: Peck WA, ed. Boneand mineral research. Vol 2. Amsterdam: Elsevier, 1984:175-243.

12 Anderson HC. Matrix vesicle calcification: review and update.In: Peck WA, ed. Bone and mineral research. Vol 3. Amster-dam: Elsevier, 1985:109-50.

13 Sedman AB, Klein GL, Merritt RJ, et al. Evidence ofaluminium loading in infants receiving intravenous therapy.N Engl J Med 1985;312:1337-43.

14 Cooper PA, Rothberg AD, Pettifor JM. Early follow-up of verylow birthweight infants after hospital discharge with respect togrowth and mineral homeostasis. J Pediatr Gastroenterol Nutr1988;7:577-82.

s5 Kulkarni PB, Hall RT, Rhodes PG. Rickets in very lowbirthweight infants. J Pediatr 1980;2:249-52.

16 Lucas A, Brooke OG, Baker BA, Bishop N, Morley R. Highalkaline phosphatase activity and growth in preterm neonates.Arch Dis Child 1989;64:902-9.

17 Greer FR, Lane J, Weiners S, Mazess RB. An accurate andreproducible absorptiometric technioque for determining bonemineral content in newborn infants. Pediatr Res 1983;17:259-62.

'8 Chan GM, Mikur L, Hansen JW. Effects of increased calciumand phosphorus formulas on bone mineralisation in preterminfants. J Pediatr Gastroenterol Nutr 1986;5:444-9.

19 Salle B, Senterre J, Putet G, Rigo J. Effects of calcium andphosphorus supplementation on calcium retention and fatabsorption in preterm infants fed pooled human milk. J PediatrGastroenterol Nutr 1986;5:638-42.

20 Schanler RJ, Abrams SA, Garza C. Bioavailability of calciumand phosphorus in human milk fortifiers and formula for verylow birthweight infants. J Pediatr 113:95-100.

21 Greer FR, Tsang RC. Calcium, phosphorus, magnesium andvitamin D requirements for the preterm infant. In: Tsang RC,ed. Vitamin and mineral requirements in preterm infants. NewYork: Marcel Dekker. 1985:99-136.

Correspondence to Dr N Bishop, Dunn Nutritional Laboratory,Downhams Lane, Milton Road, Cambridge CB4 IXJ.