Annals of the Rheumatic Diseases, 1979, 38, 401-412

Hypercalcaemia in rheumatoid arthritis:investigation of its causes and implicationsALASTAIR C. KENNEDY, BAGHAT F. ALLAM, PATRICK J. ROONEY,MICHAEL E. WATSON, ANGELA FAIRNEY,1 KEITH D. BUCHANAN,2AND CATHERINE J. HILLYARD3

From the Centre for Rheumatic Diseases and University Departments of Medicine, Pathological Biochemistry,and Surgery, Glasgow Royal Infirmary, Glasgow, the 'Department of Chemical Pathology, St Mary's Hospitaland Medical School, University ofLondon, London, the 2University Department ofMedicine, Queen's University,Belfast, and the 3Endocrine Unit, Royal Postgraduate Medical School, Hammersmith Hospital, London

SUMMARY When correction was made for hypoalbuminaemia, 23 of 50 ambulant patients withdefinite or classical rheumatoid arthritis were found to have hypercalcaemia. When these 23 patientswere studied 6 months later, 7 had hypercalcaemia as defined by the correction factor for a low serum

albumin level, and 6 of these patients had raised serum ionised calcium concentrations. Biochemicalstudies in the 23 patients indicated evidence of hyperparathyroidism, namely, hypophosphataemia,increased serum alkaline phosphatase, hyperchloraemia, and reduced tubular reabsorption ofcalcium. However, serum immunoreactive parathyroid hormone concentrations were normal. Onlyone patient had an abnormally low serum 25-hydroxy-vitamin D result: this patient had a high levelof urinary D-glucaric acid and was receiving phenobarbitone for treatment of epilepsy. The bio-chemical features suggestive of parathyroid overactivity were particularly found in patients withraised serum calcium levels. The cause of hypercalcaemia in rheumatoid arthritis remains to beexplained.

Osteoporosis is common in patients with rheumatoidarthritis even without corticosteroid therapy.Radiological studies have shown that osteoporosisis generalised, occurring to the same degree in allparts of the peripheral skeleton (Kennedy andLindsay, 1977). Rarely, osteomalacia may be foundin patients with rheumatoid arthritis, usually inelderly patients confined to home (Maddison andBacon, 1974). However, most physicians wouldprobably not consider that there was a generalmetabolic bone disorder underlying osteoporosis inrheumatoid arthritis. We were therefore surprisedto find an unexpectedly high prevalence of hyper-calcaemia in patients with the disease (KennedyAccepted for publication 20 October 1978Correspondence to Dr Kennedy at Division of Rheumatology,Dept. of Medicine, State University of New York, Buffalo,3495 Bailey Ave., Buffalo, NY 14215, USA.

et al., 1975), and this paper describes further studieson the pathogenesis of this hypercalcaemia.

Material and methods

PATIENTS STUDIEDTwenty-three of 50 randomly selected femalepatients suffering from 'definite' or 'classical'rheumatoid arthritis, as defined by the diagnosticcriteria of the American Rheumatism Association(Ropes et al., 1959) were found to be hypercalcaemic,that is, serum calcium concentration greater than2.60 mmol/l when the total serum calcium concen-tration was corrected for the corresponding albuminconcentration as outlined below. All these 23 patientswere ambulant and all participated in the studywith full knowledge of its content. All blood samples

Table 1 Clinical and laboratory data of the patients vithl rheumatoid arthritis included in the study (ii=23)

Age (years) Duration of Functional Age at Duration Rheumatoid Antinuclear Haemoglobin Erythrocytearthritis grade menopause since factor factor concen- sedimen-(jears) (years) menopause titre titre tration gll tation rate

(years) mm/l h

Mean 4 SD 51-7±15-6 6-0±6-6 1-7±0-6 45-5±2.3 15-6±7-3 18-2±22-9 217±513 12-2±1-4 52-5±33-5

401

on July 7, 2020 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.38.5.401 on 1 O

ctober 1979. Dow

nloaded from

402 Kennedy, Allam, Rooney, Watson, Fairney, Buchanan, Hillyard

were obtained without haemostasis between 8.30and 9.30 a.m., the patients having fasted and takenno medication since 6 p.m. the evening before. Priorto the study all patients were taking medication:1 patient was receiving digoxin and a diuretic, 1was on phenobarbitone for epileptic seizures, 3were receiving low-dose corticosteroid therapy, thatis, less than 10 mg prednisolone per day, and theremainder were receiving a variety of non-steroidalanti-inflammatory analgesics. The clinical and labo-ratory data of the 23 patients arc summarised in'Table 1.

DESIGN OF CORRECTION OF S.ERUMCALCIUM CONCENTRATION FORHYPOALB UMINAEMIAThe serum calcium levels were corrected for albuminemploying a formula derived in the following way. Aregression analysis ofcalcium on albumin was carriedout on the results from 150 normal healthy persons,half of whom were female. Their age ranges wereas follows: 17 to 69 years for females, and 18 to70 years for males. A regression equation was thencalculated on the total 150 subjects. It is of interestthat when the serum albumin concentration waszero the regression line cut the Y axis at 1 .43 mmol/l,which represents the ultrafiltrable calcium con-centration, and corresponds to that found byRobertson (1969) (1.40 mmol/l); Pedersen (1970)(1.40 mmol/l); Rose (1972) (1.45 mmol/l); andPayne et al., (1973) (1.42 mmol/l) Orrell (1971) founda somewhat higher value for ultrafiltrable calcium(1 .70 mmol/l), probably in part because the subjectsstudied included patients with nephrotic syndrome,a condition in which there is evidence of a tendencyto hypercalcaemia (Jones et al., 1967).The regression equation derived from the analysis

was:Y=(0.0194 x X) + C

where X=serum albumin in g/lY=serum calcium in mmollI

and C is a constant equal to 1 434(value of Y when X=O).

The mean albumin for the population was 47.0 g/l,and the following correction factor for serumcalcium concentration was constructed: Ca (cor-rected)=(47 - alb.) x 0.0194 + Ca where Ca isthe measured serum calcium is mmol/l, alb. is theserum albumin in g/l, and Ca (corrected) is the serumcalcium in mmol/l corrected for albumin.

This indicates that calcium concentration changesby 0.0194 mmol/l for every 1 g/l change in albuminand that therefore for every 1 g/l by which theplasma albumin exceeds the normal mean of 47 *0 g/l0.0194 is subtracted from the total plasma calcium

and a corresponding addition is made when plasmaalbumin is less than 47-0 g/l.

In addition, a biochemical mineral metabolismscreen was carried out on each subject. This involvedutilising data from blood and urine to calculatecertain derived values: (1) fasting urinary calcium/creatinine ratio; (2) urinary calcium excretion (CaE)per litre glomerular filtrate (Nordin et al., 1967);(3) theoretical maximum tubular reabsorptioncapacity (Tm) for calcium, TmCa (Nordin et al.,1976); (4) the most reliable indices of renal handlingof phosphate (Peterson, 1973), i.e., TmPO4 GFR(theoretical tubular maximum reabsorption ofphosphate corrected for glomerular function (Bijvoet,1969; Bijvoet et al., 1969) and IPE (index of phos-phate excretion) (Nordin and Bulusu, 1968).The derivation and rationale of employing the

CaE per litre GF and the TmCa are well describedby Marshall (1976) and by Nordin, Horsman andAaron (1976). The calculation of these indices wascarried out in the MRC Mineral Metabolism Unitat Leeds Royal Infirmary, and we are grateful forthe co-operation extended by Professor B. E. C.Nordin and colleagues.

MEASUREMENT OF SERUM IONISEDCALCIUMA 10 ml clotted sample of blood, from which airwas excluded immediately after withdrawal, wasused for the estimations of ionised calcium. Serumionised calcium concentration was measured with aflow-through electrode system (Orion 99-20) inwhich the electrodes were maintained at 370C in athermostatically controlled cabinet. At each measure-ment run the electrode was calibrated for directsodium interference as well as for its response tostandard calcium solutions, since the effect of sodiumon the electrode has been found to vary (Watson,1975). Immediately after each measurement ofionised calcium the pH of the same serum sample wasmeasured at 370C with a glass capillary electrodesystem (EIL Model SHH 33).

Corrections were applied to the observed ionisedcalcium results to take account of direct sodiuminterference, and for this purpose a linear relation-ship was assumed between log (Cal+) and sodiumconcentration (Watson, 1975). The ionised calciumat the measured pH was then adjusted to corres-pond to a standard pH value of 7.40. For any serumspecimen the relationship between the logarithm ofthe ionised calcium concentration and pH is linear,but the slope of the regression line varies withalbumin concentration and, to a lesser degree, withthe actual ionised calcium level. Accordingly, in orderto make pH corrections this slope was calculated

on July 7, 2020 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.38.5.401 on 1 O

ctober 1979. Dow

nloaded from

Hypercalcaemia in rheumatoid arthritis: investigation of its causes and implications 403

for each sample by a formula derived fromtheoretical studies of calcium binding to albumin(Pedersen, 1972) and from the results of observationsof the influence of pH on serum ionised calciumwith varying albumin and calcium concentrations(Watson, 1975). The reproducibility of this methodhas been shown to be highly satisfactory, with acoefficient of variation of 0-760% (Watson, 1975).The normal range of serum ionised calcium con-centration was established in an age- and sex-matched group of healthy persons who were eithermembers of staff or related to members of staff ofthe Centre for Rheumatic Diseases and in whomthere was no evidence of disease associated withdisturbance of calcium metabolism. These bloodsamples were also obtained in the fasting state andunder the same stringent conditions described for thepatient group.

OTHER ASSAYSSerum immunoreactive parathormone was measuredby Dr Angela Fairney of St Mary's Hospital(Fairney et al., 1973) and by Mrs Catherine Hillyardof the Royal Postgraduate Medical School in London(Berson et al., 1963). Calcitonin assays were per-formed by the latter (Coombes et al., 1974). Esti-mation of serum 25-hydroxy-vitamin D was per-formed by the method of Haddad and Chyu (1971)using vitamin-D-replete rat kidney cytosol protein(McLaughlin et al., 1974). Other assays includedserum immunoreactive gastrin (Buchanan andMcCarrol, 1971), serum immunoreactive pancreaticglucagon (Buchanan, 1973; Flanagan et al., 1974),urinary D-glucaric acid excretion (Simmons et al.,1974), and urinary hydroxyproline (Goverde andVeenkamp, 1972).

In all assays reference data were obtained fromestimations performed on fasting subjects matchedfor age and sex.

RADIOLOGICAL ASSESSMENTRadiological films of the hands and wrists wereassessed for joint space narrowing and erosions bythe method described by Sharp et al. (1971). Themetacarpal index of Barnett and Nordin (1960) wasused as a measurement of osteoporosis.

STATISTICAL METHODSThe following tests were used: Student's t test forcomparison of group means, Fisher's F test forcomparison of group variances, and linear regressioncalculations for association between variables.

Results

The results of the mineral metabolic screen are

summarised in Table 2. Both the corrected serum

calcium and the serum ironised calcium concen-trations were significantly raised (P<0.001). Therewas a significant correlation between the serumionised calcium and the uncorrected serum calciumconcentration (r=0.55, P<0 .05), and this becameeven more significant (r=0.71, P<0.005) when theformer was correlated with the corrected serumcalcium level (Figs. la and b). The serum phosphateconcentration was significantly reduced (P<0.01),and 10 patients had raised levels of serum alkalinephosphatase. Plasma chloride concentrations weresignificantly raised and 5 patients had increasedcalcium creatinine ratios. Five patients had reducedrenal tubular reabsorption of phosphate (TmPO4/GFR) and 7 had increased phosphate excretion(IPE). When CaE is plotted against total serumcalcium concentration (Fig. 2) the results are to theright of the mean and 7 are outside the limits ofnormality, indicating increased renal tubular re-absorption of calcium (Marshall, 1976).When the serum ionised calcium, serum phosphate

and alkaline phosphatase are grouped according to

2 7

- 25EE

2 3 -

E

E2a,Ln

io

3-0

7- ,

E

E 2 4

a,

* a

* .r = 0-55

0 * 0 t =302

*1 0 1-1 1.2 i.3

6- a

0 **r=0 714~~~~~~~~~~~~~~~~ t=464

0 1.1 1 2 i. 3Serum ioneed calcium mmol/I (measured)

Fig. 1 Correlation between serum ionised calcium and(1 a) total serum calcium not corrected for albuminconcentration (r=0-55; P<0-05); (lb) total serumcalcium corrected for albumin concentration (r==00.71;P<0 005).

on July 7, 2020 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.38.5.401 on 1 O

ctober 1979. Dow

nloaded from

404 Kennedy, Allam, Rooney, Watson, Fairney, Buchanan, Hillyard

Table 2 Results (mean + SD) of mineral metabolism screen in the patients with rheumatoid arthritis compared to

Serum Serum Serum Serum Serum Serum Plasma Plasmaionised calcium calcium phosphate alkaline albumin bicarb. sodiumcalcium (uncorrected) (corrected) mmol/l phosphatase gil mmol/l mmol/lmmol/l mmol/l mmol/l U/I

Patientgroup 1-13+0-06 2-41+0-13 2-58+0-12 0-98+0-2 318±(130-1490) 38-2+3-9 24-3+2.2 139-3+2-2

Normalreferencegroupranges 1-07±0-05 2-39+0-1 2-39+0-1 1-1+0-15 80-280 46-9+3-3 24-5+1-8 140+2-0

t 4-23 8-4 2-81 9-7P <0.001 NS <0.001 <0-01 * <0-001 NS NSNumberabovereferencerange 8 1 7 0 10 0 0 1

Numberbelowreferencerange 0 1 0 5 0 4 1 0

Statistical analysis invalid because of the non-normally distributed data.

Table 3 Grouping of combinations of 'normal' and'abnormal' serum ionised calcium, phosphate andalkaline phosphatase resultsSerum Serum Serum Number ofionised phosphate alkaline patients incalcium phosphatase each group

N N N 7 (30-9%)t or N or; torNor; -

N N 6 (26-0%)N 4 N 2 (8.6%)t N N 2 (8.6%)t N t 3 (13-0%)t 4. N 2 (8.6%)1 t 1(4.3%)

0

0

2u 2b 3 0 35Total serum calcium (mmol/l )

Fig. 2 Plot of calcium excretion (CaE) per litre ofglomerular filtrate, against total serum calciumconcentration. The shaded area represents the limitsof normality.

their 'normality' (Table 3) the following factsemerge. Of the 23 patients only 7 had all 3 resultsin the 'normal' range. Of the 8 patients with raisedserum ionised calcium 6 had biochemical patternswhich might be encountered in a hypercalcaemichyperparathyroid state. It can be seen in Table 4that the mean serum alkaline phosphatase lies abovethe upper limit of the reference range. Indeed, 10

of the 23 patients had raised serum alkaline phos-phatase levels, whereas there was no patient whoshowed increases of serum bilirubin, aspartate

Table 4 The results (mean i SD) ofserum alkalinephosphatase, bilirubin, SGOT and SGPT in the patientswith rheumatoid arthritis, compared to the referencegroup.

Serum Serum Serum Serumalkaline bilirubin asparate alaminephosphatase pmol/l transaminase transaminaseU/l (SGOT) (SGPT)

U/Il U/l

PatientgroupMean 318 4.7 23.4 16-7Range 130-1490 2-14 14-31 11-28

Referencerange 80-280 3-14 12-38 8-41

Number ofpatientswith resultsaboveupperlimit ofreferencerange 10 0 0 0

150-

ILo 100'

E

wau 50Q

I I-% ^ 1% I

I

on July 7, 2020 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.38.5.401 on 1 O

ctober 1979. Dow

nloaded from

Hypercalcaemia in rheumatoid arthritis: investigation of its causes and implications 405

the normal reference group. Ranges or mean ± ISD whsn appropriate

Plasma Plasma Plasma Serum Maximum Index of Calciuml TmCa/GFR Hydroxy-potassium chloride urea creatinine tubular phosphate creatinine mmol/l prolinelmmol/l mmol/l mmol/l mmol/l reabsorption excretion ratio creatinine

ofphosphate (IPE) (mmol/mmol) ratio(TmPO4 mmol/mmolGFR) mmol/l

4-1±0-5 106±3-3 5-4±1-9 83±30 0-96±0-36 +0-25±0-17 0-38±0-28 2-04±035 0-018±0-016

4-2±0-4 100±25 2-7-7-0 35-115 0-70-135 0.70-1-35 0-06-0-46 1-68-2 08 0.028-53

NS <0.001 NS NS * * * * *

1 9 3 1 2 7 5 12 6

3 0 0 0 5 0 1 0 0

transaminase (AST), or alanine transaminase(ALT). One patient had extremely high serumalkaline phosphatase, and this was associated withthe highest serum ionised calcium in our patients.The results of the urinary hydroxyproline showedthat 9 of the patients studied had levels in excess ofnormal limits (0.05-0.17 mmol/24h). However,when these results were subsequently calculated foraccuracy of interpretation in terms of the hydroxy-proline/creatinine ratio only 6 patients had abnor-mally high results. Since the urinary hydroxyprolineand the fasting calcium/creatinine ratio suggestevidence of bone destructions (Nordin et al., 1976)the results for urinary calcium and urinary hydroxy-proline levels were correlated in the patients studied(Fig. 3). It can be seen that there is a statisticallysignificant positive correlation (r=0.54, P<0 005)between these indices.The results of the immunoreactive parathormone

assay (Fairney et al., 1973) and other hormoneassays are shown in Table 5. It can be seen that nopatient had an elevated parathormone level, andindeed in 16 the level was below the lower limit ofthe normal range. In the other assay (Berson et al.,1963) all levels lay within the normal range. Onlyone patient (patient 11) had an raised serum immuno-reactive gastrin level: this patient also had raisedserum calcitonin and ionised calcium levels, a lowparathyroid hormone (PTH) level, and the highestserum alkaline phosphatase. The patient with the low25-OH vitamin D level (patient 23) had a markedlyraised urinary D-glucaric acid concentration. Thispatient had been receiving phenobarbitone forepilepsy. Serum magnesium concentrations (0 80 ±0-06 mmol/l) were within normal limits.

70 - 102

E 60 -

E

E-

a./

40-* /

0o1 * /E 30 - * *

20 0 0a/ * ~~~~~~~~~r0 54

v10- p=~~~~~~~P<0-005

0 010 020 030 0-40Unrnary hydroxyproline concentration mmol/lI)

Fig. 3 Correlation between the calcium ratio and thehydroxyproline ratio derivedfrom fasting urinesamples (r= 0 54; P<0 - 005).

Patients with an elevated serum ionised calciumlevel tended to have lower serum phosphate levelsand TmP/GFR and higher TmCa/GFR (Table 6).No significant differences were noted in the clinicaland other laboratory features between the patientswith elevated and normal serum ionised calciumlevels.

Analysis of the bone damage indices revealed asignificant correlation between the erosion score andthe metacarpal index (r==0.6, P<0.01) and theerosion score and joint space narrowing score

on July 7, 2020 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.38.5.401 on 1 O

ctober 1979. Dow

nloaded from

406 Kennedy, Allam, Rooney, Watson, Fairney, Buchanan, Hillyard

i t- t0_ 00 00I V-sA nQ u 0 00 -

Cl0Cl-.. ..000C-l O -? 0

0000000000000000000 1000A00

>*F o +> fi F*

- 0%i Cl %00cI0-egIn 1 0Oo-oooooooAooo6PI 00OI? . ... ...~.!. .. .... . ..?

NC00.o0 00000aOO,0%0 000 O ~oooooooooooooo6600066o

000 000000C%00N-0%00-Noo0%oom~~~~~0 ~0 ~u Cl0l - Clu0 000%0 0 u 0 -0

_-_ _ er -- --

-HT en0%N in

-H

co o00p .

co

o 0C Cl

00

_ l

j0 * 000%NNzF 0% 0%'Iu o0%Suf10% ClcY

^ClClc_Cl<e,i,I CC

i-ClCeN C

80X888088888°8888888V VVVVVVVV VVIVVvVVVVI

^~~~~~4 t- 8 t- g_ -8>F>oo t-} t-> _) WI N f> o

Cl00 00000000000 00000000-000

C~IN~00N~ NCl'lClC'N N C eCl

ON1lCl%C0%WI N0 N0'l %%0 0V000Cl..C~~. 00..*..

-HN 0%Cl

-0. .0

-HrzT rb'oo? o? ol oT

or

o o~ or o!'IT" " " 00 0" 00 0Vl0 l 0 IClV 0

N " 00Cl...............N ..........................- -.......

ClCllCllCllCllCllClCClCClCClCClCClC ClO

0 - 0 CC'4 If, N8 8 - 01

N'O C lcWV N 0 C

en- Cn en I -en (In en e en00ne e 0 en CleoI

ooen tV WI^ o

N Wo 0^. " ClClClf RO 0

- N * v" c t- W h 0 V " %O t- rvr4 M

_~~~~~~- --n - --0 -oo-,--e (4 N

o0 -

-H!$ 0.ClN 0

-HN*o

-H

bo

,~ X

L.Sbcoo

W.-

l E z

S E

0.t

.t

1.1

t .E

I._

0A

0A

0

0

co

V

V

00O

N Cl

AsClC. X

Ez X

o-

on July 7, 2020 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.38.5.401 on 1 O

ctober 1979. Dow

nloaded from

Hypercalcaemia in rheumatoid arthritis: investigation of its causes and implications 407

Table 6 Comparison of the biochemical data between the normocalcaemic and hypercalcaemic patients asdetermined by their ionised calcium concentration

Number Serum ionised Total serum Serum TmP/GFR TmCa/GFR Calciuml Hydroxy-calcium calcium phosphate mmol/l mmol/l creatinine prolinelmmol/l (corrected) mmol/l ratio creatinine

mmol/l ratio

Hypercalcaemicpatients >1 * 13mmol/l 8 1-20±0-04 2-66±0-12 0-90±0-24 81-63±48-16 2-2±0-14 0-36±0-3 0-018±0-010

Normocalcaemicpatients < 1 * 13mmol/l 15 1*09±0-03 2-53±0-08 1*06±0-16 103-5±25-6 1-9±0-5 0-39±0-3 0-019±0-23

t 2-.15 3.45 1*97 1*43 1*42 0-22 0-12P <0 05 <0.005 <0-1 (NS) NS NS NS NS

(r=0 - 77, P<0 * 001). No correlation existed betweenserum ionised calcium concentration and any ofthese bone damage indices.

Discussion

Bone loss is an important clinical facet ofrheumatoidarthritis. When severe osteoporosis is present,compression fractures of the vertebrae may occur,and in peripheral bones thinning of cortical bonemay complicate joint surgery. So far there has beenlittle evidence that osteoporosis in rheumatoidarthritis has a 'metabolic' basis, and it was thereforeof interest when we observed that a high proportionof patients had raised serum calcium concentrationswhen the serum calcium was corrected for the lowalbumin concentration (Kennedy et al., 1975).It seemed relevant, therefore, to study in greaterdepth the pathogenesis of hypercalcaemia in rheu-matoid arthritis. To this end 50 female patients withrheumatoid arthritis were randomly selected andscreened for possible hypercalcaemia after correctionof serum calcium for hypoalbuminaemia. Twenty-three of the^e patients with 'definite' or 'classical'rheumatoid arthritis were shown to have an elevatedcorrected serum calcium concentration and werechosen for detailed investigation of biochemicaland hormonal aspects of calcium metabolism.The first and most crucial question is whether or

not the serum ionised calcium estimations validlysupport the contention that hypercalcaemia inrheumatoid arthritis is a real entity. The study wascarefully controlled both in terms of matchingpatients for age and sex with healthy persons and inthe methods employed to determine serum ionisedcalcium concentrations. The patients and controlswere studied in the fasting state, and particular carewas taken in withdrawal of venous blood to ensurethat no air entered the syringe and that the arm wasnot constricted. The patients and controls werestudied contemporaneously and the laboratoryestimations were performed under identical con-

ditions. The temperature at which the estimationswere performed was constant (370C) for all samples,and the results were corrected for pH and sodiuminterference. Hence there are strong grounds foraccepting that 8 of the 23 patients with rheumatoidarthritis did have raised serum ionised calciumconcentrations.

It might be argued that the mean control value ofserum ionised calcium concentration was somewhatlower (1 .07 mmol/l) than that reported by otherworkers. However, it is extremely difficult to makecomparisons of normal values obtained in differentlaboratories, in particular when there are so manyvariables which can affect such results. Moreover, aliterature survey revealed that only 3 other groupshave reported serum ionised calcium determinationsusing the ion electrode method employed in thepresent study while also studying patients in thefasting state. Robertson and Peacock (1968) andLadenson and Bowers (1973) both obtained highermean values, 1 -23 and 1 .28 mmol/l respectively,but Ryden et al., (1976) obtained a mean value of1-03 mmol/l, which was lower than that obtainedin the control subjects in the present study. It shouldbe noted that Ladenson and Bowers (1973) carriedout their estimations at a temperature of 250C,which would tend to increase the serum ionisedcalcium result. Ryden and her colleagues (1976)do not, unfortunately, state the temperature at whichthey carried out their estimations. Varghese (1973),using spectrophotometry, obtained a mean valueslightly lower (1 -04 mmol/l) than that obtained inthe present study. His subjects were also fasting andthe working temperature was 370C. The repro-ducibility of the method employed in the presentstudy was high, with a coefficient of variation of0.76% in 20 samples from the same bulk source(Watson, 1975). Thus it seems reasonable to con-clude that the 8 patients with rheumatoid arthritishad definite, though in some patients slight, ele-vations in serum ionised calcium levels.

Further confirmation of the finding of raised serum

on July 7, 2020 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.38.5.401 on 1 O

ctober 1979. Dow

nloaded from

408 Kennedy, Allam, Rooney, Watson, Fairney, Buchanan, Hillyard

ionised calcium levels in rheumatoid arthritis hasbeen obtained in a study of 52 patients with thedisease studied in South Africa. Approximatelyone-third (17) of the patients had raised serumionised calcium levels when compared to a healthycontrol group matched for age and sex (Meyerset al., 1976). This method employed an Orion calciumion exchange electrode model SS-20 and the tem-perature at which estimations were performed was370C. The correction factor employed in the presentstudy is also central to the investigation, and itsvalidation is therefore crucial. As previouslyindicated, the regression line of serum calciumconcentration on serum albumin level when extendedto the Y axis when the albumin concentration iszero intersects at 1 .43 mmol/l, which correspondsto the serum ultrafilterable calcium concentrationfound by others (Robertson, 1969; Pedersen, 1970;Rose, 1972; Payne et al., 1973). Of great importanceis whether there is concordance between hyper-calcaemia estimated from the correction formula andhypercalcaemia confirmed by a high value of serumionised calcium. Of the 23 patients with rheumatoidarthritis in the present study, 7 had hypercalcaemiaas estimated by the correction formula for a lowserum albumin concentration. Six of these 7 patientshad elevated serum ionised calcium levels. Patient4 (Table 5), who had a normal value for serumionised calcium (1 .12 mmol/l) had a corrected totalserum calcium concentration just over the upperlimit of normal (2-65 mmol/l). Of the 16 patientswho had normal values for corrected serum calciumconcentrations 2 had increases of serum ionisedcalcium (patients 14 and 15; Table 5). There is thusreasonable concordance between the correctedtotal serum ionised calcium concentration and theserum ionised calcium level, and this tends to sup-port the validity of the correction formula. Furthersupport is provided by the fact that the correlationbetween serum ionised calcium concentration andthe corrected total serum calcium level (r=0 71,P<0.005) is higher than with the uncorrectedtotal serum calcium (r=0.55, P<0.05).The serum ionised calcium is the physiologically

active fraction of total serum calcium, and con-sequently, when raised, it is of clinical significance.The associated biochemical results of the mineralmetabolism screen are therefore of interest. Thefinding of hypophosphataemia, increased serumalkaline phosphatase, hyperchloraemia, reducedtubular reabsorption of phosphate, and increasedtubular reabsorption of calcium suggest possibleparathyroid overactivity. The elevation of serumalkaline phosphatase in patients with rheumatoidarthritis may be ofeither bone (Maddison and Bacon,1974) or liver origin (Webb et al., 1975). Unfortu-

nately, isoenzyme studies were not available todetermine the source of the increased serum alkalinephosphatase levels in the present study. However,none of the patients had clinical evidence of liverdisease, and all had normal serum levels of bilirubinand the enzymes AST and ALT. It is of interest thatof 6 patients with elevated hydroxyproline/creatinineratios 4 had elevated serum alkaline phosphataselevels. However, 6 patients with elevated serumalkaline phosphatase levels had normal hydroxy-proline/creatinine ratios. Increased hydroxyprolineexcretion in the urine reflects an increased turnoverof collagen, and has been reported by some (Hart-mann et al., 1969) but not all workers (Ziff et al.,1956) in patients with rheumatoid arthritis. Thesignificant correlation between the increased urinarycalcium and hydroxyproline results (Fig. 3) suggeststhat the underlying cause for this is likely to beincreased bone turnover.The question arises whether this evidence sug-

gesting increased bone turnover and the otherbiochemical abnormalities noted might be interpretedas indicating hyperparathyroid activity. It is there-fore of interest that patients who had raised serumionised calcium level tended to have lower serumphosphate levels and TmP/GFR and higher TmCa/GFR. Thus, one can only tentatively conclude thatthere is some evidence of increased bone turnoverand possible parathyroid overactivity. In this con-text the results of the hormone assays were ofparticular interest.The most accurate test for parathyroid over-

activity is regarded as the direct measurement ofserum PTH. By both methods of immunoassayemployed in this study the serum levels of PTHwere normal or reduced. Serum calcitonin levelswere elevated in two patients only, and it is ofinterest that 1 of those had the highest serum ionisedcalcium concentration (patient 11, Table 5). Thecause of the elevated calcitonin level may be a res-ponse to the elevation of serum ionised calcium ashas been shown in some mammalian species (Grayand Munson, 1969), but not yet in man (Schneiderand Sherwood, 1974).

Hypercalcaemia and hyperparathyroidism may beassociated with hypergastrinaemia (Turbey andPassaro, 1972; Barreras, 1973), and rheumatoidarthritis may also be associated with high serumimmunoreactive gastrin levels (Rooney et al., 1976).This suggests a possible link between enteric hor-mones and anomalies in calcium metabolismoccurring in rheumatoid arthritis. However, only 1of the patients in the present study (patient 11, Table5) had a raised serum immunoreactive gastrin level,though it is of interest that this patient had thehighest serum ionised calcium concentration. In

on July 7, 2020 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.38.5.401 on 1 O

ctober 1979. Dow

nloaded from

Hypercalcaemia in rheumatoid arthritis: investigation of its causes and implications 409

addition this patient had a raised serum calcitoninlevel, and it might be due to the fact that hyper-gastrinaemia can stimulate calcitonin secretion inman (Hennessy et al., 1973). Serum glucagon levelswere normal in all patients studied.The role of vitamin D in calcium metabolism in

patients with rheumatoid arthritis has not beenthoroughly investigated. In a recent study of 5patients with rheumatoid arthritis bone biopsy,showed osteomalacia and secondary hyperpara-thyroidism, and these findings were considered to besecondary to a dietary deficiency of vitamin D,assessed retrospectively by dietary histories (Mad-dison and Bacon, 1974). However, the analysis ofserum 25-hydroxy-vitamin D levels in this studyrevealed only 1 abnormally low result, in a patientwith associated high level of urinary D-glucaricacid who was receiving phenobarbitone for treat-ment of epilepsy. It seems reasonable to postulatethat the low level of vitamin D was due to livermicrosomal enzyme induction secondary to anti-convulsant therapy (H4ahn et al., 1972; Rowe andStamp, 1974). A number of antirheumatic drugsare also known to be associated with liver micro-somal enzyme induction in experimental animals(Conney, 1967), but there was no evidence of thisoccurring in patients in the present study as judgedby urinary D-glucaric acid levels.The patients in Maddison and Bacon's study were

selected because they had bone fractures and theytended to be older and more severely affected by thearthritis than the patients in this study. In additionit seems germane to note that in the radiologicalstudies of skeletal status in very large numbers ofpatients with rheumatoid arthritis previouslyreportedno circumstantial evidence of osteomalacia was

found (Kennedy and Lindsay, 1977). In addition,the absence of supportive biochemical evidence in a

prior study (Kennedy et al., 1975) would seem tosuggest that the incidence of hypovitaminosis D isprobably confined to a very small select group of therheumatoid population and indeed perhaps not toany greater extent than that in the general population.Examination of the indices of articular bone

damage showed significant correlation betweenthe erosion score and the metacarpal index, tendingto suggest that whatever is occurring at the jointsurface is also being reflected at the midpoint of theshaft of the metacarpal. Thus, the metacarpalindex would seem to provide a reasonably simpleguide to the extent of bone damage occurring in thehand of a rheumatoid patient. There was also a goodcorrelation between the erosion score (Dc) and thejoint space narrowing score (JSNc), suggesting thatthe development ofbony defects parallels the develop-ment of cartilage destruction.

None of the patients in this study had any evidenceof any alternative disease which could have explainedthe hypercalcaemia such as the milk-alkali syndrome,hypervitaminosis D, sarcoidosis, or malignancy. Itis possible that the increased serum calcium con-centrations might be due simply to erosion of bone,in which case it might be anticipated that they wouldcorrelate with activity of joint disease as judged byclinical, laboratory, and radiological parameters.There was no such correlation between hypercal-caemia and disease activity based on the patient'spain index, articular index of joint tenderness,functional grade, haemoglobin concentration, ery-throcyte sedimentation rate, and serum proteinconcentration, nor with the radiological indices ofbone damage. However, this does not totally excludethe possibility of any influence of disease activityon the production of hypercalcaemia, since thenumber of patients studied was small and the diseaserelatively long standing. Nevertheless, the fact thatthe hypercalcaemia is apparently not a constantfeature for any one patient could be consistent withvariation in the rate of bone destruction, which is notan unlikely possibility in a disease with such clinicalvariation.Another variable which might explain the elevated

serum ionised calcium levels is administration ofantirheumatic drugs. None of the patients had takenany medication within 12 hours of the study, butpersistence of drug effect cannot be discounted.Also it is possible that non-steroidal anti-inflam-matory drugs or corticosteroids may interfere withplasma binding of calcium, thus producing abnor-malities in serum ionised calcium measurements.Against, this however, is the fact that all the patientsin this study had had no drug therapy for a minimumof 12 hours before its start, and additionally thereis no reported evidence that these drugs do produceinterference in this way.

It might be argued that cessation of therapy initself could cause a release of calcium from itsbinding sites on albumin. However, it is of interestthat the patients with rheumatoid arthritis studied inSouth Africa (Meyers et al., 1976) had approximatelythe same incidence (30%) of increased serum ionisedcalcium levels and all were currently receiving non-steroidal anti-inflammatory drugs. Northover (1973),using amounts of indomethacin in excess of normaltherapeutic doses, has shown in vitro that the drugwill reduce calcium binding to the membranes ofendothelial cells, but the relevance of this singleobservation to the present clinical studies is notapparent. Careful search of the literature has failedto find any other publication on the effects of non-steroidal anti-inflammatory drugs on serum ionisedcalcium determinations.

on July 7, 2020 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.38.5.401 on 1 O

ctober 1979. Dow

nloaded from

410 Kennedy, Allam, Rooney, Watson, Fairney, Buchanan, Hillyard

If hyperparathyroidism is not the cause of theincreased serum ionised calcium concentrations andthe other biochemical abnormalities observed, thequestion arises of what they may be due to. Thesituation in rheumatoid arthritis has some resem-blance to that pertaining to increased serum ionisedcalcium concentrations in neoplasm, where in thepresence of biochemical evidence strongly suggestiveof hyperparathyroidism immunoassay of PTH maybe normal (Heath, 1976; Woodhead and Walker,1976). Conceivably in rheumatoid arthritis and incertain neoplasms polypeptides may be producedwhich have PTH-like activity but cannot yet bemeasured by the current immunoassay procedures.An alternative explanation is that the hypercalcaemiamay be the result of effects of prostaglandin E2.Prostaglandin levels are increased in the synovialfluid in rheumatoid arthritis (Higgs et al., 1974), andRobinson et al. (1975) have shown in bone culturestudies in vitro that all the bone-resorption-stimu-lating activity in rheumatoid synovial fluid can beaccounted for by prostaglandin E2.Krane (1974) also demonstrated bone resorbing

factor from synovial cell culture media but did notconsider this to be due to prostaglandin E2. To dateno measurement has been made of prostaglandin E2in the sera of patients with rheumatoid arthritis, butconceivably this, if it were raised, could cause boneresorption and hypercalcaemia. It is not knownwhether prostaglandin E2 could account for theother biochemical abnormalities of mineral meta-bolism observed in the present study, and clearlythis is worthy of further study. It is interesting tonote that sera from patients with rheumatoid arth-ritis have been shown to cause resorption of bonein tissue culture and that this effect was most markedin serum with high calcium concentrations (Kennedyet al., 1976).Another substance with PTH-like action is

osteoclast-activating factor, OAF (Horton et al.,1972), a substance which has been isolated fromlymphocytes. The effects of OAF on bone in vitroresemble those of PTH (Raisz et al., 1975), buteffects of injection of equipotent in vitro doses of the2 substances in vivo in rats differ in that PTHaffects serum calcium and phosphate while OAFdoes not, suggesting that OAF is a local rather thana systemic bone resorbing factor. This would alsoseem an obvious area for further investigation.Biochemical effects particularly regarding renalhandling of calcium and phosphate have, like PGE2,not been studied with OAF, and clearly requirefurther investigation.

In conclusion, the results of the present studyindicate that hypercalcaemia is a not uncommonfeature of rheumatoid arthritis. The hypercalcaemia

is associated with biochemical abnormalities sug-gestive of hyperparathyroidism, but serum con-centrations of immunoreactive parathyroid hormoneare normal. Since sera of patients with rheumatoidarthritis, especially those with hypercalcaemia,cause bone resorption in vitro (Kennedy et al., 1976)there seems little doubt that bone loss in rheumatoidarthritis is an active process. Finally, the fact thatcalcitonin and hydrocortisone block the resorptiveeffect of rheumatoid sera on bone in vitro (Kennedyand Lindsay, 1977) may have therapeutic possibilities.

We thank the technical staff of the University Department ofPathological Biochemistry, Glasgow Royal Infirmary, fortheir help. We wish to acknowledge the generous financialsupport of the National Fund for Research into CripplingDiseases, and the Arthritis and Rheumatism Council forResearch in Great Britain. One of us (A.C.K.) was in receiptof a Medical Research Council Clinical Research Fellow-ship.

References

Barnett, E., and Nordin, B. E. C. (1960). The radiologicaldiagnosis of osteoporosis-a new approach. ClinicalRadiology, 11, 166-174.

Barreras, R. F. (1973). Calcium and gastric secretion.Gastroenterology, 64, 1168-1184.

Berson, S. A., Yalow, R. S., Aurbach, G. D., and Potts,J. T. Jr. (1963). Immunoassay of bovine and humanparathyroid hormone. Proceedings ofthe National Academyof Sciences of the United States of America, 49, 613-617.

Bijvoet, 0. L. (1969). Relation to plasma phosphate con-centation to renal tubular reabsorption of phosphate.Clinical Science, 37, 23-36.

Bijvoet, 0. L., Morgan, D. B., and Fourman, P. (1969).The assessment of phosphate reabsorption. ClinicaChimica Acta, 26, 15-24.

Boehringer, Mannheim GMBH Diagnostica. Alkalinephosphatase optimised standard method kit. CatalogueNo. 15990.

Buchanan, K. D. (1973). Ph.D. thesis. The Queen's Universityof Belfast.

Buchanan, K. D., and McCarroll, A. M. (1971). In Radio-immunoassay Methods, p. 136. Edited by K. E. Kirkhamand W. M. Hunter. Edinburgh: Livingstone.

Conney, A. H. (1967). Pharmacological implications ofmicrosomal enzyme induction. Pharmacological Reviews,19, 317-366.

Coombes, R. C., Hillyard, C., Greenberg, P. B., and Mc-Intyre, I. (1974). Plasma immunoreactive calcitbonin inpatients with non-thyroid tumour. Lancet, 1, 1080-1083.

Fairney, A., Jackson, D., and Clayton, B. E. (1973). Measure-ment of serum parathyroid hormone with particularreference to some infants with hypocalcaemia. Archivesof Disease in Childhood, 48, 419-424.

Flanagan, R. W. J., Buchanan, K. D., and Murphy, R. F.(1974). Specificity of antibodies in the radioimmunoassayof glucagon. Diabetologia, 10, 365-369.

Goverde, B. C., and Veenkamp, F. J. (1972). Routine assayof total hydroxyproline based on resin catalysed hydrolysis.Clinica Chimica Acta, 41, 29-40.

Gray, T. K., and Munson, P. L. (1969). Thyrocalcitonin:evidence for physiological function. Science, 166, 512-513.

on July 7, 2020 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.38.5.401 on 1 O

ctober 1979. Dow

nloaded from

Hypercalcaemia in rheumatoid arthritis: investigation of its causes and implications 411

Haddad, J. G., and Chyu, K. G. (1971). Competitive proteinbinding radioassay for 25 hydroxycholecalciferol. Journalof Clinical Endocrinology and Metabolism, 33, 992-995.

Hahn, T. J., Birge, S. J., Scharp, C. R. and Avioli, L. V.(1972). Phenobarbital induced alterations in vitamin Dmetabolism. Journal of Clinical Investigation, 51, 741-748.

Hartmann, F., Rohde, J., Schmidt, A. (1969). Aktivitats-diagnostik bei der primar-chronischen polyarthritis.Zeitschrift fir Rheumatologie, 28, 263-266.

Heath, D. A. (1976). Hypercalcaemia and malignancy. Annalsof Clinical Biochemistry, 13, 555-561.

Hennessy, J. F., Gray, T. K., Cooper, C. W., and Ontjes,D. A. (1973). Stimulation of thyrocalcitonin secretion bypenetagastrin and calcium in 2 patients with medullarycarcinoma of the thyroid. Journal of Clinical Endocrinologyand Metabolism, 36, 200-203.

Higgs, G. A., Vane, J. R., Hart, F. O., and Wojulewski, J. A.(1974). The effects of anti-inflammatory drugs on prosta-glandin concentrations in synovial fluid from patients withrheumatoid arthritis. In Prostaglandin Synthetase pp. 165-171. Edited by H. J. Robinson and J. R. Vane. Raven Press:New York City.

Horton, J. E., Raisz, L. G., Simmons, H. A., Oppenheim,J. T., and Mergenhagen, S. E. (1972). Bone resorbingactivity in supernatant fluid cultured from humanperipheral blood leucocytes. Science, 177, 793-795.

Jones, J. H., Peters, D. K., Morgan, D. B., Coles, G. A., andMallick, N. P. (1967). Observations on calcium metabolismin the nephrotic syndrome. Quarterly Journal of Medicine,36, 301-320.

Kennedy, A. C., Allam, B. F., Boyle, I. T., Nuki, G., Rooney,P. J., and Buchanan, W. W. (1975). Abnormalities inmineral metabolism suggestive of parathyroid overactivityin rheumatoid arthritis. Current Medical Research andOpinion, 3, 345-358.

Kennedy, A. C., Lindsay, R., Buchanan, W. W., and Allam,B. F. (1976). Bone-resorbing activity in the sera of patientswith rheumatoid arthritis. Clinical Science and MolecularMedicine, 51, 205-207.

Kennedy, A. C., and Lindsay, R. (1977). Bone involvementin rheumatoid arthritis. Clinics in the Rheumatic Diseases,3, 403-420.

Krane, S. M. (1974). Degradation of collagen in connectivetissue diseases. Rheumatoid arthritis. In Dynamics ofConnective Tissue Macromolecules, 7th edn. McGraw Hill:New York.

Ladenson, J. H., and Bowers, G. N. (1973). Free calcium inserum II. Rigor of homeostatic control correlations withtotal serum calcium and review of data on patients withdisturbed calcium metabolism. Clinical Chemistry, 19,575-582.

McLaughlin, M., Fairney, A., Lester, E., Raggatt, P. R.,Brown, D. J., and Wills, M. R. (1974). Seasonal variationsin serum 25-hydroxycholecalciferol in healthy people.Lancet, 1, 536.

Maddison, P. J., and Bacon, P. A. (1974). Vitamin D defi-ciency, spontaneous fractures and osteopenia in rheu-matoid arthritis. British Medical Journal, 4, 433-435.

Marshall, R. W. (1976). In Calcium, Phosphate and Mag-nesium Metabolism, pp. 290-294. Edited by B. E. C.Nordin. Churchill-Livingstone: Edinburgh, London, andNew York.

Meyers, A. M., Laidley, L., Levine, E., Milne, F. J., Salant,D., and Kennedy, A. C. (1976). Ionized calciums in renaland other metabolic osteodystrophies. Communication tothe Association of Physicians of South Africa. The Uni-versity of the Orange Free State, Bloemfontein, SouthAfrica.

Nordin, B. E. C., Hodgkinson, A., and Peacock, M. (1967).The measurement and meaning of urinary calcium.Clinical Orthopaedics and Related Research, 52, 293-322.

Nordin, B. E. C., and Bulusu, L. (1968). A modified index ofphosphate excretion. Postgraduate Medical Journal, 44,93-97.

Nordin, B. E. C., Horsman, A., and Aaron, J. (1976). InCalcium, Phosphate and Magnesium Metabolism, pp. 471-473. Edited by B. E. C. Nordin. Churchill-Livingstone:Edinburgh, London, and New York.

Northover, B. J. (1973). Effect of anti-inflammatory drugs onthe binding of calcium to cellular membranes in varioushuman and guinea-pig tissues. British Journal of Pharma-cology, 48, 496-504.

Orrell, D. H. (1971). Albumin as an aid to th- interpretationof serum calcium. Clinica Chimica Acta, 35, 483-489.

Paterson, C. R. (1973). Tests of phosphate reabsorption.Lancet, 1, 1251.

Payne, R. B., Little, A. J., Williams, R. B., and Milner, J. R.(1973). Interpretation of serum calcium in patients withabnormal serum proteins. British Medical Journal, 4,643-646.

Pedersen, K. 0. (1970). Determination of calcium fractionsof serum. II. Ionized calcium in ultrafiltrates of normalserum and examination of pertinant variables. Scandi-navian Journal of Clinical and Laboratory Investigations,25, 223-230.

Pedersen, K. 0. (1972). Binding ofcalcium to serum albumin.II. Effects of pH via competitive hydrogen and calciumion binding to imidazole groups of albumin. ScandinavianJournal of Clinical and Laboratory Investigations, 29,75-83.

Raisz, L. G., Trummel, C. L., Mundy, G. R., and Luben,R. A. (1975). Immunological factors influencing boneresorption, role of osteoclast activating factor from humanlymphocytes and complement mediated prostaglandinsynthesis. In Calcium Regulating Hormones, Proceedingsof the Fifth Parathyroid Conference. Edited by R. V.Talmage, M. Owen, and J. A. Parsons. Excerpta Medica:Amsterdam.

Robertson, W. G., and Peacock, M. (1968). New techniquesfor the separation and measurement of the calciumfractions of normal human serum. Clinica Chimica Acta,20, 315-326.

Robertson, W. G. (1969). Measurement of ionized calcium inbiological fluids. Clinica Chimica Acta, 24, 149-157.

Robinson, D. R., Tashjian, A. H. Jr., and Levine, L. (1975).A possible mechanism for bone destruction in rheumatoidarthritis. Journal of clinical Investigation, 56, 1181-1188.

Rooney, P. J., Kennedy, A. C., Gray, G. H., Sturrock, R. D.,Buchanan, K. D., and Dick, W. C. (1976). Serum immuno-reactive gastrin (IRG) in rheumatoid arthritis: furtherobservations on its identity, clinical specificity and relation-ship to gastric function. Annals of the Rheumatic Diseases,35, 246-251.

Ropes, M. W., Bennet, G. A., Cobb, S., Jacox, R., and Jessar,A. R. (1959). Proposed diagnostic criteria for rheumatoidarthritis. Annals of Rheumatic Diseases, 18, 49.

Rose, G. A. (1972). A simple and rapid method for themeasurement of plasma ultrafiltrable and ionised calcium.Clinica Chimica Acta, 37, 343-349.

Rowe, D. J., and Stamp, T. C. (1974). Anticonvulsantosteomalacia and vitamin D. British Medical Journal, 1,392.

Ryden, S. E., Kirkish, L. S., and McCann, D. S. (1976).Evaluation ofserum ionic calcium measurement in a generalhospital population. American Journal of Clinical Patho-logy, 66, 634-638.

on July 7, 2020 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.38.5.401 on 1 O

ctober 1979. Dow

nloaded from

412 Kennedy, Allam, Rooney, Watson, Fairney, Buchanan, Hillyard

Schneider, A. B., and Sherwood, L. M. (1974). Calciumhomeostasis and the pathogenesis and management ofhypercalcaemic disorders. Metabolism, 23, 975-1007.

Sharp, J. T., Lidsky, M. D., Collins, L. C., and Moreland,J. (1971). Methods of scoring the progression of radio-logic changes in rheumatoid arthritis. Arthritis andRheumatism, 14, 706-720.

Simmons, C. J., Davis, M., Dordoni, B., and Williams, R.(1974). Urinary D-glucaric acid assay by an improvedenzymatic procedure. Clinica Chimica Acta, 51, 47-51.

Turbey, W. J., and Passaro, E. Jr. (1972). Hyperpara-thyroidism in the Zollinger-Ellison Syndrome. Archives ofSurgery, 105, 62-66.

Varghese, Z. (1973). Determination of plasma calciumfractions. Annals of Clinical Biochemistry, 10, 120-129.

Watson, M. (1975). Ph.D. thesis. University of Glasgow.Webb, J., Whaley, K., MacSween, R. N., Nuki, G., Dick,W. C., and Buchanan, W. W. (1975). Liver disease inrheumatoid arthritis and Sjogren's syndrome. Prospectivestudy using biochemical and serological markers ofhepatic dysfunction. Annals of the Rheumatic Diseases,34, 70-81.

Woodhead, J. S., and Walker, D. A. (1976). Assay of para-thyroid hormone in human serum and its use. Annals ofClinical Biochemistry, 13, 549-555.

Ziff, M., Kibrick, A., Dresner, E., and Giribetz, H. L. (1956).Excretion of hydroxyproline in patients with rheumaticand non-rheumatic diseases. Journal of Clinical Investi-gation, 35, 579-583.

on July 7, 2020 by guest. Protected by copyright.

http://ard.bmj.com

/A

nn Rheum

Dis: first published as 10.1136/ard.38.5.401 on 1 O

ctober 1979. Dow

nloaded from