Alkaline Phosphatase lsoenzyme Patterns in Malignant ... - CiteSeerX

2 downloads 124 Views 1MB Size Report
disease and liver or bone metastasis may increase their survival time. We have used the activity patterns of liver and bone isoenzymes of alkaline phosphatase.
CLIN. CHEM. 38/12, 2546-2551

Alkaline Viviane

0.

(1992)

Phosphatase

lsoenzyme

Van

T. Van

Hoof,’

Allan

Patterns

Oosterom,2

Early treatment of patients with malignant liver or bone metastasis may increase their

We

have

used

the

activity

of alkaline

isoenzymes

patterns

phosphatase

agarose gel electrophoresis, We studied ALP isoenzyme population

of 101

ease that might

with

disease survival liver

(ALP),

to detect patterns

patients

influence

of

Lucien and time.

and

bone

separated

by

early metastasis. in a background

no evidence

this pattern;

of any

a healthy

dis-

reference

population (n = 330); and the following three patients: 143 with malignant disease, 47 with nant liver disease, and 22 with nonmalignant

groups of nonmaligbone dis-

ease.

Cutoff and predictive values of liver ALP, highmolecular-mass (high-Me) ALP, and bone ALP were established for detecting liver and bone metastasis. The positive predictive value of liver and high-Mr ALP was higher than that of total ALP in detecting liver metastasis, but liver and high-Mr ALP did not enable us to differentiate between malignant and nonmalignant liver disease. Total

ALP

activity

was

of slightly

more

value

than

liver

and

high-M, ALP in enabling us to rule out liver metastasis. From bone ALP activity we could not distinguish between nonmalignant bone disease and bone metastasis. The negative predictive value of bone ALP in the diagnosis of bone metastasis was low, but its positive predictive was high and superior to that of total ALP. AddItIonal

Keyphrases:

cancer

liver

In patients with malignant improved with early treatment metastasis,

and

especially

and radiological ultrasound,

and

time-consuming.

CAT

metastasis,

procedures,

scans

Simple

metastasis

disease, survival (1 ). Symptoms

liver

diagnostic

bone

.

appear

may be of bone late,

such as x-rays,

are sensitive

laboratory

value

but

costly and for the in oncolog-

techniques

early detection of liver and bone metastasis ical patients are needed. Alkaline phosphatase (ALP, EC 3.1.3.1) is a membrane-bound enzyme that can be resolved into tissue nonspecific ALP (liver/bone type), placental ALP, and intestinal ALP.4 Hepatocellular and cholestatic diseases, such as acute and chronic hepatitis, cirrhosis, carcinoma of the liver, metastatic carcinoma of the liver, and acute or chronic biliary obstruction, are all associated with increased liver ALP activity (2) and, frequently, with high-molecular-mass (high-Me) M2 (3). High-Mr ALP is also known as fast-liver ALP (4), koinozyme (5, 6), and bile ALP (7). Bone ALP is mark1 Departments ogyfHypertension,

gium. 4Nonstandard high-Me ALP, characteristics. Received

2546

ofClimcal Antwerp

miry2 University

CLINICAL

12, 1992;

ALP,

accepted

CHEMISTRY,

NephrolAntwerp, Bel-

alkaline phosphatase; ALP; ROC, receiver-operator

abbreviations:

high-molecular-mass

May

Oncology, Hospital,

July

15, 1992.

Vol. 38, No. 12, 1992

in Malignant G. Lepoutre,’

Disease E. De Broe3

Marc

edly increased during growth (8) and in pathological conditions causing an increase in osteoblastic activity, e.g., healing fractures, Paget disease ofbone, osteoblastic metastases, and hyperparathyroidism (9). We have studied patterns of ALP isoenzyme activity obtained by agarose electrophoresis (10) in an attempt to establish the specificity, sensitivity, and predictive values ofliver ALP, high-Me M2, and bone ALP for detecting metastasis in patients with malignant disease. Materials

and Methods

Instruments

Total

ALP

activity

was determined

with

a Hitachi

705 automated analyzer (Boehringer, Mannheim, many). Electrophoresis was performed in Beckman trophoresis chambers and the gels were scanned computerized densitometer (Appraise; Beckman ments, Inc., Brea, CA).

Gerelec-

with

a

Instru-

Procedures Total ALP activity was determined by the method recommended by the Scandinavian Society for Clinical Chemistry (11 ), but at 25 #{176}C. Reagents for ALP were commercially available as a kit for which p-nitrophenyl phosphate was used as substrate and diethanolamine as a buffer (Baker Chemicals By, Deventer, The Nether-

lands). Serum samples were obtained by venipuncture. The samples were stored either at 4 #{176}C for short periods or at -85 #{176}C for long periods (reproducible for >3 years). Agarose electrophoresis of the ALP isoenzymes was performed with the Isopal system (Analis/Beckman, Namur, Belgium) as described previously (10).

Data Analysis Data

SPSS statistical Inc., Chicago, IL). Differences between disease-grouped data were studied by nonparametric tests (Kruskall-Wallis and median tests). The observed differences were significant to P = 0.05. Intervals were derived nonparametrically by computing the 0.05, 0.10, 0.25, 0.50, 0.75, 0.90, and 0.95 fractiles (see Figure 1). Cutoff values for the different ALP isoenzymes in the detection of metastasis were determined as follows (Figures 2-5). Two or three possible cutoffvalues were chosen for each ALP isoenzyme on the basis of the receiveroperator characteristics (ROC) curve (see Figures 2 and 4). Taking into account the different prevalences (prey.) of metastasis in the group of patients with malignant disease, we used the sensitivity (sens.) and specificity

package

(spec.)

for each

were

analyzed

for personal

of the test isoenzyme

by

use

computers

of the

(SPSS,

for the given cutoff value and with help of Bayes’

to calculate, theorem, the

isoenzyme

1400

Liver ALP

U/L

activity

was

below

the

cutoffvalue

(T

=

test

negative):

1200

spec.

1000

P(D1T)

-

prey.) (2)

spec. 800

(1

X

=

(1

X

-

prey.)

+ (1

-

sens.)

x prey.

8 0

For each

variable, the cutoff value that combined maximal accuracy with maximal P(D/T) and P(D1T) values was selected (see Figures 3 and 5). As an example, let us use an extreme cutoffvalue of 0 UIL for liver ALP: the sensitivity for detecting liver metastasis was 97% and the specificity was 1% [highest point of the ROC curve for liver ALP in Figure 1 (top panel)]; a cutoffvalue of330 UIL yielded a sensitivity of 18% with a specificity of 100% [lowest point of the ROC

600 400

_*1

200

L

0

_

C n-1O1 Background population

n-330 Reference popuIaon 800 WL

n-104 Liver mta stasis -

n-39

Liver stasis

+

curve for liver ALP in Figure 1 (top panel)]; and a cutoff value of 120 UIL combined a sensitivity of 67% with a specificity of 88%. The prevalence of liver metastasis being 27% in this population, the post-test probabilities were calculated as follows:

ALP

High-Mr

700

n-47 Non-malignant ever disease

600

500

0 0

400

0.67

300

P(D,T)

0.67

200

x

0.27

(3)

=

x

0.27

+

(1

0.88)

-

x (1

-

0.27)

0

100

0

L n-330 Reterence poputaon

1

t

0

LI

n-101 Background poputaon

:

L]

n-47 Non.malignant Dyer disease

P(D7T) n-

+

u/i-

-

1000 0

800 600

200

[J

611 n_330

r-1O1

Referenc

n-fl

Background population

populatbn

Description

0

n-47

tion-margnant bone disease

n-96

Bone meta. stasis +

Bone rneta-

stasis -

Fig. 1 . Distribution healthy reference Upperandlowerlines

of liver ALP, high-Mi ALP, and bone ALP in the population and in the different patient groups indicate the 10th and the 90th percentile, boxesthe 25th and 75th percentile, the ilne in the boxthe 50th percentile, and circles indicate outliers

probability (P) that a patient had metastasis (D = diagnosis present) when the ALP isoenzyme activity was above the cutoff value (T = test result positive): sens.

X

prey.

P(D’7T)= sens.

X prey.

and the probability metastasis

(D

=

+ (1

(P) that diagnosis

-

spec.)

(4) x

0.27

Thus, when the test is positive, the probability that the patient in this population has liver metastasis is 65%; when the test is negative, the probability that the patient does not have liver metastasis is 88%. The curve for liver ALP in Figure 2 (top panel) was obtained by changing the values for the prevalence ofliver metastasis from 0% to 100% in equations 3 and 4. By this means we could judge the discriminating value of a test: the flatter the curves, the more insensitive the test, and the steeper the curves, the more sensitive the test.

1200

400

-

104

stasis

ALP

Bone

0.88 x (1 0.27) 0.27) + (1 - 0.67) -

x (1

Uver meta-

stasis 0

1400

=

0.88

rt-39 Uver

x (1

-

this patient did absent) when

prey.) not the

(1)

have ALP

of the Populations

We studied results for a total of 330 healthy adults and 313 patients for whom total ALP activity and ALP isoenzyme activity had been routinely determined between November 1989 and February 1990. Patients’ records were analyzed retrospectively. Healthy referencepopulation (n = 330). We previously studied the ALP isoenzyme patterns in 330 healthy adults. The age and sex distribution of this population was described earlier (12). Background population (n = 101). As a background population we used a group of patients (ages 55 ± 33 years) without any ofthe conditions, such as lung embolism and chronic renal failure, known to cause an increase in total ALP or a change in ALP isoenzyme pattern. Patients with osteoporosis were not excluded (n = 13), because this condition is common in this age group and was usually not the only health problem. This population included 57 men and 44 women treated in 11 different hospital and outpatient departments. Their roeCLINICAL

CHEMISTRY,

Vol. 38, No. 12, 1992

2547

a

LIVER

METASTASIS

+1-

The most breast (n

common sites of the primary tumors were 36), lung (n = 30), and colonlrectum/sigmoid (n = 12), followed by ovary (n = 9), prostate (n 8), kidney (n = 6), skin (melanoma and spinocellular carcinoma; n = 6), testis (n = 3), non-Hodgkins lymphoma (n = 3), bone (n = 3), bladder (n = 2), and other diverse sites (n = 8). No primary tumor was found in 3 patients; 6 had multiple tumors. Evidence ofliver metastasis was obtained by liver echography or liver scan in 39 of these

100

80

-60 >‘ > U)

40 (I)

patients; 47 had bone metastases shown by x-ray or total body scan; and 19 patients had both liver and bone metastases. Some patients with liver or bone metastasis also had lung, brain, or other metastatic sites.

20

I 00-Specificity (%) +1- (excludIng

b

LIVER METASTASIS

lung

tumors)

Results

100

The distribution and the percentiles zymes in the reference population patient groups are shown in Figure

80

Background

-60 >

40 0

C,)

20

20

60

40

2. Receiver-operator

characteristics

80

100

(%)

100-Specificity

(ROC)

curves

fortotal

and 1.

ofthe ALP isoenin the different

Population

Total ALP, liver ALP, and high-Me M.2 activities were significantly increased in the background population compared with the healthy reference population. The distribution of bone ALP was greater in the background population than in the healthy reference population: both lower and higher activities were encountered. Only 1 of the 13 patients with osteoporosis had high concentrations of bone ALP (112 U/L). Of the 41

U)

Fig.

=

ALP,

liver ALP, and high-M, ALP in the detection

of liver metastasis 39) was tested against the

The liver metastasis positive group (n = liver metastasis negative group (n = 104) in (a) the group ofpatients with malignant disease and (b) the group with malignant disease but with exclusion of the patients with lung tumors (31 patients, 7 of whom had liver metastasis)

patients

with

trations ofliver with the healthy

respiratory problems, ALP or high-Me M, reference population.

the intensive care department was days and had increased placental

9 had high concenor both, compared One patient from intubated for several ALP, detectable by

electrophoresis. ords contained no evidence of liver, bone, or malignant disease. Most of the patients (n 41) had lung disease (chronic obstructive lung disease, asthma, pneumonia, sarcoidosis); 12 patients had cardiac disease (myocardial infarction, cor pulmonale, angina pectoris, pericarditis), 7 patients had diabetes, 3 had urolithiasis, and 6 were severely ill (sepsis, viral encephalitis, Leyll’s syndrome, adult respiratory distress syndrome); 49 ofthese patients had diverse and multiple health problems. Patients with nonmalignant liver disease (n = 47). This group included 23 men and 24 women (ages 58 ± 20 years). Chole(doco)lithiasis was present in 16 of these patients; 6 had liver steatosis; 8 had toxic and 2 had chronic viral hepatitis; 4 had primary biliary cirrhosis; 6 had several other types ofhepatic cirrhosis; and 5 had other, different, liver problems. Patients with nonmalignant bone disease (n = 22). Nine men and 13 women (ages 54 ± 15 years) composed this group. Secondary hyperparathyroidism was diagnosed in 15 patients (associated with chronic renal failure, post-renal transplant, or hyperthyroidism); 5 had Paget disease ofbone; 1 had primary hyperparathyroidism; and 1 had myositis ossificans. Patients with malignant disease (n = 143). This group included 72 men and 71 women (ages 58 ± 14 years). 2548

CLINICAL

CHEMISTRY,

Vol. 38, No. 12, 1992

Nonmalignant

Liver Disease

Liver ALP and high-Me M were significantly increased in these patients compared with the background population. Bone ALP was also higher in the group with liver disease. None of the isoenzymes could be used to differentiate between the patients with liver disease and the patients with both malignant disease and liver metastasis.

Nonmalignant

Bone Disease

Total and bone ALP were higher in these patients than in the background population. None of the isoenzymes could be used to distinguish between the group with bone disease and the patients with both malignant disease and bone metastasis. Patients

with

Malignant

Disease

Patients without liver metastasis. Concentrations of liver and high-Me M2 did not enable us to distinguish between the group of patients without liver metastasis and the background population. Patients with liver metastasis. Total ALP, liver ALP, and high-Me MY were significantly higher in the patients with liver metastasis compared with both the

a

+1-

METASTASIS

TotaIALP>

17011/I.

90

12011/1.

LiverALP Liver

.70

ALP

>

and Hh-MALP

92 U/I. >61111.

60 High-Mr

MY

>

10 U/I.

50 40 30

2 10 0

10

20

30

40

50

60

70

80

90

100

PRE-Test Probability (%)

b

UVER

LUNG

100

TUMORS)

Total ALP >.170 U/I.

90

12011/1.

LIVerALP> j80 liver ALP andHigh-MALP

.70 60

Hh-Mr

>

ALP

92 U/I. >611/1. >

10 U/i.

50 40 I1-30 C’)

H#{149}

10 0 0

10

20

30

40

PRE-Test

50

60

Probability

70

80

90

of 70%,

100

(%)

Fig. 3. Graphic representation of the probability that a patient with malignant disease has liver metastasis if total ALP, liver ALP, or high-Mr ALP, or the combination of liver and high-Mr ALP, are above cutoff values [P(D7T4)]. Also shown is the probability that this patient does not have liver metastasis if these variables are below their cutoff values [P(DiT)], taking into account different prevalences of liver metastasis in the background population with malignant disease (Bayes’ theorem), in (a) the whole population with malignant disease, and (b) the whole population with malignant disease but excluding patlents with lung tumors

background population and the patients without liver metastasis. Of the 19 patients with both liver and bone metastasis, 6 had higher than normal bone ALP activity. Only one of the patients with liver metastasis without bone invasion had a high bone ALP activity.

ROC curves were calculated and used to determine the relation between true-positive and false-positive rates for the detection of liver metastasis in oncological patients. Different cutoff values were used for each of the following isoenzymes: total ALP, liver ALP, and high-Mr ALP. As shown in Figure 2, the discriminating value of the isoenzymes was slightly better when lung cancer patients were excluded. When we used a cutoffvalue of 120 UIL for liver ALP, the sensitivity for detecting liver metastasis was 67%, with a specificity of 88% and an accuracy of 83%. When we used a cutoff value of 10 U/L for high-Me A12, the sensitivity was 69%, the specificity was 88%, and the accuracy was 83%. When lung tumors were excluded, these values were 65%, 93%, and 85% for liver ALP and 68%, 91%, and 85% for high-Mi ALP, respectively. When the results ofliver ALP and high-Mr ALP were combined, the cutoff values could be lowered to 92 U/L for liver ALP and to 6 U/L for high-Me ALP. When both

were specificity for the

and 70%, respectively, and the P(D/T) were 88%, 88%, and 89%, respectively. When with lung carcinoma were not included, the incidence of liver metastasis and the P(D1T) values remained the same, whereas the P(DPFi values improved to 77%, 75%, and 73%, respectively. Patients without bone metastasis. The activity of bone ALP did not enable us to distinguish between the 68%,

+1- (EXCLUDING

METASTASIS

above these cutoff levels, a sensitivity of 88%, and accuracy of 84% were obtained whole group of patients. These values increased to 71% for sensitivity, 90% for specificity, and 85% for accuracy when patients with lung carcinoma were excluded. It can be seen in Figure 3 that, with a pretest probability of27% (the prevalence ofliver metastasis in the population studied) and with cutoff values of 120 UIL for liver ALP, 10 UIL for high-Me M2, and >92 U/L and >6 UIL for the combination of liver ALP and high-Me A’Y’ respectively, the P(D/T) values were isoenzymes

100

68%,

values patients

patients without bone metastasis and the background population. Patients with bone metastasis. Total ALP, liver ALP, and high-Me M2 were higher in the patients with bone metastasis compared to the background population. Bone ALP activity did not differ between these groups. Total and liver ALP were higher in the group of patients with bone metastasis than in the group of patients without such metastasis. The difference in bone ALP activity between these groups was not significant. Fifteen of the 19 patients with both bone and liver metastasis had increased liver or high-Me A1P activities, or both, explaining the significantly higher liver and high-Me activities in patients with bone metastasis. However, 10 ofthe 28 patients with bone metastasis but without evidence ofliver metastasis had increased liver or high-Me A’ activities, or both, and 6 of these 10 patients had lung carcinoma. ROC curves were calculated for bone ALP and used in the detection ofbone metastasis (Figure 4). For a cutoff value of 90 UIL, the sensitivity for detecting bone metastasis was only 33% for a specificity of 97%; the accuracy of the test was 76%. As shown in Figure 5, a pretest probability of 33% (the prevalence of bone metastasis in the population studied) and a cutoff value of 90 U/L yields 84% for P(DITi and 74% for P(D1T). Discussion The technique we describe here for the detection of bone and liver metastasis in cancer patients is sensitive and reproducible and poses no major technical problems. The agarose gels are precast and the reagents are ready for use. Native serum and serum treated for a short period (5 mm) with a polyclonal antiserum that reacted with placental ALP and intestinal ALP are usually run side by side. When bone ALP exceeds 50% of the total ALP activity, neuraminidase must be added to enhance the separation ofliver and bone ALP (10, 12). The whole procedure, including scanning of the gels, takes about 3 h for 10 samples. The cost of the test is low compared CLINICAL

CHEMISTRY,

Vol. 38, No. 12, 1992

2549

BONE

METASTASIS

+1BONE

100

METASTASIS

+1-

100

Total ALP

170

>

11/1.

90 80

Bone

ALP

>

90 U/I.

,70 60 60 >

50 U)

Bone

ALP

40 I-

40 C/)

I-

30

(I)

20

20

10

0 20

40

1 00-Specificity

Fig. 4. Receiver-operator

in the detection

0

100

80

60

(ROC)

curve

of bone ALP

was tested against the bone

47)

with the costs of radiographic and scintigraphic procedures and various other tests, such as CEA dosage with radioisotopes (which in Belgium costs about twice as much as ALP electrophoresis). The use of this method for determining the distribution of the different ALP isoenzymes according to age and sex in a healthy population was described preyously (12). In the present study we found that, compared with the previous healthy reference population, a background population ofpatients with no evidence for liver or bone disease had nonspecific increases in liver, highM, and bone ALP. This justified our use of this background population in the present study ofisoenzymes in malignant disease. Electrophoresis of a serum sample from one of the patients in this background population revealed a placental ALP fraction. With the Isopal system, the major placental ALP fraction comigrates with bone ALP (10); it is differentiated from the intestinal variant ALP fraction by its sensitivity to neuramimdase and its resistance to heat (65 #{176}C, 10 mm). In this patient, the

placental

nature

by its reaction ALP. Specific

ofthe

with dosage

abnormal

a monoclonal of placental

fraction

was confirmed

antibody to placental ALP determined with

an enzyme-linked

immunosorbent assay (more sensitive electrophoretic procedure) was not routinely performed in this study. The finding ofplacental ALP in a patient on artificial respiration and with no signs of malignancy is not surprising, because increased serum concentrations of this isoenzyme are found in several lung diseases. Such increases are thought to be caused by an increase in alveolar epithelial and endothelial permeability with leaking of placental ALP, produced by type I pneumocytes, into the circulation (13, 14). Bone ALP was increased in the group with liver disease. This might be explained by inadequate intestithan

nal biliary

the

absorption obstruction

of vitamin

D, in

or cirrhosis,

mm D metabolism of the liver, opment of osteomalacia (15). 2550

CLINICAL

CHEMISTRY,

30

PRE-Test

of bone metastasis =

20

40

50

60

Probability

70

80

90

100

(%)

(%)

characteristics

The bone metastasis positive group (n metastasis negative group (n = 96)

10

cases

or by

an

resulting

of prolonged impaired

vita-

in the devel-

Vol. 38, No. 12, 1992

Fig. 5. Graphic

representation

of the probability

that a patient

with

malignant disease has bone metastasis if total ALP or bone ALP is above the respective cutoff values [P(D/T)], and the probabllfty that this patient does not have bone metastasis if these variables are below their cutoff values [P(D1T)], taking into account different prevalences of bone metastasis in the background population with

malignant

disease

The bone metastasis negative group

(Bayes’ theorem) positive group was tested against the bone metastasis

Because

serum samples are taken at regular intervals the follow-up of all patients with malignant disease in our hospital, and because we did not select the patients but processed the results from consecutive routine samples, the 143 oncological patients we studied can be considered representative for the population of the oncological department in our hospital. A review of the medical records for this population showed that the prevalences ofliver and bone metastases were 27% and during

33%,

respectively.

Liver

Metastasis

High-Me

consists of liver ALP attached to fragcell membrane; variously sized vesides are visible in these fragments when examined in the electron microscope (6). High-Me M1 is present in the serum of patients with well-defined biliary obstruction (5, 16, 1 7) and in patients with hepatic metastasis of solid tumors (18, 19). Viot et al. (20) studied 202 patients with different malignant conditions by cellulose acetate electrophoresis. These investigators concluded that high-Me M was highly correlated with the presence ofliver metastasis and that assays ofthis ALP isoform provided greater specificity (90%) and sensitivity (97%) than those of -y-glutamyltransferase and total ALP; however, they did not include a large variety of benign pathologies in their study. Nishio et al. (18) examined the diagnostic value of high-Me M2 separated by cellulose acetate electrophoresis in 126 lung cancer patients and 15 controls with benign respiratory diseases, and concluded that this isoenzyme was most useful in patients with small cell lung cancer, which often shows widespread hepatic metastasis. They found a sensitivity of 71%, a specificity of 89%, and an accuracy of 86% for high-Me ALP. Mayne et al. (21 ) studied total, liver, and high-Me M2 in 140 patients with ments

A’

of the liver

breast cancer. They found lower sensitivities than the other authors, but, as might be expected, higher specificities and an accuracy of85% for total ALP and of 88% and 86% for liver and high-Mr ALP, respectively. They attributed the low sensitivity of the isoenzymes to the diagnosis of liver metastasis occurring at an earlier stage compared to other studies (more false negatives).

The

sensitivity,

specificity,

and accuracy

of high-Me

ALP high (18).

for detecting metastasis in our study were not as as those found by Viot et al. (20) and Nishio et al. High-Mr ALP was definitely better than total ALP and only slightly better than liver ALP for detecting liver metastasis. Lower cutoff values could improve sensitivity but also resulted in a lower accuracy and straighter P(DiT)IP(D7T) curves, parameters that were not studied by the other authors. Positive predictive values of liver and high-Me M.2 increased slightly when lung cancer patients were excluded. This might be

explained

either

by the presence

ofliver

metastasis

not

detected by echographic and scintigraphic procedures or by nonspecific increases in liver and high-Me ALP in lung cancer without liver metastasis. The presence of increased liver and high-Me M2 activities in some patients with benign lung disease (see background population) supports the latter hypothesis. Although combining the activities of liver and high-Me M2 yielded the best results for detecting liver metastasis, total ALP proved to be more useful for ruling out liver metastasis. Bone

Metastasis

in total, liver, and high-Mr M2 bewith and without bone metastasis were the presence ofliver metastasis in 19 ofthe patients group that was positive for bone metastasis. Determination of bone ALP activity did not differ significantly between the group that was positive for bone metastasis and the group that was negative for this condition. Because bone ALP is an indicator of osteoblastic activity, it is unlikely to increase in cases of osteolytic bone metastasis (e.g., most breast tumors). Thus, a normal bone ALP activity has only a very limited value in ruling out bone metastasis, whereas increased bone ALP activity is of value in detecting osteoblastic metastasis. None ofthe ALP isoenzymes enabled us to distinguish between benign liver disease and liver metastasis or between benign bone disease and bone metastasis. Therefore, when liver, high-Me, or bone ALP is increased without obvious cause, malignancy should be ruled out by further diagnostic tests. Once malignancy has been diagnosed, determination of ALP isoenzyme patterns can provide a noninvasive test for monitoring a patient between radiographic and scintigraphic procedures. An increase in the activity of liver, high-Me, or bone ALP above the aforementioned cutoff values in a patient with malignant disease without previous signs of metastasis should also indicate the need for further evaluation of the patient. The tween due to in the

differences the patients

We thank all the technicians from the clinical chemistry department of our hospital, and especially Marina Van Mullem, for excellent technical assistance. We are also indebted to the heads and clinicians from the departments ofendocrinology, hematology, physiotherapy, gastroenterology, intensive care, and pneumology who allowed us, and eventually helped us, to go through their patients’ records. We are especially grateful to Dirk De Weerdt of the department ofnephrology/hypertension forlayout ofthe figures. References 1. De Vita VT, Hellinann S, Roeanberg SA, eds. Cancer. Principles and practiceof oncology, 3rd ad. Philadelphia: Lippincott, 1990. 2. Rhone DP, Mizuno FM. Profiles of alkaline phosphatase isoenzymes in serum using cellulose acetate electrophoresis and organspecific inhibitors. Am J Clin Pathol 1973;59:531-41. 3. Jennings RC, Brocklehurst D, Hiret M. A comparative study of alkaline phosphatase enzymes using starch-gel electrophoresis and Sephadex gel-filtration with special reference to high moleciilax weight enzymes. Clin Chim Acta 1970;30:509-17. 4. Moss DW. Alkaline phosphatase isoenzymes [ Review]. Clin Chem 1982;28:2007-16. 5. De Broe ME, Borgers M, Wieme RJ. The separation and characterization ofliver plasma membrane fragments circulating in the blood ofpatients with cholestasis. Clin Chim Acta 1975;59: 369-72. 6. De Broe ME, Roels F, Nouwen EJ, Wieme EJ. Liver plasma membrane: the source of serum high molecular weight alkaline phosphatase in human serum. Hepatology 1985;5:118-28. 7. Price CP, Sammons HG. The nature of the serum alkaline phosphatase in liver disease. 8. Clark LC, Beck A. Plasma

Normative 9. Epstein alkaline 10. Van Broe ME.

J Clin Pathol 197427:392-8. “alkaline” phosphatase activity. I. data for growing children. J Pediatr 1950;36:335-41. E, Kiechie FL, Artiss JD, Zak B. The clinical use of phosphat.ase enzymes. Clin Lab Med 1986;84:695-8. Hoof VO, Lepoutre LG, Hoylaerts MF, Chevign R, De Improved agarose electrophoretic method for separating phosphatase isoenzymes in serum. Clin Chem 1988;34:

alkaline 1857-62. 11. Scandinavian

Society for Clinical Chemistry and Clinical Physiology. Recommended methods for the determination of four enzymes in blood. Scand J Clin Lab Invest 1974;33:291-303. 12. Van Hoof VO, Hoylaerth MF, Geryl H, Van Mullem M, Lepoutre LG, De Broe ME. Age and sex distribution of alkaline phosphatase isoenzymes by agarose electrophoresis. Clin Chem

1990;36:875-8. Eestermans G, Nouwen EJ, Demey HE, Bosaaert LL, De Backer W, De Broe ME. Human placental alkaline phosphatase and lung iijury. Chest 1987;92:961. 14. Nouwen EJ, Buysaens N, De Bros ME. Heat stable alkaline phosphatase as a marker for human and monkey type-I pneumocytes. Cell Tissue Res 1990260:321-35. 15. Diamond TH, Stiel D, Lunzer M, McDowall D, Eckstein RP, Posen S. Hepatic osteodystrophy. Gastroenterologjr 1989;96:213-21 16. Brocklehurst D, Wilde CE, Doar JWH. The incidence and likely origins of serum particulate alkaline phosphatase and lipoprotein-X in liver disease. Clin Chim Acts 1978;88:509-15. 17. Crofton PM, Elton RA, Smith AF. High molecular weight alkaline phosphatase: a clinical study. Cliii Chim Acta 1979;98: 13.

263-75. 18. Nishio H, Sakuma T, Nakamura SI, Horai T, Ikegami H, Matsuda M. Diagnostic value of high molecular weight alkaline phosphatase in detection of hepatic metastasis in patients with lung cancer. Cancer 1986;57:1815-9. 19. Viot M, Thyss A, Viot G, Ramaioli A, Cambon P, Schneider M, Lalanne CM. Comparative study of gamma glutamyl transferase, alkaline phosphatase and its 1 isoenzyme as biological indicators ofliver metastases. Clin Chim Acta 1981;115:349-58. 20. Viot M, Thyss A, Schneider M, Viot G, Ramaioli A, Cambon P, Lalanne C. Alpha-i isoenzyme of alkaline phosphatases. Clinical importance and value for the detection ofliver metastases. Cancer

1983;52:140-5. 21. Mayne

Identification measurement

J Clin Pathol

Thakrar 5, Rosalki SB, Foe AY, Parbhoo S. of bone and liver metastases from breast cancer by ofplasma alkaline phosphatase isoenzyme activity. 1987;40:398-403.

PD,

CLINICAL

CHEMISTRY,

Vol. 38, No. 12, 1992

2551