Adrenocortical Carcinoma and Hypoglycemia: Evidence for Production of Nonsuppressible Insulin-Like Activity by the Tumor

T. HYODO,* K. MEGYESI, tt C. R. KAHN, #| J. P. MCLEAN,§ AND H. G. FRIESEN*

* Department of Physiology, University of Manitoba, Guest Worker, Semmelweis Medical University, Budapest, Hungary, Diabetes Branch, National Institute of Arthritis, Metabolism, and Digestive Diseases, National Institutes of Health, Bethesda, Maryland 20014, and §St. Boniface Hospital, Manitoba, Canada

ABSTRACT. Serum and tumor extracts from a pa- tient with adrenocortical carcinoma and hypogly- cemia were studied using conventional bioassay and radioreceptor assays. Immunoreactive insulin in all sera and extracts was low or undetectable. Increased insulin-like activity (about 2 x normal) was detected in the insulin radioreceptor assay and in the fat cell bioassay, suggesting the pres- ence of some other insulin-like material in blood. An increased level of NSILA-s related peptides was even more prominent (4 x normal) when the more specific NSILA-s/MSA radioreceptor assay was employed. Delineation of the tumor as the proba- ble source of the NSILA-s related peptides was highly dependent on studying the proper type of extract. Acetone-ethanol extracts of tumor were 4- to 20-fold more active in the NSILA-s/MSA receptor

assay than unaffected tissues; these elevated levels of NSILA-s related peptides were confirmed by in- sulin radioreceptor assay and fat cell bioassay. Acid-ethanol extracts showed the same general pat- tern, but due to the much lower activity material obtained by this method, it was detectable only in the more sensitive NSILA-s/MSA receptor assay. The great variability observed among extraction methods and the lack of specific and sensitive assays in the past may have accounted for the dif- ficulty in detecting this insulin-like peptide in previ- ously reported patients. These data again suggest that in some patients with tumor hypoglycemia, elevated levels of NSILA-s related peptides may play an important pathogenetic role and that these pep- tides are produced by the tumor itself. (J Clin Endocrinol Metab 44: 1175, 1977)

S INCE the initial report of Doege in 1930 (1) over 200 cases of hypoglycemia associated with extrapancreatic tumor have been reported (2). The types of tumor as- sociated with this syndrome have been of a wide variety; most frequently, however, this syndrome occurs with mesenchymal tumors, hepatomas, and adrenocortical car- cinomas (2,3). Several mechanisms have been suggested as being responsible for the hypoglycemia. Recently we presented evi- dence that in about one-third of patients with extrapancreatic tumors and hypo- glycemia the plasma concentration of an insulin-like peptide, NSILA-s1 or some

closely related peptide, was elevated (4). NSILA-s is a serum peptide with molecu- lar weight of about 6000 that has insulin- like activity in vitro and in vivo and can- not be suppressed by anti-insulin antibodies (5). We suggested that the hypoglycemia in these patients was due to the elevated NSILA-s or a similar peptide. However, the source of the NSILA-s related peptides in these patients, as well as their source in normals, is unknown.

In the present study we report a patient with adrenocortical carcinoma and hypogly- cemia. In this case we were able to demonstrate an elevated concentration of NSILA-s related peptides not only in the serum, but also in tumor tissue extracts, suggesting that these peptides are being syn- thesized in the tumor itself. Further we find that the demonstration of NSILA-s- like activity in the tumor is highly de- pendent on the method of extraction, and this finding may account for part of the difficulty in demonstrating NSILA-s or other

Received October 19, 1976.

Supported by the Medical Research Council of Can- ada and the National Institutes of Health.

| To whom reprint requests should be addressed.

’ Abbreviations used in this paper are NSILA-s, nonsuppressible insulin-like activity soluble in acid ethanol; ILA, insulin-like activity; MSA, multiplication stimulating activity; KRP, Krebs-Ringer Phosphate buffer; BSA, bovine serum albumin; KRB-A, Krebs Ringer Bicarbonate with 3% BSA.

insulin-like peptides in tumor extracts in previous studies.

Case Report

In 1971 a 42-year old man presented with ab- dominal pain and a mass in the right upper quadrant. At laparatomy a large tumor adherent to the inferior vena cava, which could not be removed, was found. Biopsy was consistent with adrenocortical carcinoma, although urinary levels of 17-hydroxy- and 17-ketosteroids were within normal levels. Radiation therapy and treatment with cytotoxic drugs did not decrease the size of the tumor, and in 1971 an ileo- transverse colostomy was performed to bypass an obstruction caused by the tumor. In 1974, the tumor size further increased, and chemo- therapy was discontinued. In 1975, the patient presented with hypoglycemia for the first time; his blood sugar was 28 mg/dl. Plasma immuno- reactive insulin levels were normal. Growth hor- mone levels were low, despite the hypoglycemia. Attempts to raise blood glucose with diazoxide or glucagon failed. The patient died after a hypo- glycemic episode. At autopsy a 2000 g right adrenal tumor and a 3000 g liver (75% of it occupied with metastases) were found.

Materials and Methods

Serum was stored at -20 C until assay. Tissues were extracted using 5 different methods: 1) acid-ethanol extraction as described by Grodsky et al. (6), 2) ethanol-acetone extraction as described by Jacob et al. (7), 3) extraction with 0.1M ammonium bicarbonate (pH 8.2), 4) extrac- tion with 10% acetic acid (pH 2.3), and 5) extraction with distilled water. In the present study we present results only with the first two since no insulin-like activity could be de- tected by insulin radioreceptor assay in the latter three extracts.

Insulin radioreceptor assay

125 I-porcine insulin was prepared using a modi- fication of the chloramine-T method (8) and puri- fied by gel filtration (Sephadex G-50 column, 1.9 x 43 cm, 25 mM Tris HCI pH 7.6 as buffer). The specific activity of 125I-insulin ranged be- tween 80-100 uCi/ug and was stable for two months stored at -20 C. Membranes from late

pregnant (50-55 days of gestation) guinea pig kidneys were prepared using the method previ- ously employed by Tsushima and Friesen for rabbit liver (9).

Serum samples were acidified and fractionated on a Sephadex G-50 column (0.9 x 55 cm) in IM acetic acid pH 2.3 with a 125I-insulin marker. All fractions were neutralized using Tris buffer (pH 10.7) to give a final pH of 7.6.

For the assay, 125I-insulin, guinea pig mem- brane (100 µg protein/tube) and insulin standard or unknown fraction were incubated at 4 C for 20 h in a 0.6 ml total incubation volume. The incubation was performed in 25 mM Tris HCI buffer at pH 7.6 containing 10 mM MgCl2 and 0.1% bovine serum albumin (BSA). After incubation 3 ml Tris-BSA buffer was added, and the bound and free 125I-insulin were counted in an autogamma counter.

As in the insulin radioreceptor assays using human lymphocytes and rat liver membranes (10), insulin analogues competed for 125I-insulin binding to guinea pig kidney membranes in direct proportion to their biological potency in vitro (Fig. 1). Serial dilution of the fraction from the serum gel filtration at a place corres- ponding to the peak of insulin-like activity showed a dilution curve similar to the porcine insulin standard (Fig. 2). The intra-assay coef- ficient of variation of serum fractions was 30% at low concentrations and 14% at high concen- trations of insulin-like activity.

The NSILA-s/MSA radioreceptor assay

Plasma NSILA-s and related peptides were measured by a modification of the radioreceptor assay we have previously described (4,11). In the present study we used multiplication stimu- lating activity (MSA), generously provided by Drs. S. P. Nissley and M. M. Rechler of the National Institutes of Health, as tracer and stand- ard.

MSA was purified from the culture medium of buffalo rat liver cells (12) and iodinated by a modification of the chloramine T method to specific activities of 80-150 uCi/ug (11). The MSA used in these experiments had a specific insulin-like activity of about 100 mU/ mg as measured in the glucose oxidation bio- assay (data not shown). Highly purified rat liver membranes were used as a source of re- ceptors in this study (13). Incubations were

FIG. 1. The insulin radioreceptor assay. 125I-insulin was incubated with guinea pig kidney mem- branes and unlabeled peptides indicated in the figures as de- scribed in Materials and Methods. At the end of the incu- bation the insulin bound to the membranes was separated by centrifugation and the radioac- tivity counted. The 125I-insulin bound (cpm) was plotted as a function of the concentration of unlabeled peptide added.

12

10

O - Prl

125; - INSULIN BOUND (x103 CPM)

h - PL

h - GH

Pork Insulin

P - Glucagon

8

Human Bovine Ovine Rabbit Rat

6

Insulin

Guinea

Pork Proinsulin

Pig

Insulin

4

2

Chicken Insulin

1

10

102

103

104

HORMONE CONCENTRATION (ng/ml)

conducted at 20-22 C for 3 h Krebs-Ringer phos- phate buffer, pH 7.4 as previously described (4,14).

We have previously characterized in some de- tail the specificity of this assay using a 70 mU/ mg preparation of NSILA-s kindly provided by Drs. E. R. Froesch and R. E. Humbel as a tracer and standard (11). We found that MSA and somatomedin A competed for binding to the receptor, whereas insulin, proinsulin, epidermal growth factor, nerve growth factor, fibroblast growth factor and somatomedin B did not. The only available preparation of somatomedin C (2000 U/mg, kindly provided by Ray Hintz) also showed little or no competition for the labeled NSILA-s binding at concentrations up to 1 ug/ml. We also showed that MSA could be used as the labeled peptide in this assay in an interchangeable fashion with NSILA-s.

Since MSA has been more readily available to us we have attempted to characterize further this receptor assay using the MSA tracer and standard. We have reassayed a number of plasma samples and have found a close correlation between the results obtained with MSA and those obtained with NSILA-s standard and tracer (data not shown). Furthermore, using MSA as a tracer, Rechler, Nissley, and co-workers have confirmed and extended our observations on the specificity of this assay (15,16). It appears that the liver membrane receptor assay, whether using MSA or NSILA-s, shows more specificity than similar

assays using either human or chick embryo fibroblasts (15-17). MSA, NSILA-s, and somatom- edin A, but not insulin, compete with labeled MSA for the liver membrane receptor. Although we have not been able to obtain any purified somatomedin C for direct measurement in the liver membrane assay since converting to MSA standard and tracer, as in our previous studies using labeled and unlabeled NSILA-s, we con- tinue to find normal levels of NSILA-s related peptides by this receptor assay in acromegalic sera, even when plasma somatomedin C is elevated as measured by the placental membrane radioreceptor assay.2 Thus, at present there is no direct or indirect data to suggest that this assay measures somatomedin C. For the sake of simplicity we have referred to everything which is measured in this receptor assay as NSILA-s, or NSILA-s related peptides. Whether the activity measured in this assay corresponds to the recently purified (18) peptides termed NSILA-I and NSILA-II is unknown.

Isolated fat cell bioassay and insulin immuno- assay

Insulin-like bioactivity was studied by measur- ing the conversion of [U-14C]glucose to 14CO2 (19). Fat pads from 2 rats were digested with

2 Megyesi, K., L. Underwood, J. J. Van Wyk, and C. R. Kahn, unpublished observations.

FIG. 2. Comparison of plasma insulin-like activity with pork insulin standard in the radioreceptor assay. One ml serum was acidified and gel filtered on Sephadex G-50 column in IM acetic acid (see Ma- terials and Methods). Serial dilutions of the peak fraction of insulin-like activity were assayed as de- scribed in the legend to Fig. 1. 125I-insulin bound, expressed as per cent of maximum, is plotted as a function of the unlabeled insulin concentration (solid line) or dilution of the serum fraction (dashed line).

100

125| - INSULIN BOUND (% of Maximum)

80

60

1/10

1/4

1/2

1/1

40

Serum Dilution

20

1

10

100

PORK INSULIN STANDARD (ng/ml)

collagenase (2 mg/ml) at 37 C for.60 min. The cells were filtered through nylon mesh, washed 3 times in Krebs Ringer Bicarbonate containing 3% bovine serum albumin (KRB-A), and sus- pended in 15 ml of KRB-A. For assay 0.2 ml of the cell suspension was added to 1.8 ml of KRB-A containing 0.3 mM glucose and 10 uCi of [U-14C]glucose in 30 ml polyethylene bottles. The incubation vessels were gassed with 95% O2:5% CO2, capped, and incubated in a metabolic shaker at 37 C for 2 h. At the end of the incu- bation period, the contents of the vessels were acidfied with 0.3 ml 2 NH2SO4 and the evolved 14CO2 collected in 0.2 ml hyamine hydroxide in hanging wells. All glucose oxidation assays were performed in duplicate or triplicate.

Immunoreactive insulin was measured by the double antibody method (10).

Results

In an attempt to determine the etiology of the hypoglycemia in this patient, sera obtained from the patient were subject to four assays for insulin-like peptides: insulin immunoassay, NSILA-s/MSA and insulin radioreceptor assays and glucose oxidation bioassay (Fig. 3). Circulating insulin in the patient’s serum, as determined by radioim-

FIG. 3. Studies on serum. Serum of the patient was studied in NORMAL RADIO RECEPTOR ASSAY BIOASSAY TUMOR 4 assay systems compared with HYPOGLYCEMIA normal sera (see Materials and Methods for details). For the 500 radioreceptor assays (NSILA-s and insulin), serum was ex- tracted by acid gel filtration. In each panel, the data have 300 been normalized by considering the value obtained on the normal sera to represent 100%. The PERCENT OF NORMAL columns represent the mean of 100 the determinations, and the vertical lines indicate the stand- ard error of the mean. For the insulin and NSILA-s/MSA radio- receptor assays of the patient serum, each fraction of the gel filtration was assayed, and the results summed to give total insulin-like activity NSILA-s (13). Gel filtration was performed on two separate sera for insulin- like activity, but only one serum for NSILA-s. A second serum was studied without gel filtration and also had an increased level of NSILA-s related peptides. For both of these assays, 10-12 normal sera were used to determine the mean normal value. The bioassay is the mean of three determinations each done in triplicate on one sample of the patient's serum and three normal sera. Immunoreactive insulin was measured on three samples of the patient's serum and a large group of normals.

IMMUNOREACTIVE INSULIN

NSILA-S

INSULIN-LIKE ACTIVITY

munoassay, was not elevated (8 + 3 / U/ml), as compared to normal basal levels (7 ± 1 U/ml). These low insulin levels are even more impressive when one considers the fact that the patient could only go 4 h without feeding, and thus these values were obtained in the early post-prandial period.

In contrast, the concentration of NSILA-s related peptides, as measured by NSILA- s/MSA (see Materials and Methods) radio- receptor assay, of the patient serum was about four times higher than the mean con- centration observed in normal serum (36,300 ng/ml versus 9,000 ng/ml based on present standard). Both the insulin radioreceptor assay and the fat cell bioassay demonstrated about a two-fold increase in total insulin- like activity in the patient’s serum. Total insulin-like activity was 1609 ± 122 u U/ml versus 903 ± 100 µU/ml by insulin radioreceptor assay, and 950 ± 200 µU/ml versus 500 ± 250 µU/ml by bioassay (Figs. 3 and 4).

Whereas the latter three assays showed a consistent elevation in the insulin-like ac- tivity of the patient serum compared to normal serum, it is important to note that the marked increase observed in the more specific NSILA-s/MSA radioreceptor assay is partially obscured in the insulin radio- receptor and bioassays. Also note that the difference observed in the conventional bioassay may not be significant due to the

FIG. 4. Stimulation of glucose oxidation in isolated adipocytes by serum. Serial dilutions of normal serum and the patient's serum are compared to the pork insulin standard. Each point is the mean of 3 dif- ferent assays in triplicate (see Materials and Methods).

100

STIMULATION OF GLUCOSE OXIDATION (% of maximal)

80

Insulin

60

Hypoglycemia Tumor

40

Normal Serum

20

1:2000

1:200

SERUM DILUTION

1:20

0.1

[INSULIN] (ng/ml)

1.0

10

great variation observed with normal serum (despite the fact that the data shown in Fig. 4 represent the mean of 3 assays each done in triplicate). Further, the non-parallel dose response of serum compared to insulin sug- gests that inhibitors of the bioassay are present in unextracted serum (Fig. 4). Gel

FIG. 5. Insulin-like activity in the patient's serum. One ml serum, to which a trace of 125I-insulin was added, was acidified and gel filtered on Sephadex G-50 in 1M acetic acid (see Materials and Methods). The absorbance of each fraction at 280 nM (O --- O) was measured, and the 125I-insulin ( --- ) de- termined by counting in a gamma counter. Each fraction was then neutralized and as- sayed in the insulin radiorecep- tor assay as described in Ma- terials and Methods. The in- sulin-like activity is expressed as ng-equivalents of insulin

30

125

Insulin

O -O ABSORBANCY AT 280 nm

—. RADIOACTIVITY (x 102 cpm/tube)

25

25

ILA IN RADIORECEPTOR ASSAY (ng/ml)

10

20

ILA in Pt

20

8

15

15

6

10

10

4

8

5

6

2

4

2

10

15

20

25

30

FRACTION (ml)

FIG. 6. Stimulation of glucose oxidation by isolated adipocytes by tissue extracts. Serial dilu- tions of the tumor extracts and control tissue extracts were as- sayed in the glucose oxidation bioassay as described in Ma- terials and Methods. The results are expressed as per cent maxi- mal stimulation compared to the pork insulin standard.

ETHANOL-ACETONE EXTRACT

ACID-ETHANOL EXTRACT

GLUCOSE OXIDATION (% maximal stimulation)

100

Insulin

80

Liver Metas.

60

Lung

40

Adrenal Tumor

20

Normal Liver

Liver Met.

.2

Normal Liver

[INSULIN] (ng/ml)

.375

.75

1.5

0

.375

.75

1.5

Adrenal Tumor

[EXTRACT] 100 (g/ml)

500

1000

100

500

1000

filtration of serum did indicate that the insulin-like activity detected in the insulin radioreceptor assay eluted ahead of 125]- insulin and in a position consistent with the elution volume of NSILA-s (Fig. 5) (20).

Insulin-like activity in extracts of tumor tissue was elevated as compared to that of normal tissue, suggesting that the NSILA-s was produced by the tumor itself. As shown in Fig. 6 (left), the stimulation of glucose

oxidation by tissues extracted by ethanol- acetone was clearly greater for the liver metastases and the adrenal tumor than for samples of normal liver from the same pa- tient. Interestingly, extracts of normal lung also showed higher stimulation than those of normal liver, suggesting the presence of either some insulin-like material or a non- specific stimulator in the extract. Used at the same protein concentration the acid-

FIG. 7. Insulin radioreceptor assay of tissue extracts. Serial dilutions of the extracted control and tumor tissues were assayed as described in Materials and Methods, the results are expressed as per cent 125I-insulin bound compared with the pork insulin standard. For clarity, only one liver tumor and one adrenal tumor are shown in each panel, but these were representative of the other extracts.

ETHANOL - ACETONE METHOD

ACID - ETHANOL METHOD

100

100

Liver Tumor Ext.

125| - INSULIN BOUND (% of Maximum)

Control Liver Ext. Lung Ext.

Adrenal Tumor Ext.

80

Adrenal Tumor Ext.

80

Liver Tumor Ext.

60

60

Pancreas Extract

Pork Insulin

40

40

Pork Insulin

20

0 — 0 --- 0 — 0 — O Pancreas Ext.

20

1

10

100

1000

1

10

Pork Insulin (ng ml)

PORK INSULIN (ng/ml)

100

10

100

1000

10

100

1000

EXTRACT ( g/ml)

EXTRACT (19’ml)

FIG. 8. NSILA-s/MSA radiore- ceptor assay of tissue extracts. NSILA-s was measured in each tissue extract using the radio- receptor assay. Results are ex- pressed as ng-equivalents of our present standard per mg of ex- tract.

ETHANOL-ACETONE EXTRACTION

ACID-ETHANOL EXTRACTION

400

300

☒ LIVER METASTASIS

ADRENAL TUMOR

NSILA-s (ng/mg of extract)

CONTROL LUNG

CONTROL LIVER

200

100

25

0

1

2

3

1

2

1

2

3

0

1 2

ethanol extracts of the same tissues showed very low or undetectable activity (Fig. 6, right).

Similar results were observed in the in- sulin and NSILA-s radioreceptor assays (Figs. 7 and 8). Adrenal tumor and liver metastasis showed substantial amounts of in- sulin-like activity when extracted by etha- nol-acetone, whereas the activity was lower by acid-ethanol extraction. The NSILA-s/ MSA radioreceptor assay again detected greater differences among tissue extracts, consistent with its greater specificity. The concentrations of NSILA-s related pep- tides in three different samples of liver me- tastases and two different samples of adrenal tumor were 4- and 20-fold higher than those found in control tissues obtained from the same patient (Fig. 8). On gel filtration, the activity detected in the NSILA-s/MSA in the extract of the liver metastasis and in the serum migrated with an apparent molecular weight of 5,000-10,000 (Fig. 9).

In all assays, regardless of whether the data are expressed as uU/mg extract, or uU/g tissue, the liver metastases and the adrenal tumor had significantly higher in- sulin-like activity than the normal liver (Table 1). In general, the NSILA-s/MSA

radioreceptor assay showed higher values and greater differences between the normal and tumor tissues. This may reflect the greater sensitivity and specificity of this as- say, or the presence of some NSILA-s- like material which does not have insulin- like bioactivity or receptor activity. Alter- natively, the lower values obtained in the glucose oxidation bioassay could be due to inhibitors in the extracts. Though it is true that none of the NSILA-s values detected in the tumor tissues exceeded that of the plasma, no recovery data are available on these extracts, and our previous experience with extraction would suggest that recovery is, at best, about 10-15% using these methods.

The immunoreactive insulin concentra- tions in the tissue extracts (with the excep- tion of the pancreas) were very low and could not account for the insulin-like ac- tivity as determined by either the bioassay or radioreceptor assays (Table 2).

Discussion

Of the endocrine syndromes associated with malignancy, fasting hypoglycemia has been one of the most difficult to explain.

FIG. 9. Gel filtration of serum and tumor extracts from a patient with adrenocortical carcinoma and hypo- glycemia. Serum (0.5 ml) was gel filtered in a Sepha- dex G-75 column (0.9 x 55 cm) in a IM acetic acid and each fraction lyophylized twice and recovered in KRP, pH 7.4. The fractions were then assayed for NSILA-s related peptides using the radioreceptor assay. The tumor extract (0.5 ml) was filtered on Sephadex G-50 column (0.9 × 55 cm) in IM acetic acid and fractions were treated as above. The shaded bars represent the NSILA-s activity found in each fraction, the solid line ( .-. ) the O.D. 280 of the fractions, and the two arrows represent the void column and 125I-MSA markers, respectively.

Vo

”$ I - MSA

40

4

PLASMA

SEPHADEX G75 (.9 x 55 cm)

30

3

20

2

10

1

[NSILA-s] (ug/ml)

0

0

O. D. 280

16

20

24

28

32

36

40

0.4

0.8

TUMOR EXTRACT

0.3

SEPHADEX G50 (.9 x 55 cm)

0.6

0.2

0.4

0.1

0.2

0

0

16

18

20

22

24

26

30

FRACTION

Mechanisms which have been suggested in- clude: 1) secretion of insulin or insulin- like substances (21-27), 2) excessive glu- cose utilization by the tumor (28-32), 3) destruction of the liver by the tumor (3), 4) suppression of counter-regulatory hor- mones, and 5) inhibition of glycogenolysis and gluconeogenesis (3,32-35).

Prior to the development of radioreceptor assay, demonstration of insulin-like activity depended upon careful extraction pro- cedures and tedious bioassays. Despite these adversities, several investigators were able to demonstrate that serum or tumor

TABLE 1. Insulin-like activity of tissue extracts from a patient with tumor hypoglycemia
Fat cell bioassayRadioreceptor assays
InsulinNSILA-s/MSA
p U of ILA/mg Extract"
Liver metastasis15- 2030- 4028-40
Adrenal tumor2- 44- 1320-30
Normal liver<1<13- 3.5
pU of ILA/gm Tissue
Liver metastasis180-350350-400450-500
Adrenal tumor25- 4040-160300-500
Normal liver<15<1550

* The calculations were based on equating I ng ILA = 25 uU in the bioassay and insulin radioreceptor assay, and 100 ng MSA = pU in the NSILA-s radioreceptor assay. The data in the upper panel are ex- pressed as pU of insulin-like activity (ILA) per mg extract (dry weight). In the lower panel the ILA is expressed per gm tissue (wet weight). Both sets of data are calculated from the acetone-ethanol extracts.

tissue of some patients with tumor hypo- glycemia had increased levels of insulin- like activity which could not be suppressed by anti-insulin antibody and could not be accounted for by immunoreactive insulin (21-27). With the development of the NSILA-s and insulin radioreceptor assays, detection of insulin-like peptides in serum was simplified. Using a NSILA-s radio- receptor assay, we demonstrated that in about one-third of the patients with this syndrome, plasma levels of NSILA-s related peptides were elevated and could account for the hypoglycemia in these cases (4,14).

In the present report we have studied in detail serum and tumor extracts of one patient with hypoglycemia due to adreno- cortical carcinoma. The clinical and labora-

TABLE 2. Immunoreactive insulin concentrations of tissue extracts
(ng/mg extract)
Liver metastasis (1)0.20
Liver metastasis (2)N.D .*
Liver metastasis (3)0.05
Adrenal tumor (1)0.08
Adrenal tumor (2)0.04
Control lung0.10
Control liver (1)0.09
Control liver (2)0.02
Control liver (3)0.02

* N.D. = Not done.

tory features in this patient were typical of those observed in other patients with this syndrome. Failure to find elevated con- centrations of immunoreactive insulin led us to study the serum in several other assays. As we have previously reported in other cases, the concentration of NSILA-s-like peptides was markedly elevated when meas- ured by the more specific NSILA-s/MSA radioreceptor assay. This assay cannot dis- tinguish NSILA-s related peptides from MSA-like or somatomedin-A-like peptides, but clearly distinguishes these from insulin, proinsulin, and a variety of other growth factors. Present data would suggest that some component of the somatomedin ac- tivity related to somatomedin-C cross-reacts poorly or not at all in this receptor assay (see Materials and Methods and ref. 11).

The conclusion that NSILA-s related pep- tides were elevated might also have been reached using the insulin radioreceptor assay which demonstrated increased in- sulin-like activity which eluted on G-50 Sephadex in a position ahead of the 125I- insulin marker. On the other hand, bioassay of serum by measurement of glucose oxidase by adipocytes showed an increase of insulin- like activity of questionable significance. This was further complicated by the non- parallel dose response curves and sugges- tion of inhibitory activity at high serum concentrations. It is possible that the bio- assay would have detected more significant differences, had it been used in conjunction with gel filtration, or some other effective extraction procedure.

Our finding of increased concentrations of NSILA-s in tumor extracts as compared to extracts of normal tissues is highly sug- gestive of the fact that the NSILA-s related peptides were produced by the tumor itself. Unger (36) had suggested that some tumors may “non-specifically” trap peptide hor- mones present in the circulation. However, this seems unlikely in this case, since adja- cent normal tissues exposed to the same serum contained low levels of NSILA-s. As in the studies of serum, the NSILA-s/

MSA radioreceptor assay showed the great- est differences due to its greater sensitivity and specificity.

The most striking finding in our study of the tissue extracts was the great difference observed between the ethanol-acetone ex- traction method and the acid ethanol ex- traction method. These differences are es- pecially important, since most previous studies have relied on acid ethanol extrac- tion (21,24). The great variability observed among extraction methods and lack of specific and sensitive assays in the past may have accounted for the difficulty in de- tecting these insulin-like peptides in previ- ously reported patients.

Acknowledgments

The authors wish to acknowledge Ms. C. Hendricks and K. Baird for technical assistance and Dr. Jesse Roth for review of the manuscript.

References

1. Doege, K. W., Fibro-sarcoma of the mediastinum, Ann Surg 92: 955, 1930.

2. Laurent, J., G. Derby, and J. Floquet, Hypogly- cemic Tumors, Excerpta Medica, Amsterdam, 1971.

3. Marks, V., and R. C. Rose, Hypoglycemia, Black- well Scientific Publications, Oxford, England, 1965.

4. Megyesi, K., C. R. Kahn, J. Roth, and P. Gorden, Circulating NSILA-s in man: Preliminary studies of stimuli in vivo and of binding to plasma components, J Clin Endocrinol Metab 41: 475, 1975.

5. Oeltz, O., E. R. Froesch, H. F. Bünzli, R. E. Humble, and W. J. Ritschard, Insulin-like activity, In Steiner, D. R., and N. Freinkel (eds.), Hand- book of Physiology, vol. 1, Williams and Wilkins, Baltimore, 1972, p. 685.

6. Grodsky, G. M., and C. G. Peng, Extractable insulin measured by immuno-chemical assay: Ef- fect of tolbutamide, Proc Soc Exp Biol Med 101: 100, 1959.

7. Jakob, A., Ch. Hauri, and E. R. Froesch, Nonsup- pressible insulin-like activity in human serum. III. Differentiation of two distinct molecules with nonsuppressible ILA, J Clin Invest 47: 2678, 1968.

8. Hunter, W. M., and F. G. Greenwood, Preparation of iodine-131 labelled human growth hormone of high specific activity, Nature 194: 495, 1962.

9. Tsushima, T., and H. G. Friesen, Radioreceptor

assay for growth hormone, J Clin Endocrinol Metab 37: 334, 1973.

10. Gavin, J. R., III, C. R. Kahn, P. Gorden, J. Roth, and D. M. Neville, Jr., Radioreceptor assay of insulin: Comparison of plasma and pancreatic insulin and proinsulins, J Clin Endocrinol Metab 41: 438, 1975.

11. Megyesi, K., C. R. Kahn, and J. Roth, The NSILA-s receptor in liver plasma membranes, J Biol Chem 250: 8990, 1975.

12. Dulak, N. C., and H. M. Temin, A partially purified polypeptide fraction from rat liver cell conditioned medium with multiplication-stimula- tion activity for embryo fibroblasts, J Cell Physiol 81: 153, 1973.

13. Neville, D. M., Jr., Isolation of an organ specific protein antigen from cell-surface membrane of rat liver, Biochem Biophys Acta 154: 540, 1968.

14. Megyesi, K., C. R. Kahn, J. Roth, and P. Gorden, Hypoglycemia in association with extrapancreatic tumors: Demonstration of elevated plasma NSILA-s by a new radioreceptor assay, J Clin Endocrinol Metab 38: 931, 1974.

15. Rechler, M. M., J. M. Podskalny, and S. P. Niss- ley, Interaction of multiplication-stimulating ac- tivity with chick embryo fibroblasts demonstrates a growth receptor, Nature 259: 134, 1976.

16. Nissley, S. P., and M. M. Rechler, Multiplication stimulating activity (MSA): A somatomedin-like polypeptide from cultured rat liver cells, 3rd Decennial Research Conf., Lake Placid, NY, Sept. 13-17, 1976, Natl Cancer Inst. Monograph (In press.)

17. Zapf, J., M. Mader, M. Waldvogel, D. S. Schlach, and E. R. Froesch, Specific binding of nonsup- pressible insulin-like activity to chick embryo fibroblasts and to a solubilized fibroblast receptor, Arch Biochem Biophys 168: 630, 1975.

18. Rinderknecht, E., and R. E. Humbel, Polypeptides with nonsuppressible insulin-like and cell-growth promoting activity in human serum: Isolation, chemical characterization, and some biological properties of forms I and II, Proc Natl Acad Sci USA 73: 2365, 1976.

19. Rodbell, M., Metabolism of isolated fat cells. I. Effects of hormones on glucose metabolism and lipolysis, J Biol Chem 239: 375, 1964.

20. Megyesi, K., C. R. Kahn, J. Roth, E. R. Froesch, R. E. Humbel, J. Zapf, and D. M. Neville, Jr., Insulin and non-suppressible insulin-like activity (NSILA-s): Evidence for separate plasma mem- brane receptor sites, Biochem Biophys Res Com- mun 57: 307, 1974.

21. Abbasi, A., and L. Power, Insulin and insulin- like activity in extracts of tumors associated with hypoglycemia, Diabetes 22: 762, 1973.

22. Subauste, C., R. Calderon, L. A. Lierena, and E. Carrion, Insulin and insulin-like activity in tumor tissue and plasma of a patient with a fibrosar- coma associated with hypoglycemia, Metabolism 14: 881, 1965.

23. Volpe, R., J. Evans, D. Phil, D. W. Clarke, N. Forbath, and R. Ehrlich, Evidence favoring the sarcomatous origin of an insulin-like substance in a case of fibrosarcoma with hypoglycemia, Am J Med 39: 540, 1965.

24. August, J. T., and H. A. Hiatt, Severe hypogly- cemia secondary to a nonpancreatic fibrosarcoma with insulin activity, N Engl J Med 258: 17, 1958.

25. Field, J. B., H. Keen, P. Johnson, and B. Herring, Insulin-like activity of nonpancreatic tumors as- sociated with hypoglycemia, J Clin Endocrinol Metab 23: 1229, 1963.

26. Whitney, J. E., and B. J. Heller, Increased in- sulin-like activity of the serum in a patient with spontaneous hypoglycemia associated with a retro- peritoneal fibrosarcoma, Am J Med 30: 633, 1961.

27. Steinke, J., S. Soeldner, and A. E. Renold, Insulin- like activity of extracts from large sacromatous tumors associated with hypoglycemia, J Clin In- vest 41: 1403, 1962.

28. Handa, S. P., Hypoglycemia in primary carcinoma of the liver: Report of a case, Am J Dig Dis 11: 989, 1966.

29. Carey, R. W., T. G. Pretlow, E. Z. Ezdinli, and J. F. Holand, Studies of the mechanism of hypoglycemia in a patient with massive intra- peritoneal leiomyosarcoma, Am J Med 40: 458, 1966.

30. Sellmann, J. C., G. T. Perkoff, F. C. Null, J. R. Kimmel, and F. H. Tyler, Hypoglycemia associated with massive intra-abdominal mesothelial-cell sar- coma, N Engl J Med 260: 847, 1959.

31. Landau, B. R., N. Wills, J. W. Craig, J. R. Leonards, and A. Moriwaki, The mechanism of hepatoma- induced hypoglycemia, Cancer 15: 1188, 1962.

32. Unger, R. H., Editorial of the riddle of tumor hypoglycemia, Am J Med 40: 325, 1966.

33. Silverstein, M. N., Tumor hypoglycemia, Cancer 23: 142, 1969.

34. Silverstein, M. N., K. G. Wakim, and R. C. Bahn, Tryptophan hypoglycemia associated with neo- plasia, Cancer 19: 2024, 1966.

35. Jakob, A., V. A. Meyer, R. Flury, W. H. Ziegler, A. Labhart, and E. R. Froesch, The pathogensis of tumor hypoglycemia: Block of hepatic glucose release and of adipose tissue lipolysis, Diabeto- logia 3: 506, 1967.

36. Unger, R. H., J. D. Lochner, and A. M. Eisen- traut, Identification of insulin and glucagon in a bronchogenic metastasis, J Clin Endocrinol Metab 24: 823, 1964.