CLIN. CHEM.22/9, 1516-1521 (1976)
Protein-Bound Carbohydrates in Breast Cancer.
Liquid-Chromatographic Analysis for Mannose,
Galactose, Fucose, and Sialic Acid in Serum
J. E. Mrochek,1 S. R. Dinsmore,1 D. C. Tormey,2 and 1. P. Waalkes2
We describe high-resolution chromatographic analysis
for protein-bound sialic acid in serum, with use of a cerate
oxidimetric detector. Values for sera from normal women
averaged 680.5 mg/liter, with a coefficient of variation of
23%. Including data obtained by previously developed
chromatographic procedures forprotein-boundmannose,
galactose, and fucose, we assessed sera from breastcancer patientswhose malignancy had been categorized
as either stable, responsive, or progressive (based on
clinicalobservations spaced from two to fivemonths
apart). All of 12 responsive patients had decreases of
protein-bound fucose averaging 34.5% (SD, 16.1) and all
of 10 patients with progressive disease had increases
averaging 38.3% (SD 21.5). Changes in fucose averaged
less than 6.7% (SD, 4.9) for eight patients with clinically
stablebreastcancer.Changes inprotein-boundmannose,
galactose, and sialic acid did not correlate as well as did
fucose with the clinical disease status of the patients.
laboratory animals having induced tumors (3). However, increased concentrations of serum glycoproteins
in serum are not a specific sign of malignant disease;
they may appear in patients
with renal, rheumatic,
or
infectious diseases, cardiovascular disorders, or biliary
obstructions (4-7).
The real role that carbohydrates play in moderating
the activity
of protein is one of the least-understood
phenomena in human biochemistry. Some interesting
studies on the role of sialic acid in this context have been
reported recently.Watkins et al.(8) reported in vitro
studiesshowing that serum with abnormally high pro-
tein-bound
sialic acid from a cancer-free human
non-
specificallyblocked host immune response to tumor
cells, and this blocking effect could be eliminated
by
partialenzymic removal of sialicacid from serum gly-
coproteins
with neuraminidase
(EC 3.2.1.18). Other
studies, performed by injecting animals with radioactivelylabeled glycoproteins,have demonstrated that
AdditIonal Keyphrases: cerate oxidimetric detector
#{149}
monitoring cancer #{149} serum glycoproteins
#{149} normal
glycoproteins whose terminal sialic acid residues were
values
stripped
off by neuraminidase
to expose galactose
The existence of glycoproteins,
and their importance
residues were subject to rapid removal from the circuin many forms of life, has been recognized for more than
15 years.However, much about the biologicalfunction
of the sugar moieties in glycoproteins remains a mystery. In some cases their function is fairly obvious. For
example, mucins have a lubricating function, which can
be related to carbohydrate residues. In other cases a
particularsugar residue may act as a signalfor recognition, either by other proteins
or by whole cells (1).
Thus, the carbohydrate moieties may influence growth
and cell-to-cell interactions and be of importance in the
development of cancer. Increased concentrations of
serum glycoproteins have been reported in human
subjects with malignant diseases (2) and in serum from
‘Oak Ridge National
National
Institutes
Operated
by Union
2
Received
1516
Laboratory,3
Oak Ridge, Tenn. 37830.
of Health, Bethesda,
Md. 20014.
Carbide Corporation
for the ERDA.
May 4, 1976; accepted
CLINICAL CHEMISTRY,
June
9, 1976.
Vol. 22, No. 9, 1976
lationby the liver.The same glycoproteinswith terminal sialic residues intact displayed normal turnover in
the circulation
(see ref. 1). The implication
of these
studies seems to be that terminal sialic acid residues
might protect a foreign protein from rejection caused
by host immune response.
Rosato et al.(9) reported an apparent relationship
between increased
concentrations
of protein-bound
fucose in serum and the presence of malignant breast
masses. Previously, we reported methodology
for the
chromatographic
analysis of the three neutral protein-bound
carbohydrates (mannose, galactose, and
fucose) in serum and analytical data on samples from
normal women and those with metastatic breast cancer
(10). Concentrations of the three carbohydrates
were
generally increased in hydrolysates
of serum from the
patients with cancer, the increase in fucose being pro-
a
PEAK EA
(orbltrory
,n,.)
Fig. 1. Linearresponseof the cerate oxidimetric detector to sialic
acid eluted from an anion-exchange column with ammonium
acetate
portionately higher than that in mannose and galactose.
The present study extends the analytical methodology
to include sialic acid in addition to the three neutral
carbohydrates and describes the variation of these four
protein-bound
carbohydrates
with relation to the
clinical status of breast cancer. The purpose of this work
isto investigate the potential of these molecular entities
as biochemical markers of breast cancer, not so much
in a diagnostic sense, but more as a means of measuring
the population of malignant cells to guide chemotherapeutic treatment
in maximizing
their destruction.
neutral
carbohydrates
system used for the analysis of
of glycoproteins
was
de-
scribed in a previous publication (10). This system was
modified for the chromatographic analysis of proteinbound sialic acid. A shorter column (100 X 0.22 cm i.d.)
was used, and its jacket temperature was maintained
at 60 #{176}C
by a constant-temperature circulating bath.
The detection system used was the cerate oxidimetric
detector as described previously (10). A linear relation
of both the peak area and peak height to concentration
was demonstrated
eluent
consisting
for amounts
of sialic acid ranging
from 6.5 to 39.1 g eluted from the anion-exchange
column (Figure 1).
Initial separations of hydrolyzed
serum samples were
performed by using a concentration gradient of ammonium acetate/acetic acid (pH 4.3). The gradient
caused sialic acid to be eluted after about 7 h. Few extraneous chromatographic
peaks were observed (no
other major peaks); consequently,
conditions
were
sought that would enable a more rapid analysis of sialic
acid without the necessity of column regeneration. A
number of different concentrations
of the acetate buffer
were tested to find isocratic elution conditions that
would enable separation of the peak corresponding
to
sialic acid from the large peak appearing at sample
breakthrough.
We found that the combination
of an
of 0.2 mol/liter
ammonium
acetic acid (pH 4.4) and a column temperature
acetate/
of 60 #{176}C
yieldeda satisfactory
separationof serum hydrolysates
(Figure 2).
Method
The neutral
System
The chromatographic
the
Fig. 2. Isocratic elution of a normal human serum hydrolyzed
under mild conditions to liberate sialic acid
Analytical
Methods and Materials
Chromatographic
TIME (MI
protein-bound
carbohydrates
of serum
glycoproteins were analyzed as described previously
(10). Sialic acid, more labile than the carbohydrates, was
analyzed separately as described below:
(1) Precipitate the protein in 0.5 ml of serum by
adding 3 ml of ethanol (95%) followed by centrifugation
(15 mm, 5000 rpm); discard the supernate.
(2) Wash the precipitate
by suspension in 3 ml of
ethanol, again centrifuge and discard the supernate.
(3) Dissolve the protein by gentle agitation in 1.0 ml
of 0.1 mol/liter NaOH.
(4)
Add 2.0 ml of 0.1mol/literH2S04; mix by swirling,
deaerate with nitrogen, and seal tightly with a septum-sealed cap (Teflon-faced septum).
(5)Place the sample in a heating block maintained
at 80 #{176}C
for 1 h; then cool to room temperature and
neutralize to an approximate pH of 4 with 0.25 ml of 1
mol/liter NaOH. Centrifuge the hydrolysate and remove
the supernate for chromatographic
analysis.
Evaluation
of Experimental
Variables
The neutral carbohydrates of glycoproteins were
hydrolyzed in 1.0 mol/liter HC1 for 4 h at 100 #{176}C
as described previously (10). Under similar hydrolytic conditions, sialic acid loses about 25% of its nitrogen content in the form of ammonia (11, 12). To avoid degradation of this labile compound, milder hydrolysis conditions are generally used for the release of proteinCLINICAL
CHEMISTRY,
Vol. 22, No. 9, 1976
1517
To further test the method, we did multiple analyses
of a pooled serum sample to establish a baseline value
for the sample. Known amounts of sialicacid were
added to the precipitated protein from this pooled
serum and the samples were analyzed in the usual
manner. The results (Table 1) showed the average recovery to be 92.4% of the added material, with a standard deviation of 3.8%.
Effect of Freeze-Thaw Cycles
on Sialic Acid Analysis
MUCIN
I,,g(
Fig. 3. Proportional recovery of protein-bound sialic acid from
bovine submaxillary
mucin after hydrolysis in 0.1 mol/liter
acid
at 80 #{176}C
for 1 h
bound sialic acid: 50 mmol/liter
H2S04 at 80 #{176}C
for 1 h
(13, 14). We used these conditions
in our work.
Five samples of N-acetylneuraminic
acid were exposed to these hydrolysis
conditions
for sialic acid,
cooled to room temperature, adjusted to about pH 4
with 1 mol/liter sodium hydroxide, and analyzed by
high-resolution liquid chromatography. Analytical recoveries averaged 97.2% (CV, 2.2%), indicating that
sialic acid is stable under these hydrolysis conditions.
To evaluate the analytical
method,
we hydrolyzed
and then analyzed bovine submaxillary
mucin (Sigma
Chemical Co., St. Louis, Mo. 63178) for sialic acid. The
glycoprotein,
in amounts
ranging from 1 to 4.9 mg, was
hydrolyzed according to the just-described
procedure
and then analyzed by high-resolution
anion-exchange
chromatography.
Proportional liberation
and recovery
of sialic acid were demonstrated
(Figure 3). A mean of
71.9 tg of sialic acid per milligram of mucin (CV, 5.4%)
was found by least-squares
nine samples.
analyses
Table 1. Analytical
of results
for the
A pooled serum was prepared from freshly thawed
sera (stored at -70 #{176}C).
The pooled sample was thawed
for about 1 h, sampled for analysis, and then refrozen
for three days before thawing again. This thaw-freeze
cycling was continued through three cycles. Analytical
results indicated thaw-induced changes in sialic acid
content with an increase of 7.8% after the second cycle
and a decrease of 18.3% after the third cycle (Table 2).
These results suggest that samples may undergo at least
two thaw-freeze cycles without significant change in
sialic acid content.
Clinical Classification of Breast Cancer
Patients Included in Study
Patients included in this study of the variation of
protein-bound carbohydrates (mannose, galactose, fucose, and sialic acid) with changing malignancy status
had clinically demonstrable and biopsy-proved metastatic breast cancer. Local therapy used for these patients was either a radical or a modified radical mastectomy; in a few patients this was followed by radiotherapy. Further therapeutic regimes employed were:
(a) bilateral oophorectomy, (b) diethylstilbestrol,
or (c)
28-day cycles of either cyclophosphamide, methotrexate, and 5-fluorouracil;
or cyclophosphamide,
adriamycin, and 5-fluorouracil; or adriamycin and dibromodulcitol. Samples collected from patients receiving
chemotherapy were obtained at least 10 to 21 days after
the preceding dose of drug.
The patients were classified into three different
clinically defined disease categories: responsive, pro-
Recovery of Sialic Acid Added to Three Pooled Samples of Serum
Initial analysis
Added
Pool
Found
Sialic acid, mg/liter
I
639.6
(1.4)a,b
403.6
II
III
628.0
493.2
(2.9)a,b
403.6
(4.0)a,c
558
558
558
558
Recovery, %
1019.7
987.6
994.9
1039.6
1007.4
993.8
(1.5)a.b
94.2
(4.0)b
89.0
89.9
97.9
mean
a Coefficient
of variation,
%.
bThree analyses of the same hydrolysate.
C Four
analyses of four separate hydrolysates.
1518
CLINICAL CHEMISTRY,
Vol. 22. No. 9, 1976
92.2
89.7
92.4
±
3.8
STABLE
Table 2. Effect of Thawing and Refreezing Serum
on Analysis for Sialic Acid
Sialic acid,
mg/liter
Thaw cycle
First
PROGRESSIVE
lb
CV, %
3.6a
Second
657
708
Third
537
34
#{176}Four
analyses of four different
RESPONSIVE
00
2.5 a
90
hydrolysates.
:
a
70-
-
8’
gressive,and stable.A responsive classification
was
defined, for lesions measured in two dimensions, as at
least a 50% reduction in the sum of the products of the
longest perpendicular cross-sectional measurements of
the lesions in the organ system. In situations where lesions were measurable in only one dimension (e.g., liver
enlargement) a 30% reduction was required. An estimated 50% return to normality was required for tumors
that could be evaluated but not measured (e.g., lymphangitic
pulmonary
metastases). This reduction in
tumor burden had to persist for at least 28 days in at
least half of the involved organ systems to qualify for a
disease response classification. The second classification-progressive
disease-was
defined
as the
ap-
pearance of any new lesion or a greater than 25% increase in any existing lesion, measured as described
previously, so long as the increase was at least 2 cm2 in
bidirectionally
measured lesions and 2 cm in unidirectionally measured lesions and persisted for more than
STABLE
RESPONSIVE
PROGRESSIVE
-
60’-
S
5C-
a40b
30.
20-
Fig. 5. Comparison of variations in protein-bound sialic acid for
breast-cancer patients with clinical status of the malignancies
six days. A stable clinical status was designated in those
situations where criteria did not yet exist for either responsive or progressive disease status.
Results and Discussion
Protein-Bound Sialic Acid in Serum
from Healthy Women
I
We did chromatographic analyses for protein-bound
sialic acid in sera from 19 healthy women subjects. The
mean was 680.5 mg/liter (CV, 23.3%). The range was 294
to 919 mg/liter; the two SD range for the results was 363
to 998 mg/liter.
We have previously reported data on the proteinbound neutral carbohydrates
in sera from normal
women (10). Mannose averaged 444 mg/liter, galactose
436 mg/liter, and fucose 48 mg/liter for nonfasting
healthy women.
Effects of Changing Clinical Status of Breast
Cancer on Protein-Bound Carbohydrates
Fig. 4. Comparison of variations in protein-bound fucose for
breast-cancer patients with clinical status of the malignan-
cies
In an attempt to evaluate the efficacy of proteinbound carbohydrates as potential biochemical markers
of breast cancer, we assayed a series of serum samples
from patients with clinicafly defined breast cancer.
Patients in each of the three disease categories-stable,
responsive, and progressive-were selected, and at least
two serum samples, obtained over periods of time
ranging from two to five months were analyzed for
protein-bound
mannose, galactose, fucose, and sialic
acid. All analyses were performed blind; i.e., patient
CLINICAL
CHEMISTRY,
Vol. 22, No. 9,
1976
1519
a
F
I
S
I
FUCOSE
SIALIC
ACID
MANNOSE
GALACTOSE
Fig. 7. Comparison of variations in four protein-bound carbo-
hydrates for breast-cancer patients with clinically responsive
malignancies
Each individual patient has the same symbol for all analyses
Fig. 6. Comparison of variations in four protein-bound carbo-
hydrates for breast-cancer patients with clinically stable
malignancies
Each Individualpatient has the same symbol for
classification
was not revealed
allanalyses
until after the results
were collated.
Figure 4 shows a comparison of the changes in protein-bound fucose for breast-cancer patients whose
disease has been classified into the three clinical categories. Figure 5 compares, for many of the same patients
(same symbols in each category represent the same
patient), protein-bound
sialic acid for each of the three
clinical categories.
It is immediately apparent that
changes in protein-bound
fucose in every case correctly
mirrored the clinical assessment of the patient’s malignancy,
whereas
this
was not true
for sialic
CLINICAL CHEMISTRY,
Vol. 22, No. 9,
1976
I
I
acid.
Changes in protein-bound fucose averaged less than
6.7% (SD, 4.9%) for eight patients whose disease was
classified as clinically stable. All 12 patients classified
as having clinically responsive disease projected decreases in protein-bound
fucose that averaged 34.5%
(SD, 16.1%). All 10 patients with progressive malignancies displayed increases in fucose, averaging 38.3%
(SD, 21.5%). Decreases in sialic acid were noted for the
five patients whose clinical status was classified as
stable, with a rather dramatic decrease noted for one
patient (Figure 5). Six of eight patients whose disease
was responding to treatment
showed declines in sialic
acid, while only three of eight patients showed increases
where the malignancy was progressive. Note that most
of the data for sialic acid are within ±2SD of the mean
for the normal data, whereas 67% of the initial fucose
measurements for responsive disease and 60% of the
final measurements for progressive disease exceeded
+2SD from the mean.
Figures 6, 7, and 8 compare fucose, sialic acid, mannose, and galactose for the three clinical conditions. A
single patient has the same symbol for each of the four
1520
I
Fig. 8. Comparison of variations in four protein-bound carbo-
hydrates for breast-cancer patients with clinically progressive
malignancies
Each individual patient has the same symbol for all analyses
Note that 10 of 12 patients
showed declines in mannose and galactose for responsive disease
(Figure 7), paralleling the fucose data. One of the two
patients showing a reversal in this trend with increased
galactose also showed corresponding
increases in
mannose and sialic acid. The decrease in fucose for this
patient (solid reversed triangle, Figure 7) was not as
significant as noted for the remainder of the responsive
patients; thus it is possible that this particular patient
was undergoing
a change in the status of her malignancy.
In general, for progressive disease, decreases in one
of the three carbohydrates
(mannose, galactose, or sialic
acid) also appeared to be reflected in the other two
carbohydrates.
(Figure 8). Additional
References
patients
1. Ashwell, G., and Morell, A., The dual role of sialic acid in the hepatic recognition and catabolism of serum glycoproteins. Biochemical
Society Symposium No. 40, The Metabolism and Function
of Glycoproteins, R. M. S. Smellie and J. G. Beeley, Eds., 1974, The Bio-
follow-up analyses are needed on
this type of response, to ascertain
if such analytical measurements may be predictive of
changes in the clinical status of their malignancies.
Both fucose and sialic acid are terminal molecules in
the carbohydrate chains of glycoproteins, and so it
might be anticipated that an increase in the number of
residues of fucose might be accompanied by a corresponding decrease in sialic acid. However, our initial
results do not seem to show such an interrelationship
between these two compounds. Our results do show
good correlations of increasing protein-bound
fucose
with progressive breast cancer and decreasing fucose
with responsive malignancy, and this is encouraging.
Further studies are needed on other types of malignancies, with use of analytical methods of high specificity and with emphasis on protein-bound fucose and
sialic acid.
Changes in protein-bound sialic acid, mannose, and
galactose show only limited correlation with the clinical
status of the malignancy. Of necessity, because of a
delay in developing methodology, the samples which
were analyzed for sialic acid had previously been
thawed, analyzed for fucose, mannose, and galactose,
then refrozen. Although the effect of one freeze-thaw
cycle on sialic acid analysis seems relatively small (Table
2), it would be desirable to perform all analyses at the
time the sample was first thawed.
who display
Research
sponsored by the National
USPHS Contract NIH-CA71-59.
Cancer
Institute,
NIH, under
chemical Society, London.
2. Apffel, C. G., and Peters, J. H., Tumors and serum glycoproteins.
The symbodies. Prog. Exp. Tumor Res. 12, 1 (1969).
3. Harshman, S., and Bryant, G., Serum mucoid levels in rats bearing
Walker carcinoma 256 and the effect of surgical extirpation of the
malignance.
Cancer Res. 24, 1625 (1964).
4. Heiskel, C. L., Carpenter, C. M., Weimer, H. E., and Nakagawa,
S., Serum glycoproteins in infectious and inflammatory diseases. Ann
N. Y. Aced. Sci. 94, 183 (1961).
5. Kelly, V. C., Adams, F. H., and Good, R. A., Serum mucoproteins
in patients with rheumatic fever. Pediatrics
12,607 (1953).
6. Shetlar, M. R., Shetlar, C. L., Richmond, V., and Everett, M. R.,
The polysaccharide content of serum fractions in carcinoma, arthritis,
and infections. Cancer Res. 10,681 (1950).
7. Winzler, R. J., Plasma proteins in cancer. Adu. Cancer Res. 1,503
(1953).
8. Watkins, E., Jr., Gray, B. N., Anderson, L. L., et al., Neuraminidase-mediated augmentation
of in vitro immune response of patients
with solid tumors. mt. J. Cancer 14,799(1974).
9. Rosato, F. E., Seltzer, M., Mullen, J., and Rosato, E. F., Serum
fucose in the diagnosis of breast cancer. Cancer (Philadelphia)
28,
1575 (1971).
10. Mrochek, J. E., Dinsmore, S. R., and Waalkes, T. P., Liquidchromatographic analysis for neutral carbohydrates
in serum glycoproteins. Clin. Chem. 21, 1314 (1975).
11. Spiro, M. J., and Spiro, R. G., Composition of the peptide portion
of fetuin. J. Biol. Chem. 237, 1507 (1962).
12. Graham, E. R. B., Murphy, W. H., and Gottachalk, A., Mucoproteins. IX. The susceptibility to alkali and hydroxylamine
of the
predominant
carbohydrate-peptide
linkage in ovine submaxillary
gland glycoproteins. Biochim. Biophys. Acta 74, 222 (1963).
13. Svennerholm, L., Quantitative estimation of sialic acid. III. An
anion exchange method. Acta Chem. Scand. 12, 547 (1958).
14. Warren, L., Sialic acid in human semen and in the male genital
tract. J. Clin. Invest. 38, 775 (1959).
CLINICALCHEMISTRY,Vol. 22, No. 9, 1976
1521