| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
Bone and Mineral Division, Henry Ford Hospital, Detroit, Michigan 48202
Address all correspondence and requests for reprints to: A. M. Parfitt, M.D., Department of Endocrinology and Metabolism, University of Arkansas for Medical Sciences, 4301 West Markham Street, Slot 587, Little Rock, Arkansas 72205-7199.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Parathyroid adenoma cell birth rates measured by tritiated thymidine labeling (3) or by the prevalence of the mitotic karyotype (4) cannot be related to the rates in normal parathyroid tissue because the necessary data are unavailable. The Ki-67 antigen has a short half-life, labels only cycling cells (6), and with the MIB-1 antibody and microwave antigen retrieval can be applied to archival tissue (7). Using this method, we have recently reported a geometric mean cell birth rate for normal human parathyroid glands of about 5%/yr (8), demonstrating the parathyroid gland to be a conditional renewal tissue of very low turnover (9, 10). Values for abnormal cell proliferation can now be compared for the first time with normal values obtained by the same method, and we report the application of this method to parathyroid adenomas and to presumably suppressed nonadenomatous tissue in the same glands. We found significant differences, but also substantial overlap, among the three types of tissue. We discuss the results in relation to concepts of parathyroid tumor pathogenesis in general and the set-point hypothesis in particular (2, 3, 4, 5).
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
We studied 55 patients with primary hyperparathyroidism due to a
single adenoma. Eighteen patients (group A) were selected because they
had been under the care of one of the authors (A.M.P.) between 1972 and
1980, and detailed records were available. Thirty-seven patients (group
B) were selected because fresh parathyroid tissue had been harvested
between 1992 and 1995 for the study of cell proliferation by
bromodeoxyuridine incorporation, a method subsequently found to be
unsatisfactory. Demographic and biochemical data in the two groups are
compared in Table 1
. In group B, the mean
age was higher, and there were relatively more male and fewer black
patients, but these differences were not significant. The mean plasma
calcium level was lower in group B; this might have reflected a change
in case-finding methods or in criteria for surgical referral, but the
ranges were similar. Also, because of a change in method, the reference
range was lower than that during the earlier period. Mean plasma
phosphate and creatinine levels did not differ between the groups. In
group B, the mean intact PTH level (11) was 126 (SD = 91)
pg/mL, with reference values of 10–58. PTH was not measured in most
patients in group A. In group B, mean total alkaline phosphatase,
measured by automated analysis (12), was 117 (SD = 46)
IU/L, with reference values of 27–83. In many patients in group A,
alkaline phosphatase had been measured only by the Bodansky method.
|
Paraffin-embedded blocks were retrieved, and new sections of 5 µm thickness were cut; one was stained with hematoxylin and eosin, and the others were used for either proliferating cell nuclear antigen (PCNA) (7, 13) or Ki-67 (7, 8, 14) immunohistochemistry. After deparaffinization, the slides were incubated for 5 min in 3% hydrogen peroxide to block endogenous peroxidase. For PCNA, the slides were rinsed in phosphate-buffered saline, covered with 5% horse serum for 40 min, and incubated for 14–16 h overnight at 4 C with a 1:1500 dilution of monoclonal antibody to PCNA (DACO clone PC10). For Ki-67, the slides were transferred to plastic Coplin jars filled with 10 mmol/L citrate buffer (pH 6.0) and heated in a microwave oven twice for 5 min each time at a power of 1080 watts. After cooling at room temperature for 15 min, the slides were covered with 5% horse serum for 40 min and incubated overnight at 4 C with 1:100 MIB-1 monoclonal antibody (Immunotech, Westbrook, ME; catalogue no. 0505). For both PCNA and Ki-67, the slides were treated with biotinylated horse antimouse IgG for 30 min, followed by incubation with peroxidase-conjugated streptavidin for 60 min. Finally, the slides were developed for 5 min with 3-amino-9-ethyl carbazole (Sigma Chemical Co., St. Louis, MO). The mean specimen age was 21.8 yr for group A and 1.9 yr for group B. In each batch, new sections of human tonsil were used as a positive control (8). The sections were derived from two different blocks that had been treated in exactly the same way and had a mean specimen age of 2.0 yr.
Microscopic procedures
In each section, the gland profile was divided whenever possible into adenoma and nonadenoma tissue, using standard criteria (15). Each tissue type was subdivided into 2, 3, or 4 regions depending on size. In the adenomas, at least 20 fields were selected by systematic random sampling (16) at x40 magnification. Using an eyepiece graticule with an unbiased square counting frame with 100 test points, the number of positive cells was counted in 1 large square (=100 small squares) and the total number of chief cells was counted in 1 small square at x200 magnification. All cell profiles with any portion inside the frame, provided they did not touch or intersect the exclusion edges or their extensions, were counted (16). Only strongly staining chief cell nuclei without cytoplasmic staining were considered positive cells. In some cases, total chief cells were counted in the corresponding field of the adjacent hematoxylin- and eosin-stained section. The procedure was continued in each case until at least 150 total chief cells had been counted in small squares, corresponding to 15,000 cell profiles in the large squares. If no positive cell was found, a value of 0.5, equivalent to half the detection limit, was assigned (4). Results were expressed as the labeling index (LI) = positive cells/10,000 cells.
In 10 cases, all from group B, not differing in any feature from the other 27 cases, the entire adenoma profile in Ki-67-stained sections was examined. By systematically moving the stage row by row, the eyepiece graticule and counting frame were superimposed successively on adjacent square fields. In each counting frame, both total and positive cells were recorded as previously described, and the distribution of positive cells in the entire tumor was reconstructed (8). At the periphery of the tumor profile, fields in which fewer than 50 of the small squares overlay the tissue section were disregarded. The reconstructed map of the section was divided into outer and inner regions (8).
Calculation of cell birth rate and life span
For reasons given below, only Ki-67 data were used for these calculations. The Ki-67 antigen is expressed in all stages of the cell cycle, but decays rapidly, with a half-life of less than 1 h and so becomes undetectable very soon after the completion of mitosis (6). The fraction of Ki-67-positive cells is the same as the average fraction of time that each cell spends in the cycling, rather than in the noncycling (G0), state (17). The duration of the cell cycle in the parathyroid gland has never been measured, but a representative mean value for nonneoplastic cells in a variety of tissues and species is 24 h; the majority of values fall between 18–32 hs (9, 10). With this assumption: mean cell birth rate (%/yr) = LI (/10,000) x 365/10,000 x 100 = LI x 3.65. If the current cell birth rate were to be maintained, the corresponding mean cell life span (years) would be 100/(LI x 3.65). For such growth-related variables and for gland weight, the geometric mean was calculated as exp (mean loge x), and the geometric (multiplicative) SD was calculated as exp (SD loge x).
Statistics
Differences between tissue types were evaluated initially by
one-way ANOVA, followed by appropriate pairwise comparison, and
differences between groups and regions were determined by unpaired or
paired Student’s t tests as appropriate. Relations between
variables were tested by linear regression analysis. Nonparametric
methods were used when indicated. The frequency distribution of
positive cells among different graticule fields was compared with the
expected Poisson distribution by computation of the 
2
value (18).
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
. The LI was significantly higher for
PCNA than for Ki-67 (P = 0.002), although there was a
significant correlation between the two measurements (r = 0.53;
P < 0.001). Because positive and negative cells are
more easily distinguished and for other reasons described below,
further analysis was restricted to the Ki-67 results. The Ki-67 LI was
significantly lower in group A than in group B adenomas, presumably
because loss of antigen during prolonged storage of the blocks was
refractory to microwave retrieval. For the entire series, there was a
significant regression of LI on specimen age (Ln LI = 1.00 -
0.0457 x specimen age; r = -0.32; P =
0.017), and this regression was used to correct all LI values to a
specimen age of zero (corrected Ln LI = observed Ln LI +
0.0457 x specimen age). In nonadenomatous tissue, there was no
difference in Ki-67 LI between group A and group B and no significant
regression on specimen age. This difference might reflect smaller
sample size or stronger expression in normal than in abnormal tissue
and consequent lesser susceptibility to decay with time. The
age-corrected LI was almost 4 times higher in the adenomas than in the
nonadenomatous tissue, and there was no significant correlation between
the LI values.
|
. There were significant
differences among the three types of tissue. The cell birth rate was
about twice normal in the adenomas and about half normal in the
nonadenomatous tissue. The rates conformed to a log normal distribution
with very wide ranges and substantial overlap (Fig. 1). Only nine adenomas had values above
the 95% confidence interval for the normal glands (expected number,
1.4), and only 16 (29%) had values above the corresponding range. Only
one nonadenomatous value was below the 95% confidence interval for the
normal glands, and only three were below the corresponding range.
|
|
. A
total of 473 Ki-67-positive cells were found in approximately 1,243,000
total chief cells. The Ki-67 LI and calculated cell birth rate were
significantly higher by about 2-fold in the peripheral than in the
central regions, in which the rates did not differ significantly from
normal. The frequency distribution of number of positive cells per
field showed significantly more fields than expected with no labeled
cells, or more than two labeled cells (Table 5). The values for the entire tumor did
not differ from the values previously obtained by the less rigorous
method and were very similar to the values for the whole series based
on random sampling, although the variability was lower because of
greater precision. The mean number of cell profiles per graticule field
in different subjects ranged from 399-1702; the mean of 905 was not
significantly different from the value of 1178 cells/graticule field in
normal glands (8), but the variability between fields and between
subjects was greater. If, in each case, only the fields with the most
positive cells had been counted, the arithmetic mean label index would
have been 29.4 instead of 3.6/10,000 cells, an 8-fold overestimation.
If x400 rather than x200 magnification had been used, the referent
cell number would have been reduced 4-fold, and the bias would have
been increased an additional 4-fold.
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The prevalence of cell cycle markers has usually been used as a diagnostic and prognostic tool in patients with neoplasia (7, 20, 21, 22) and much less often to study rates of cell proliferation. Until recently, it was standard practice in diagnostic histopathology to express the prevalence of cell cycle markers per high power field, but because the areal density of cell profiles is so variable, it is much more satisfactory to use a defined number of cell profiles as the referent (27); in the present study, the number of cells per graticule field varied over more than a 4-fold range between patients. The interpretation of data concerning cell cycle markers depends heavily on the method of sampling within the section. In diagnostic work it has been usual to make measurements only in the most heavily labeled regions of the section (20). In the present study, this procedure would have overestimated the proportion of labeled cells and the calculated cell birth rate by at least 8-fold, and possibly much more, depending on the magnification used (8). We avoided this bias and estimated even very low prevalence values without unreasonable expenditure of time by using a much smaller graticule area to estimate the referent cell profile number than to count the positive cells (16).
The calculation of cell birth rate and life span assumed a duration of
Ki-67 expression of 24 h, an assumption previously justified in
detail (28). If the duration of expression is shorter, the calculated
cell birth rate would be higher and vice versa. The duration
of the cell cycle is generally longer in malignant tumors than in
normal tissue (10). If this applied also to parathyroid adenomas, the
difference in cell birth rate would be even smaller than we have
calculated. In previous work, the geometric mean rate of cell
proliferation was estimated to be 13.7%/yr in sporadic parathyroid
adenomas, assuming the duration of S phase to be 12 h (3), and
6.4%/yr in radiation-associated parathyroid adenomas, assuming the
duration of mitosis to be 0.5 h (4). Thus, three independent
methods, all using an appropriate sampling procedure, have produced
results that varied over little more than a 2-fold range. Considering
the uncertainties in the estimates of the durations of various
subdivisions of the cell cycle, this is reasonable agreement. Fresh
tissue (3) may have given higher results than archival tissue (4)
(Table 4
) because of unrecognized changes during tissue processing and
storage. It is also possible that the rates are indeed somewhat lower
in radiation-associated than in sporadic adenomas (4), and that such
tumors are closer to their growth asymptote (5).
The proportion of Ki-67-positive cells was approximately 50 times
smaller in the present study than in studies that used selective and
consequently biased sampling (Table 6)
(29, 30). The similarity in LI results when the whole section was
examined (2.89 vs. 2.72) demonstrates that our sampling was
unbiased. We believe that our detection method has been well validated
(8, 28) and that the absence of sampling bias is the most likely
explanation for our lower results. Very high results despite less
biased (although nonrandom) sampling probably reflect an unusually
sensitive method for PCNA that detected more cells undergoing DNA
repair (25). Even higher values for the proportion of cycling cells are
given by flow cytometry (Table 6) (31), but this method is not
applicable to low turnover tissue, partly because of cellular debris
(32) and partly because of inability to distinguish between the
elective tetraploidy of cells that have completed S phase and
constitutive tetraploidy. This is a feature of many normal tissues (2)
and of tumors of several endocrine glands, including the parathyroid
(33, 34).
|
We found a significantly higher proportion of cycling cells in the peripheral than in the central regions of parathyroid adenomas and a significant deviation from random distribution. By contrast, in normal parathyroid glands, there is no difference between peripheral and central regions and close conformity to a Poisson distribution (8). A clustered distribution of cycling cells has been reported in nodular secondary hyperparathyroidism and also in some parathyroid adenomas (24), although this conclusion was based on simple inspection of the sections rather than on statistical evaluation. We did not find such clusters (indeed, the largest number of cycling cells in any graticule field was six), but, rather, observed a general tendency for cell birth rate to be higher closer to the perimeter of the tumors. In a large, rapidly growing, malignant tumor, the central region may be relatively ischemic and hypoxic (40), but there is no reason to suppose that blood supply would be compromised in the center of a small, slowly growing, benign tumor of a highly vascular tissue (41), and there was no histological evidence for such a phenomenon. A differential cell birth rate appears to be a characteristic of parathyroid tumor growth, for which an explanation is required.
It was previously demonstrated (3, 4) that cell birth rates in
parathyroid adenomas had declined substantially during their life span,
assuming that they arose from a single mutant cell. In
radiation-associated adenomas, the initial cell birth rate can be
estimated by fitting the data to the Gompertz function (5). The
geometric mean and 95% confidence limits of these estimated rates are
compared with observed rates at the time of excision in Table 7. For this purpose, data in the 10
completely mapped cases were used, because the precision of individual
values was greater and the range narrower. The estimated decline in
cell birth rate, based on geometric means, was 20- to 25-fold, but the
ranges were very wide, so that the extent of decline could have varied
substantially between cases. Because of the differential cell birth
rate previously discussed (Table 4), the decline had been greater in
the center than in the periphery of the tumor. Tumors enlarge mainly by
interstitial growth, a process that must be accompanied by cell
displacement, but it is reasonable to assume that the mean age of cells
is greater in the center than in the periphery, which suggests that the
magnitude of decline in cell birth rate in parathyroid adenomas is a
function of cell age.
|
One way out of this dilemma, the only way in our opinion, is provided
by the set-point hypothesis. It is proposed (2, 3, 4, 5, 8, 46) that a
hereditary change in phenotype occurs in a single cell that decreases
its sensitivity to its ambient calcium concentration. This change does
not result from a mutant form of the serpentine calcium sensing
receptor (47, 48), but could result from a reduction in the number of
normal receptors (49), which is the main determinant of the secretory
set-point (50). By whatever means the change in sensitivity was brought
about, the abnormal cell would behave in the same way as a normal cell
exposed to hypocalcemia. It would increase its rate of PTH secretion,
but to no avail, and eventually would be driven to divide. The clone of
new cells would continue to proliferate, but ever more slowly, until it
was large enough to raise the patient’s plasma calcium level to the
new secretory set-point. Because of the very long time interval between
successive divisions, the decline in cell birth rate would be very slow
and would have progressed further in older cells (Table 4). The
increase in cell proliferation in the early stages of tumor growth
would increase the mutagenic risk (51), and so set the stage for the
mutations that have been detected. Although not yet supported by direct
evidence, the set-point hypothesis accounts for the current clinical
characteristics of primary hyperparathyroidism, the main abnormality in
PTH secretion, the low rates of cell proliferation at the time of
surgical excision, the difference in cell birth rate between the
central and peripheral regions, the inferred pattern of asymptotic
growth, and the high prevalence of mutations, all by means of a single
mechanism.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received February 4, 1997.
Revised September 4, 1997.
Revised November 24, 1997.
Accepted December 4, 1997.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  G. J. Strewler A 64-Year-Old Woman With Primary Hyperparathyroidism JAMA, April 13, 2005; 293(14): 1772 - 1779. [Full Text] [PDF]
|
|
|
|
 |
|
 N. Garcia de la Torre, I. Buley, J. A. H. Wass, D. G. Jackson, and H. E. Turner Angiogenesis and Lymphangiogenesis in Parathyroid Proliferative Lesions J. Clin. Endocrinol. Metab., June 1, 2004; 89(6): 2890 - 2896. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. M. Mallya, J. J. Gallagher, and A. Arnold Analysis of Microsatellite Instability in Sporadic Parathyroid Adenomas J. Clin. Endocrinol. Metab., March 1, 2003; 88(3): 1248 - 1251. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. S. Rao, M. Honasoge, G. W. Divine, E. R. Phillips, M. W. Lee, M. R. Ansari, G. B. Talpos, and A. M. Parfitt Effect of Vitamin D Nutrition on Parathyroid Adenoma Weight: Pathogenetic and Clinical Implications J. Clin. Endocrinol. Metab., March 1, 2000; 85(3): 1054 - 1058. [Abstract] [Full Text]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
