| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
Sections of Neurosurgery (J.T.K.), General Internal Medicine
(A.C.J.), and Endocrinology (D.C.A.), Department of Veterans Affairs
Medical Center; and the Department of Neurosurgery (J.T.K.); the
Program in Health Care Research (J.T.K., A.C.J., D.C.A.); the Division
of General Internal Medicine, Department of Medicine (A.C.J., D.C.A.);
the Division of Clinical and Molecular Endocrinology, Department of
Medicine (D.C.A.), Case Western Reserve University and University
Hospitals, Cleveland, Ohio 44106
Decision analysis: A
systematic and quantitative approach to decision making under
conditions of uncertainty. The probabilities of each possible event and
the consequences of those events, given various conditions and
assumptions, are stated explicitly. A mathematical model of a problem
and its possible treatments, incorporating options, probabilities, and
outcomes, is used to calculate the "best" decision.
"Best" is
determined based on expected values, the calculated average outcomes
from the branches of the decision tree.
Decision
tree: A flow chart of a decision and its probabilistic
consequences. The initial choice and outcomes are graphically
represented in branches of the tree. These branching points are either
chance nodes that have biologically determined outcomes or decision
nodes (usually only one) in which the outcomes are decided by choice of
the physician or patient.
Discount rate: The rate
used to compute the present value of a clinical or monetary events that
occur in the future (typically 5ü.
Health-related
quality of life: As a construct, health-related quality of life
(HRQOL) refers to the impact of the health aspects of an individual’s
life on that person’s quality of life or overall well-being; also used
to refer to the value of a health state to an individual.
Incremental cost: The cost of one alternative minus
the cost of another.
Incremental QALY: The number
of QALY of one alternative minus the number of QUALY of another.
Incremental cost/QALY: The cost of one alternative
minus the cost of another for one additional QALY.
Incremental cost-effectiveness (ratio): The ratio of
the difference in costs between two alternatives to the difference in
effectiveness between the same two alternatives.
Marginal benefit: The added benefit generated by the
next unit consumed.
Marginal cost: The added cost
of producing one additional unit of output.
Markov
model: A type of mathematical model containing a finite number of
mutually exclusive and exhaustive health states. During each time
period of uniform length, patients can move from one state to another
based on probability rules.
Present value: The
value to the decision maker now of outcomes occurring in the future. In
economic evaluations (present value analysis), future costs and
benefits are expressed in current dollars. Future costs and benefits
are multiplied by a discount factor to convert them to current dollars.
QALYs: A measure of health outcome that assigns to
each year a weight, ranging from 0–1, corresponding to the quality of
life during that year. By convention, perfect health is assigned a
value of 1.0, and death is assigned a value of 0.0. Total QALYs are
obtained by adding the product of the quality of life value and the
number of years in that state (e.g. quality of life
value = 0.7, yr = 10, QALYs = 0.7 x 10 =
7.0).
%Sensitivity analysis: A method to assess the
effects of key assumptions or values on the final result of a
mathematical model. The assumptions are varied over a range of values
to determine their effects on the result, i.e. a test of the
stability of the conclusions of an analysis over a range of probability
estimates, value judgments, and structural assumptions. Large
differences in effects indicate that the analysis is "sensitive"
to
the assumption. One-way sensitivity analysis varies one variable at a
time. Two-way sensitivity analysis varies two at a time,
etc.
Time horizon: The period of time
for which costs and effects are measured in a cost-effectiveness
analysis.
These definitions were derived and modified from
Refs. 73 and 74.
Address all correspondence and requests for reprints to: David C. Aron, M.D., M.S., Medical Service 111(W), Department of Veterans Affairs Medical Center, 10701 East Boulevard, Cleveland, Ohio 44106. E-mail: aron.david{at}cleveland.va.gov
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionGlossaryReferences |
|---|
A decision analytic Markov model was used to determine the incremental cost-effectiveness of four clinical management strategies: 1) expectant management, 2) PRL screening, 3) an endocrine screening panel (PRL, insulin-like growth factor I, and 1-mg dexamethasone suppression test), and 4) magnetic resonance imaging (MRI) follow-up. The model incorporated the natural history of incidental microadenomas, test characteristics, pharmacological and surgical treatment outcomes, patient’s quality of life, discounting, and the costs of hormone testing, bromocriptine, MRIs, hospitalization for surgery, and physician services.
PRL screening, endocrine screening panel, and MRI follow-up all provided slightly greater quality-adjusted survival than expectant management, but the costs increased disproportionately more than the benefits. The incremental cost per quality-adjusted life year for PRL screening is $1,428, and that for the endocrine screening panel is $69,495. These results are most sensitive to patient anxiety about the microadenoma; increased anxiety shifts the recommended strategy to the endocrine screening panel.
We conclude that in patients with an incidental asymptomatic pituitary microadenoma, a single PRL test may be the most cost-effective management strategy.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionGlossaryReferences |
|---|
Current recommendations for the management of incidental microadenomas are controversial. Some authorities recommend extensive routine biochemical screening to diagnose hormonally active lesions secreting PRL, GH, ACTH, and the glycopeptide hormones (TSH, LH, and FSH) (2, 10, 11). Some also recommend follow-up serial MRI scans to detect microadenoma growth (2). The limited available data suggest that clinical practice varies widely. When surveyed about their management of a hypothetical 25-yr-old woman with a 5-mm incidental pituitary mass, normal menses, no galactorrhea, and no symptoms except headaches, Cleveland metropolitan area endocrinologists recommended a mean of 3.5 hormone tests (range, 0–9) (8).
To determine the cost-effectiveness of various strategies, we developed a decision analytical model based on a review of the endocrinology literature. Our approach was guided by the following questions. 1) Does an incidental microadenoma put the patient at increased risk for an adverse outcome? 2) Can individuals with treatable syndromes be accurately diagnosed? 3) Is the treatment of these syndromes more effective in presymptomatic patients? 4) Do the beneficial effects of presymptomatic detection and treatment of these patients justify the costs incurred?
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionGlossaryReferences |
|---|
A patient with an incidental pituitary microadenoma is
susceptible to three types of adverse outcomes: 1) mortality; 2)
endocrinological or neurological morbidity; or 3) anxiety from knowing
about a tumor that might cause problems in the future. Symptomatic
pituitary tumors are quite rare; epidemiologic studies show a
prevalence of 0.00020 and an incidence of 0.00002 (Table 1
) (12, 13, 14). Fewer than 50% of these
diagnosed cases were microadenomas. Screening studies show that
prolactinomas, the most common hormone-secreting tumors of the
pituitary, have a prevalence of less than 0.0005 (15, 16, 17). (Note that
not all prolactinomas are symptomatic.) Acromegaly and Cushing’s
disease are less common, with prevalences of 0.00007 and 0.00004,
respectively (17, 18, 19, 20). The prevalence of pituitary tumors that secrete
TSH, LH, or FSH are even lower (21, 22, 23, 24). Nonsecreting pituitary
lesions, such as primary pituitary tumors, sarcoidosis, and metastases,
do not cause excess morbidity or mortality unless one or more of the
following is present: pituitary hormone deficiency, hyperprolactinemia
due to pituitary disinhibition, or neurological changes from mass
effects such as visual field defects. Consequently, these lesions do
not lend themselves to hormonal screening. They are included in the
model only to the extent that that further evaluation would follow the
identification of such clinical manifestations.
|
Studies of the natural history of incidentally discovered pituitary microadenomas are limited. Reincke et al. followed 7 microadenoma patients with normal endocrine function for up to 8 yr. Only 1 showed an increase in tumor size (from 5 to 9 mm), 1 patient showed regression in tumor size from 8 to 4 mm, and 5 patients remained stable (25). Donovan and Corenblum followed 15 endocrinologically normal microadenoma patients for a mean of 6.7 yr. One tumor progressed in size (from 4 to 5 mm), and regression in lesion size was seen in 6 patients (26). The authors of both of these studies and others have recommended observation for microadenomas in concert with hormonal screening (11, 25, 26).
Accuracy of diagnosis
Approaches to the clinical and biochemical diagnosis of hormone-secreting pituitary tumors are well established, but suffer from shortcomings in test sensitivity and specificity (27). The diagnosis of prolactinoma has rested primarily on basal PRL levels and neuroradiological studies for the identification of a sellar mass. Although basal PRL levels above 200 µg/L are virtually diagnostic of a macroprolactinoma, microprolactinomas are often associated with lower PRL levels (28, 29). Moreover, the differential diagnosis of nonphysiological hyperprolactinemia is extensive, and test specificity falls as the diagnostic cut-off point is decreased (30).
The biochemical diagnosis of acromegaly rests on measurements of GH and somatomedin C [insulin-like growth factor I (IGF-I)] (31, 32). Single measurements of GH are not entirely reliable, because GH secretion is episodic and increased GH levels can be observed in other conditions (33). Dynamic testing for GH response to oral glucose administration is more specific, as is a single measurement of IGF-I (33, 34). However, these studies involved only patients with clinically apparent disease. Test characteristics in patients with subclinical disease are unknown, but are likely to be less accurate due to spectrum bias (30, 35).
The clinical diagnosis of Cushing’s syndrome, or persistent inappropriate hypercortisolism, may be difficult (36). Endogenous hypercortisolism is demonstrated by the presence of abnormal cortisol suppressibility with low dose dexamethasone or by increased cortisol secretion as reflected in basal 24-h urinary free cortisol measurements. Both tests have excellent sensitivity, but serious problems with specificity, especially with the low dose dexamethasone suppression test (36). Moreover, the test is likely to be less sensitive in patients lacking classic manifestations of the disease. The relative rarity of clinically diagnosed glycopeptide-producing pituitary tumors makes it difficult to generalize about screening test characteristics.
The high frequency of incidental microadenomas has complicated the radiological evaluation of the pituitary (37, 38, 39). Findings on sella tomograms have been shown to bear little relation to findings at autopsy or surgery (39, 40). The lack of specificity of computed tomographic scanning has also been demonstrated (41, 42). Magnetic resonance imaging has been shown to have superior sensitivity compared with CT in patients with biochemically diagnosed hormone-secreting pituitary tumors (43, 44). Magnetic resonance imaging has approximately 90% sensitivity and a similar specificity for pituitary tumors in patients with most syndromes of pituitary hormone excess; the sensitivity for ACTH-secreting pituitary adenomas is, at best, 60–75% (45). A recent report of 100 healthy volunteers using high resolution MRI scanning found a 10% prevalence of focal pituitary abnormalities (7). Moreover, closer examination of the data indicated that findings in postgadolinium films were interpreted to be indicative of pituitary micoadenoma by at least one of the three radiologists in 34 of 70 women (48.6%) and 8 of 30 men (26.7%). Similar findings have been reported in another small series (46). In the absence of studies correlating MRI and histopathological findings, one cannot be certain that the MRI lesions represent microadenomas. Thus, only inferences about their natural history can be drawn.
Effectiveness of treatment
Treatment for pituitary microadenomas depends upon the endocrinological activity of the tumor (8). Prolactinomas are usually treated initially with bromocriptine, which reduces PRL secretion and tumor volume. Bromocriptine effectively treats endocrinological or neurological symptoms in 90% of patients with symptomatic prolactinomas (11). The effects of bromocriptine on patients with asymptomatic prolactinomas are unknown.
Patients who evidence microadenoma growth by neuroimaging or who
develop endocrinological or neurological symptoms refractory to
bromocriptine are candidates for surgery. The transsphenoidal route is
the surgical approach of choice to pituitary microadenomas (8).
Morbidity, mortality, and cure rates from transsphenoidal surgery for
pituitary adenomas vary with the underlying endocrinopathy or
neurological symptoms (Table 1
) (47, 48). External beam radiation and
stereotaxic radiosurgery have been used for recurrent or inoperable
tumors, but are considered second line therapies in patients with
microadenomas (49). Little is known about whether presymptomatic
treatment for patients with biochemical endocrinopathies is
beneficial.
Cost-effectiveness of the evaluation and treatment of incidental pituitary microadenomas
We used cost-effectiveness analysis to compare four strategies
for the management of an asymptomatic patient with an incidental
pituitary microadenoma: 1) expectant management, 2) PRL screening, 3) a
panel of hormone screening tests (PRL, IGF-I, and 1 mg dexamethasone
suppression of cortisol), and 4) follow-up MRI screening. A schematic
diagram of our decision tree is shown in Fig. 1. (A technical report with the complete
decision tree is available to interested readers.)
|
PRL screening. Patients undergo only a single PRL screening, and hyperprolactinemia confirmed by a repeat test is treated with bromocriptine. Those in whom bromocriptine is not effective are followed and undergo surgery if they become symptomatic. Patients with normal screening PRL assays undergo no further testing or treatment unless they develop endocrinological or neurological symptoms in the future.
Endocrine panel. Patients are screened for abnormal levels of PRL, IGF-I, and cortisol after 1 mg dexamethasone administration. Similar to the PRL screening scenario, confirmed hyperprolactinemia is initially treated with bromocriptine; those in whom bromocriptine is ineffective are followed and undergo surgery if they become symptomatic. Confirmed elevations in either cortisol (using 24-h urinary free cortisol and plasma ACTH) or GH (via repeat IGF-I) result in surgery. Patients with normal screening PRL, IGF-I, and 1 mg dexamethasone suppression tests undergo no further testing or treatment unless they develop endocrinological or neurological symptoms in the future.
MRI follow-up. Enlargement of the incidental pituitary microadenoma on follow-up MRI scans at 6 and 12 months precipitates hormone screening and subsequent pharmacological or surgical treatment. Patients with enlarging tumors undergo endocrine testing. Enlarging PRL-secreting tumors are initially treated with bromocriptine. Patients with enlarging tumors that do not secrete PRL or that fail to respond to bromocriptine treatment undergo surgery. If MRI scans performed at 6 and 12 months show no evidence of growth, no further workup is performed unless the patient develops endocrinological or neurological symptoms in the future.
Quality of life values were assigned to each of the health states in
our model (Table 1) (50, 51, 52, 53). The clinical effectiveness of each
management strategy was expressed in quality-adjusted life years
(QALYs). A societal perspective was used to estimate the direct medical
costs (e.g. drugs, tests, hospitalizations, physician fees,
etc.) associated with each strategy. Costs, expressed in
1995 U.S. dollars, were estimated using Medicare diagnostic related
groups, Medicare physician fee schedules, and the Drug Topics Red Book
(Table 2) (54, 55, 56). Indirect costs,
i.e. earnings lost as a result of disability or death, were
not included in the analysis. We defined the incremental
cost-effectiveness of the various strategies as the net change in costs
divided by the net change in clinical effectiveness associated with
each scenario. Because many of the costs and benefits in our model
accrue in the future, we used a 10-yr time horizon and a 5% discount
rate to estimate their present values (57, 58, 59, 60). The incremental
cost-effectiveness was calculated using Markov modeling techniques (61, 62) and commercially available software (data version 3.0.4, TreeAge
Software, Boston, MA). Baseline assumptions of the model were tested in
one- and two-way sensitivity analyses by altering the input values of
variables and observing the impact on the model’s conclusions.
|
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionGlossaryReferences |
|---|
). This
slight difference results from the interplay of patient anxiety from
their incidental pituitary microadenoma, low positive rates of
screening tests, low incidence of endocrinological and neurological
disease, and effective treatments with low morbidity and mortality. The
incremental costs ranged from $34 for PRL screening relative to the
least costly expectant management strategy, to $1549 for MRI follow-up
relative to the endocrine panel strategy (Table 3). The cost variation
derives primarily from the large difference in cost between hormone
testing and MRI scanning.
|
).
Compared to expectant management, the marginal cost per QALY of PRL
screening is $1,428. The additional QALY benefit provided by extending
PRL screening to include tests for acromegaly and Cushing’s syndrome
has a marginal cost per QALY of $69,495. MRI follow-up is both less
effective and more expensive than either hormone testing strategy, and
thus by definition is dominated by these strategies. Considered another
way and not making adjustments for quality of life, evaluation of a
hypothetical cohort of 10,000 patients at 10 yr would result in 99.92
deaths with the expectant management strategy, 99.72 deaths with the
PRL strategy, 99.84 deaths with the endocrine panel strategy, and 99.74
deaths with the MRI follow-up strategy.
Sensitivity analyses were performed with the ranges of values listed in
Table 1. One-way sensitivity analyses showed that our results were
highly dependent on patient anxiety (Fig. 2) and relatively unaffected by time
horizon, disease prevalence, surgical outcomes, or MRI or laboratory
costs. The results were also moderately sensitive to the diagnostic
test characteristics in asymptomatic patients. Two-way and
multianalyses showed similar results for disease prevalence and
surgical outcomes. Our baseline assumption is a 1% decrease in quality
of life from perfect health, i.e. a value of 1.0' due to
anxiety in patients who know that they harbor a microadenoma
(i.e. quality of life value = 0.99; see Table 1). PRL
screening is the only cost-effective strategy in these individuals.
Figure 2 shows that the cost-effectiveness of each of the strategies
increases as patient anxiety increases. At a quality of life value of
0.98, the endocrine panel meets the $50,000/QALY threshold. At a
quality of life value of 0.97, the MRI follow-up strategy meets the
$50,000/QALY threshold. However, at all levels, the endocrine panel was
most effective, i.e. resulted in the most QALYs.
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionGlossaryReferences |
|---|
In the sensitivity analyses this recommendation is quite stable, even when most model input values are altered within reasonable ranges. The exception is patient anxiety; the recommendation of a single screening PRL level is predicated upon slight patient anxiety concerning the microadenoma. By comparison, rules of thumb attribute a quality of life value of 0.98–0.995 for moderate health problems (64). In a study of patients with the diagnosis of testicular cancer who were disease free for less than 2 yr, the quality of life value determined by the time trade-off method was found to be 0.95 after orchidectomy and 0.97 after chemotherapy (65). Similarly, a value of 0.95 was suggested as quality of life value for a patient with an asymptomatic unruptured intracranial aneurysm (66). These data suggest that our choice of a value of 0.99 for the baseline case is not unreasonable. The majority of the QALY benefit from the testing strategies compared to expectant management is from anxiety reduction. Most patients who undergo endocrine or radiological screening will have negative test results, which provides reassurance and improves quality of life. In the expectant management strategy, only the "tincture of time" provides this anxiety relief. Thus, if a patient has no anxiety, there is no reassurance from a negative screening test, and expectant management is the preferred strategy. As patient anxiety increases, the negative test results from the active screening strategies become increasingly beneficial, and the endocrine panel becomes a cost-effective strategy. Physician framing can significantly influence patient attitudes toward disease and treatment (67, 68, 69, 70). Physicians must take care not to create inappropriate anxiety in patients by overemphasizing the importance of an incidental finding unless it is associated with an established clinical risk. We suggest that if there is no increased risk, this should be conveyed to the patient by their physician, which should alleviate anxiety. If there is a small amount of risk, there is reason for the patient to feel a small amount of anxiety. At the same time physicians must recognize the uncertainty that results from the limitations of the available data; the absence of evidence of an outcome is not the same as evidence of the absence of that outcome.
Our review of the endocrinological literature suggests that patients with an incidental microadenoma have a slightly increased risk of morbidity and mortality, which implies a benefit of early diagnosis. Unfortunately, our ability to identify the approximately 0.5% of incidental microadenoma patients at increased risk for endocrine or neurological dysfunction is poor. As a result, many patients are subjected to unnecessary testing and treatment, which carry their own set of risks. Black and Welch have addressed the advances in diagnostic imaging and overestimation of disease prevalence and the benefits of therapy (71). The identification of an incidental pituitary microadenoma prompts endocrinologists to perform multiple hormonal screening tests (8). However, extensive testing is expensive and may result in further expense and harm. As false positive results are pursued, a cascade effect is produced, in which a "chain of events tends to proceed with increasing momentum, so that the further it progresses the more difficult it is to stop" (72).
A patient with an incidental microadenoma has a slightly increased risk of morbidity and mortality, primarily from the complications of a prolactinoma. These patients should undergo a single PRL screening test. This test will be normal in most cases. A normal screening PRL test can effectively reassure both physician and patient that subclinical disease is not present. Reassurance is not a trivial benefit; the sensitivity analysis demonstrates the importance of patient anxiety in determining the optimal management of an incidental microadenoma. Abnormal PRL screening tests should be confirmed, and the appropriate pharmacological or surgical treatment initiated. Because of the extremely low prevalence of other endocrinopathies, further screening in patients with normal PRL levels confers minimal benefit and is not cost-effective. MRI follow-up is quite expensive and in asymptomatic microadenoma patients provides no better clinical results than hormone screening. Both prudence and the results of our cost-effectiveness analysis suggest that a single PRL test, repeated only if the initial test is abnormal, may be the most appropriate approach to the incidental pituitary microadenoma.
|
Glossary |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionGlossaryReferences |
|---|
|
Footnotes |
|---|
Received April 25, 1997.
Revised July 9, 1997.
Accepted July 17, 1997.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionGlossaryReferences |
|---|
-subunit hypersecretion in patients with pituitary
tumors: clinically non-functioning and somatotroph adenomas. J
Clin Endocrinol Metab. 70:859–864.[Abstract]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 |
