Which assay using 24-hour urine is considered the best single screening test for pheochromocytoma?

• PDF
• Split View
• • Article contents
  • Figures & tables
  • Video
  • Audio
  • Supplementary Data

CATECHOLAMINE-SECRETING TUMORS ARE rare neoplasms of chromaffin cells that arise from the adrenal medulla (pheochromocytoma) or paraganglia (paraganglioma). Biochemical testing for a catecholamine-secreting tumor is typically performed as part of an evaluation for secondary causes of hypertension, unexplained spells, incidental adrenal masses, or, less commonly, patients with a family history of pheochromocytoma.

Biochemical testing for a catecholamine-secreting tumor typically has included measurements of 24-h urinary excretion of total metanephrines and catecholamines (1, 2). Recently, measurements of fractionated plasma metanephrines were found to have high sensitivities and specificities for detecting pheochromocytoma and have been recommended as the single biochemical test of choice (3–7). Our aim was to determine the diagnostic efficacy of fractionated plasma metanephrine measurements compared with those of urinary total metanephrines and urinary catecholamines in detecting adrenal and extra-adrenal catecholamine-secreting tumors in an outpatient tertiary-care setting.

Subjects and Methods

Experimental subjects

A computerized search identified all patients at Mayo Clinic Rochester in whom fractionated plasma metanephrines and 24-h urinary total metanephrines were measured from January 1, 1999, until November 27, 2000, and their medical records were reviewed. Hospitalized patients were excluded secondary to potential physiologically appropriate elevations of catecholamines and their metabolites. We identified 349 consecutive outpatients (including 33 patients with histologically confirmed pheochromocytoma or paraganglioma) in whom fractionated plasma metanephrines and 24-h urinary total metanephrines, with or without urinary catecholamines, were ordered and measured concurrently (24-h urinary catecholamine measurements were available in 292 patients).

The criteria for performing the plasma and urinary measurements were: hypertension alone (refractory, new onset, severe, or recent unexplained exacerbation), spells (headache, palpitations, lightheadedness, excessive perspiration, pallor, or flushing) with or without sustained or paroxysmal hypertension, adrenal abnormality (adrenal mass) on imaging, and patients at high risk (high-risk familial syndromes, previous surgically cured pheochromocytomas or paragangliomas). All patients with catecholamine-producing tumors had histologic confirmation. All patients without pheochromocytoma were assigned a different clinical diagnosis by their treating physician at the completion of their evaluation.

The Institutional Review Board of Mayo Foundation approved the study, and signed research authorization was verified for all medical records used. There was no sponsor involvement or funding for the study.

Biochemical assays

Fractionated plasma metanephrines were measured by the Endocrine Laboratory at Mayo Medical Center using liquid chromatography with electrochemical detection and reported as metanephrine and normetanephrine fractions (8). Blood was drawn in the Mayo Clinic Endocrine Testing Center from all 349 patients for measuring fractionated plasma metanephrines. Plasma normetanephrine measurements were available in all 349, but the plasma metanephrine fraction was not reported by the laboratory for 6 patients because of interfering substances. Acetaminophen is a potential interfering substance, and the test was delayed until acetaminophen was withdrawn, if possible, or noted in the patient chart if used in the previous 48 h. No other medications were routinely withdrawn as part of this testing protocol. If interfering substances were noted by the laboratory upon measurement of plasma or urinary values, no specific testing was performed to identify these substances.

Twenty-four-hour urinary catecholamines were also measured by liquid chromatography and electrochemical detection (in 292 patients), whereas urinary total metanephrines were measured by spectrophotometry (in all patients), both at Mayo Medical Laboratories (9–11). For six patients who had urinary catecholamines measured, an epinephrine fraction was not available because of interfering substances, although norepinephrine and dopamine measurements were available in those cases and included in the analyses. For urinary total metanephrines, a normal bell-shaped spectral curve has its maximum absorption at 347 nm, and maximal absorption at a different wavelength could be consistent with drug interference. The spectral curve was abnormal in samples from 27 patients; hence, urinary total metanephrine measurements were not available for these patients.

Analysis of measurements

For fractionated plasma metanephrines, a plasma metanephrine value of at least 0.5 nmol/liter or a plasma normetanephrine value of at least 0.9 nmol/liter was considered positive, on the basis of a reference range established by Mayo Medical Laboratories. Similarly, a urinary total metanephrine content of at least 6.6 μmol/24 h (≥1.3 mg/24 h) was considered positive, based on our institutional experience (12). For urinary catecholamines, a 24-h urinary content of norepinephrine greater than 1005 nmol (>170 μg), epinephrine greater than 191 nmol (>35 μg), or dopamine greater than 4571 nmol (>700 μg) was considered positive, also based on our institutional experience (12). A positive 24-h urinary total metanephrine and catecholamines test result was defined by either the urinary total metanephrine or any of urinary catecholamine fraction measurements being increased above the set cut-offs.

Statistical analysis

Sensitivities and specificities of plasma and urinary measurements (using the cut-offs of positivity described above) were compared using McNemar’s test (SPSS 10, SPSS, Inc., Chicago, IL; Ref. 13). Comparisons of sensitivities and specificities were only performed in patients who underwent measurement of fractionated plasma metanephrines as well as 24-h urinary measurements of total metanephrines and catecholamines. For sensitivities and specificities, 95% confidence intervals (CI) were calculated using Wilson’s method (14). The association of continuous variables with outcomes was evaluated using logistic regression, and the association of nominal variables with outcomes was evaluated using Fisher’s exact test for all patients who underwent measurement of fractionated plasma metanephrines (JMP version 4, SAS Institute, Inc., Cary, NC). Spearman rank correlations were used to identify associations between variables for all subjects who had fractionated plasma metanephrine measurements (JMP version 4). Likelihood ratios for positive tests (and their 95% CI), representing the likelihood of a positive result in patients with disease divided by the likelihood of a positive test result in patients without disease, were calculated using the log method (for patients who underwent measurement of plasma and both urinary parameters; Refs. 14–16). The post-test probability (percentage chance that an individual patient with a positive test has the disease) was then calculated for several clinical scenarios, using prevalence (pretest probability) estimates from the literature and the calculated likelihood ratios (15, 16). Receiver-operating characteristic (ROC) curves were constructed for individual plasma and urinary measurements as well as scores derived from multiple logistic regression, incorporating the combination of plasma measurements and the combination of urinary measurements, respectively, and the areas under the curve (AUC) were calculated (SPSS 10, SPSS, Inc.; Ref. 17). Only patients for whom measurements of all plasma and urinary variables were available were included in the logistic regression model (24 patients with pheochromocytoma and 234 patients without pheochromocytoma). All reported P values were two-sided.

Results

The study group consisted of 349 outpatients, including 33 patients with histologically proven catecholamine-secreting tumors. In patients without catecholamine-secreting tumors, an alternative clinical diagnosis was noted. The median age of patients studied was 52 yr (range, 10–83 yr), including 159 females and 190 males. Of the 33 patients with catecholamine-secreting tumors, 16 were extra-adrenal and 17 were malignant; 8 were associated with familial syndromes [3 familial malignant paragangliomas, 2 von Hippel-Lindau disease, 1 multiple endocrine neoplasia (MEN) type 2A (MEN 2A), 1 MEN 2B, and 1 familial multiple benign paraganglioma]. Distributions of all plasma and urinary measurements in patients with or without adrenal or extra-adrenal catecholamine-secreting tumors are shown in Figs. 1 and 22. The patients with adrenal or extra-adrenal catecholamine-secreting tumors were combined in the category of pheochromocytoma in the rest of the study.

Open in new tabDownload slide

Plasma metanephrine (A) and normetanephrine (B) values in pheochromocytoma and those without pheochromocytoma. Dashed line marks the upper limit of normal. ▵, Adrenal pheochromocytoma; ▴, extra-adrenal catecholamine-secreting tumor; ○, no pheochromocytoma.

Open in new tabDownload slide

Twenty-four-hour urinary excretion of total metanephrines (A), norepinephrine (B), epinephrine (C), and dopamine (D) in patients with pheochromocytoma and those without pheochromocytoma. Dashed line marks the upper limit of normal. ▵, Adrenal pheochromocytoma; ▴, extra-adrenal catecholamine-secreting tumor; ○, no pheochromocytoma.

The sensitivity of fractionated plasma metanephrine measurements was 97% (30 of 31 patients; 95% CI, 84–99%), compared with a sensitivity of 90% (28 of 31 patients; 95% CI, 75–97%) for measurements of urinary total metanephrines in combination with urinary catecholamines (P = 0.63). Of note, in a subgroup of patients with sporadic catecholamine-secreting tumors, the sensitivity of measurements of fractionated plasma metanephrines was identical to that of the combination of 24-h urinary total metanephrines and catecholamines at 96% (23 of 24 patients). Measurements of fractionated plasma metanephrines were, however, significantly less specific at 85% (221 of 261 patients; 95% CI, 80–89%), compared with measurements of urinary total metanephrines and catecholamines, which had a specificity of 98% (257 of 261 patients; 95% CI, 96–99%; P < 0.001). The likelihood ratio for a positive test was 6.3 (95% CI, 4.7–8.5) for fractionated plasma metanephrine measurements and 58.9 (95% CI, 22.1–156.9) for the combination of urinary total metanephrine and catecholamine measurements (Table 1). In contrast, the likelihood ratio for a negative test was 0.04 (95% CI, 0.006–0.26) for fractionated plasma metanephrine measurements and 0.10 (95% CI, 0.03–0.29) for the combination of urinary total metanephrine and catecholamine measurements.

Table 1.

Comparison of diagnostic efficacy of biochemical assays for detection of pheochromocytoma

Among the 316 patients without pheochromocytoma, 47 had false-positive measurements of fractionated plasma metanephrines (normetanephrine fraction increased in 40 of 47). In patients without pheochromocytoma, increasing age was associated with false-positive fractionated plasma metanephrine measurements (P = 0.008) and correlated with increasing levels of plasma normetanephrine (r = 0.249; P < 0.001) and plasma metanephrine (r = 0.126; P = 0.03; Fig. 3). Characteristics of the 47 patients with false-positive measurements of fractionated plasma metanephrines are shown in Table 2.

Open in new tabDownload slide

Correlation of age with plasma normetanephrine in nonpheochromocytoma patients. The dashed line indicates the upper limit of normal.

Table 2.

Characteristics of patients with false-positive measurements of fractionated plasma metanephrines

An extra-adrenal paraganglioma was missed by plasma screening in one patient who had a dopamine-secreting paraganglioma of the neck and an elevated 24-h urinary dopamine measurement. Of the patients with false-negative urinary total metanephrine and catecholamine values, all three had adrenal pheochromocytomas [two had familial syndromes (MEN 2A and MEN 2B), and one had an incidentally discovered vascular adrenal mass]. Under collection of the 24-h urinary specimens was unlikely because the concurrently collected urinary creatinine measurements were appropriate for weight in these cases. The two patients with MEN had relatively small tumors, less than or equal to 2 cm in diameter, and the patient with the incidentally found adrenal mass had a 4.5 × 2.2 × 2.2-cm tumor. None of these three patients was taking any antihypertensive medication. The systolic blood pressure of the subject with MEN 2A was 94/52 mm Hg, that of the subject with MEN 2B was 136/90 mm Hg, and that of the patient with an incidentaloma was 148/90 mm Hg at the time of initial assessment. The plasma metanephrine measurements were 0.74, less than 0.2, and 0.5 nmol/liter with normetanephrine measurements of 0.76, 1.76, and 0.9 nmol/liter, respectively, in the subjects with MEN 2A, MEN 2B, and the incidentaloma.

The ROC curves for measurements of plasma metanephrine and normetanephrine fractions as well as 24-h urinary measurements of total metanephrines, norepinephrine, epinephrine, and dopamine are shown in Fig. 4A. The ROC curve generated by logistic regression for plasma metanephrine fractions (AUC, 0.965; 95% CI, 0.918–1.013), compared with the combination of 24-h urinary metanephrine and catecholamine measurements (AUC, 0.979; 95% CI, 0.953–1.005), is shown in Fig. 4B. Of note, when the combined plasma score and combined urinary score generated by logistic regression were compared at the same sensitivity of 92% (95% CI, 74–98%; 22 of 24 patients) using the ROC curve, the specificity of fractionated plasma metanephrines was 94% (95% CI, 90–96%; 213 of 234 patients) and was not significantly different from that of combined urinary measurements at 96% (95% CI, 92–98%; 224 of 234; P = 0.302).

Open in new tabDownload slide

ROC curves for 24 patients with pheochromocytoma and 234 patients without pheochromocytoma. A, The AUC values for individual measurements were: plasma normetanephrine [(yellow line) AUC, 0.959; 95% CI, 0.915–1.004; P < 0.001 for the difference from the null hypothesis of the AUC of 0.5 for the reference line (dotted line)]; plasma metanephrine fraction [(light blue line) AUC, 0.670; 95% CI, 0.527–0.813; P = 0.006]; urinary total metanephrines [(red line) AUC, 0.901; 95% CI, 0.800–1.002; P < 0.001]; urinary norepinephrine [(green line) AUC, 0.860; 95% CI, 0.746–0.973; P < 0.001]; urinary epinephrine [(dark blue line) AUC, 0.579; 95% CI, 0.421–0.738; P = 0.081]; and urinary dopamine [(purple line) AUC, 0.592; 95% CI, 0.465–0.719; P = 0.138). B, ROC curves of combined plasma compared with combined urinary measurements generated by logistic regression. For plasma measures (green line), the AUC was 0.965 [95% CI, 0.918–1.013; P < 0.001 compared with reference line (dotted line) of AUC 0.5]. For urinary measures (red line), the AUC was 0.979 [95% CI, 0.953–1.005; P < 0.001 compared with reference line (dotted line)].

Discussion

Catecholamine-secreting tumors release variable amounts of norepinephrine, epinephrine, and dopamine, with subsequent conversion of norepinephrine and epinephrine to normetanephrine and metanephrine by catechol-O-methyl-transferase (18). Pheochromocytomas secrete heterogeneous patterns of catecholamines and their metabolites, so it has been traditionally recommended that more than one analyte be measured concurrently (19). Yet, it has been recently suggested that a single measurement of fractionated plasma metanephrines may be superior to combinations of other biochemical tests in detecting or excluding pheochromocytoma (7). Measurement of fractionated plasma metanephrines is attractive, given the high sensitivity and convenience of this test compared with combined 24-h urinary measurements. Furthermore, in considering the direct cost of biochemical testing, measurement of fractionated plasma metanephrines is less expensive than combinations of other biochemical tests (20). However, in choosing a biochemical testing strategy, one must consider not only the direct costs of the assays, but also the costs of resultant imaging, patient anxiety over false-positive testing, time lost from work, surgical costs for potentially needless surgery, travel, consultation costs and, of course, potential lives saved. Thus, the estimation of the costs of a biochemical testing strategy can be more complex than a simple estimate of assay costs, particularly with testing strategies that lack specificity.

It is clinically relevant to consider the individual patient’s post-test probability of pheochromocytoma, given a positive biochemical test, in different clinical scenarios. One may estimate the post-test probability of disease incorporating the positive likelihood ratios of 6.3 for fractionated plasma metanephrines and 58.9 for urinary total metanephrines with catecholamines, respectively, as prevalence of pheochromocytoma quoted from the literature [0.5% among screened hypertensive patients (Ref. 21); 5.1% among incidentally discovered adrenal masses at least 1 cm in diameter in absence of symptoms of adrenal disease (adrenal incidentalomas; Ref. 22); and 42% among patients with MEN 2A syndrome presenting with medullary thyroid cancer (Ref. 23)]. For a patient with positive fractionated plasma metanephrines, the post-test chance of pheochromocytoma would be as follows: 3% in the patient with hypertension, 25% in the patient with an adrenal incidentaloma, or 82% in the patient with MEN 2A. Similarly, for a patient with positive urinary total metanephrines or fractionated catecholamines, the post-test chance of having a pheochromocytoma would be as follows: 23% in the patient with hypertension, 76% in the patient with an adrenal incidentaloma, or 98% in the patient with MEN 2A. The above examples illustrate the finding that a positive combined urinary measurement (either a positive fractionated urinary total metanephrine or catecholamine value) is more meaningful in ruling in the diagnosis of pheochromocytoma than a positive fractionated plasma metanephrine measurement, given the high specificity of urinary testing.

One may also estimate the post-test probability of disease in the above clinical circumstances given a negative test result (using the negative likelihood ratio of 0.04 for fractionated plasma metanephrine measurement and 0.10 for the combination of urinary metanephrine and catecholamine measurements). For a patient with negative fractionated plasma metanephrines, the post-test chance of pheochromocytoma would be estimated to be 0.02% in the patient with hypertension, 0.2% in the patient with an adrenal incidentaloma, or 2.7% in the patient with MEN 2A. Similarly, for a patient with normal urinary total metanephrines and fractionated catecholamines, the post-test chance of having a pheochromocytoma would be 0.05% in the patient with hypertension, 0.5% in the patient with an adrenal incidentaloma, or 6.6% in the patient with MEN 2A. The above examples indicate that negative measurements of fractionated plasma metanephrines or 24-h urinary metanephrines with catecholamines are effective in ruling out the diagnosis of pheochromocytoma.

There are several limitations of our study. We performed a retrospective study of limited sample size, which may have affected power to detect a difference in sensitivities. Furthermore, urinary catecholamine measurements were not available in all patients. Also, our study was only conducted in one center, so generalizability may be limited, although it is notable that the biochemical measurements were performed in a national reference laboratory that is widely used in the United States. Furthermore, absence of a catecholamine-secreting tumor was based on experienced clinicians’ evaluations and alternative clinical diagnoses, not histologic confirmation by blinded pathologists. Moreover, patients with an abnormal spectral curve or drug interference as reported by the laboratory were excluded from the analyses. Under collection of 24-h urinary specimens could have resulted in falsely low 24-h urinary total metanephrine and catecholamine measurements. One of the strengths of our study is that screening was carried out in a broad spectrum of patients with and without the disorder, including those with early manifestations of the disease (preclinical catecholamine-secreting tumors) and patients with a variety of different disorders that can be confused with pheochromocytoma (such as refractory essential hypertension, adrenal incidentaloma, and anxiety disorder). Moreover, histologic confirmation was obtained in all cases of catecholamine-secreting tumors.

In conclusion, we have shown that measurements of fractionated plasma metanephrines are highly sensitive but lack specificity when compared with the combination of 24-h urinary total metanephrines and catecholamines. Yet, we observed that false-negative urinary measurements may be seen in asymptomatic patients with a vascular adrenal mass or a high-risk familial syndrome. In contrast, a negative fractionated plasma metanephrine measurement effectively ruled out the diagnosis of pheochromocytoma, even in high-risk individuals, with the possible exception of a dopamine-secreting tumor. Thus, we suggest that measurement of fractionated plasma metanephrines may be the biochemical test of choice in high-risk patients (those with a familial syndrome or vascular adrenal mass). However, in the more common clinical setting when sporadic pheochromocytoma is sought, particularly older hypertensive patients, measurement of 24-h urinary metanephrines and catecholamines may provide adequate sensitivity, with a lower rate of false-positive tests. Further research should focus on which patients benefit the most from testing (and in whom testing can be deferred) as well as performance of different tests and different positivity cut-offs in different risk groups.

Acknowledgements

We thank Dr. Charlie Goldsmith and Dr. Douglas G. Altman for advice on the calculation of likelihood ratios and confidence intervals.

Abbreviations:

 
 
 

• multiple endocrine neoplasia;

 
 

• receiver operating characteristic.

3

, , , , , , , ,

Plasma metanephrines in the diagnosis of pheochromocytoma.

4

, , , , , , ,

Diagnostic efficacy of unconjugated plasma metanephrines for the detection of pheochromocytoma.

5

, , , , ,

Plasma normetanephrine for detecting pheochromocytoma in von Hippel-Lindau disease and multiple endocrine neoplasia 2.

6

, , , ,

NIH conference: recent advances in genetics, diagnosis, localization, and treatment of pheochromocytoma.

7

, , , , , , ,

Biochemical diagnosis of pheochromocytoma: which is the best test?

16

, , Sackett DL; for the Evidence-Based Medicine Working Group

Users’ guides to the medical literature. III. How to use an article about a diagnostic test. B. What are the results and will they help me in caring for my patients?

18

, , , , , , ,

Regional release and removal of catecholamines and extraneuronal metabolism to metanephrines.

20

Special features