Measurement of urinary metanephrine levels is more sensitive than measurement of urinary vanillylmandelic acid or plasma catecholamine levels [2-5]. However, metanephrine excretion is under- or overestimated when the period of urine collection is shorter or longer than 24 hours; underestimation leads to false-negative results, and overestimation leads to false-positive results. Such errors can be avoided by indexing urinary metanephrine levels by urinary creatinine levels. We retrospectively compared the operating characteristics of two tests used to diagnose pheochromocytoma: 1) measurement of the ratio between urinary metanephrine and creatinine levels and 2) measurement of urinary metanephrine levels. We also analyzed the causes of false-positive test results.

Methods

Top Methods Results Discussion Author & Article Info References

Between November 1992 and October 1993, 1831 patients were referred by their physicians or other hospitals for a work-up of hypertension. Urinary metanephrine levels were assayed in 1013 patients who were considered to be at risk for pheochromocytoma because their hypertension was severe, resistant, or associated with diabetes or paroxysmal symptoms [1, 2]. Urinary metanephrine levels were measured by liquid chromatography with electrochemical detection [6]; urinary creatinine levels were measured by the Jaffe reaction [7] in an acidified 24-hour urine sample. In our laboratory, the upper limit of normal for urinary metanephrine levels is 3.70 µmol/d [6]. The presence of pheochromocytoma was confirmed at surgery. In patients with slightly increased metanephrine excretion, the absence of pheochromocytoma was documented by negative results of retesting or negative findings on one or more of the following imaging procedures: adrenal ultrasonography, computed tomography of the abdomen and pelvis, digitalized angiography, and whole-body metaido-benzyl guanidine (MIBG) scintigraphy.

Values are expressed as the mean ± 1 SD. We used the unpaired Student t-test for comparisons between groups. Regression lines were fitted by the least-squares method. Multiple linear regression analysis was used for multivariate analysis, and between-test agreement was assessed by calculating the score. The probability of a type 1 error was set at 0.05.

Results

Top Methods Results Discussion Author & Article Info References

Of the 1013 patients, 20 had pheochromocytoma and 993 (including 16 with a history of resected pheochromocytoma) were considered free of the disease (Table 1). Among the patients without pheochromocytoma, urinary metanephrine and creatinine levels were higher in men than in women (metanephrine level, 1.95 ± 0.92 µmol/d in men compared with 1.55 ± 0.80 µmol/d in women [P < 0.001]; creatinine level, 13.4 ± 4.1 mmol/d in men compared with 9.0 ± 2.9 mmol/d in women [P < 0.001]), but metanephrine-to-creatinine ratios were lower (0.152 ± 0.074 in men compared with 0.181 ± 0.090 in women; P < 0.001). Urinary levels of metanephrines and creatinine were correlated (r = 0.417; P < 0.001). Using multiple linear regression, we analyzed one model in which the metanephrine level was the dependent variable and one in which the metanephrine-to-creatinine ratio was the dependent variable. In the first model, age, body mass index, and urinary creatinine levels, but not sex, were significant predictors of urinary metanephrine levels (multiple correlation coefficient, 0.454; P < 0.001). In the second model, age and sex, but not body mass index, were significant predictors of the metanephrine-to-creatinine ratio (multiple correlation coefficient, 0.392; P < 0.001).

The upper limit of normal for urinary metanephrine levels used in our laboratory (3.70 µmol/d [6]) corresponded to the mean + 2.23 SDs of the values observed in the study sample. Thus, the mean + 2.23 SDs of the metanephrine-to-creatinine ratio (0.354) was used as the upper limit of normal for this variable. The operating characteristics of the measurement of urinary metanephrine levels and measurement of the metanephrine-to-creatinine ratio were as follows: sensitivity, 95% and 100%; specificity, 98% and 98%; positive predictive value, 46% and 47%; and negative predictive value, 100% and 100% (Table 2). The value was 0.603, indicating moderate agreement between the two tests.

Of the 20 patients with pheochromocytoma, 1 had a normal urinary metanephrine level (3.20 µmol/d) but a high metanephrine-to-creatinine ratio (0.552). Of the 993 patients without pheochromocytoma, 38 had positive urinary test results: Fifteen had high metanephrine levels (range, 3.7 to 6.2 µmol/d), 16 had high metanephrine-to-creatinine ratios (range, 0.361 to 0.618), and 7 had both high metanephrine levels (range, 3.8 to 9.8 µmol/d) and high metanephrine-to-creatinine ratios (range, 0.387 to 0.856). Urinary creatinine levels were higher in patients with false-positive results of the measurement of metanephrine levels than in those with false-positive results of the measurement of the metanephrine-to-creatinine ratio (15.4 ± 6.0 µmol/d compared with 7.7 ± 4.5 µmol/d; P < 0.001). Pheochromocytoma was excluded in these patients in the following manner. Of the 15 patients with high urinary metanephrine levels, 8 had negative results on repeated determinations and all had at least one negative (and no positive) imaging test results (11 patients had ultrasonography, 11 had angiography, 4 had computed tomography, and 2 had MIBG scintigraphy). Of the 16 patients with high metanephrine-to-creatinine ratios, 6 had negative results on repeated determinations and all had at least one negative (and no positive) imaging test results (14 patients had ultrasonography, 8 had angiography, and 3 had computed tomography). Of the 7 patients with high metanephrine levels and high metanephrine-to-creatinine ratios, 1 had an adrenocortical adenoma, 5 had a normal ratio at repeated determination, and 6 had negative imaging test results (4 patients had ultrasonography, 2 had angiography, 5 had computed tomography, and 2 had MIBG scintigraphy).

Fourteen patients with acute conditions likely to induce hyperadrenergia had false-positive results and, therefore, high urinary catecholamine metabolite levels: Four patients had had heart failure, 3 had had stroke, 2 had had respiratory failure, 2 had had cancer during chemotherapy, 1 had diabetic ketoacidosis, 1 had depression, and 1 had an adrenocortical adenoma. Thirteen of these patients had a high metanephrine-to-creatinine ratio, but only one had an isolated high metanephrine level.

Discussion

Top Methods Results Discussion Author & Article Info References

Because pheochromocytoma is a life-threatening disease, biochemical tests to detect the condition should have a sensitivity close to 100%; a high specificity is relatively less important. Biochemical tests for pheochromocytoma are based on the measurement of catecholamine or catecholamine metabolite levels in plasma or urine. The limitations of these tests are well known. For example, many pheochromocytomas intermittently secrete catecholamines. This variable secretion, combined with the short half-life of catecholamines, explains why the result of a single plasma catecholamine assay may be negative if the blood is collected during the intervals between surges of high blood pressure [5]. Urinary tests have the advantage of integrating tumoral secretion over the collection period. Measurement of metanephrine excretion is preferred to measurement of catecholamine or vanillylmandelic acid excretion [1-5] because many pheochromocytomas show methylating activity. Metanephrines are therefore a direct tumor marker as well as an indirect marker of catecholamine release [3, 8]. Use of liquid chromatography to measure urinary metanephrine levels almost eliminates false-positive results that are caused by interference from food or drugs [6, 9]. However, the inaccuracy of urine collection is a serious problem with urinary tests [1, 10]. Lenders and colleagues [10] recently found that measurement of plasma metanephrine levels had a sensitivity for pheochromocytoma of 100%, whereas measurement of urinary metanephrine levels had a sensitivity of 89%. In this study, however, plasma metanephrines were assayed by high-performance liquid chromatography and urinary metanephrine levels were measured using a less reliable colorimetric method [11].

We assayed urinary metanephrines in a large sample of hypertensive adults by using high-performance liquid chromatography and metanephrine excretion normalized for creatinine levels; we thus corrected for inaccurate urine collection. This normalization improved the sensitivity of measurement of urinary metanephrine levels for pheochromocytoma. However, several limitations must be considered when this finding is interpreted. First, sensitivity improved from 95% to 100% because 1 patient with pheochromocytoma had normal urinary metanephrine levels but a high metanephrine-to-creatinine ratio. Second, among 1831 patients referred in 1 year to our academic hypertension unit for a work-up of hypertension, urinary metanephrine levels were determined for 1013 patients (55.3%); 20 of these patients (1.1%) had pheochromocytoma. The proportion of patients tested for pheochromocytoma and the prevalence of the tumor were therefore high in our study sample; this reflects a wide interpretation of the clues for pheochromocytoma [1, 2] and referral bias. Finally, we could not measure metanephrine levels a second time for 13 patients with mildly elevated urinary levels. It is therefore remotely possible that some patients with false-positive urinary test results had ectopic, extra-abdominal tumors associated with mildly increased metanephrine excretion. However, we examined in detail patients who did not have documented pheochromocytoma but did have high urinary metanephrine levels, high metanephrine-to-creatinine ratios, or both. Each patient had had at least one of the following imaging tests: ultrasonography, computed tomography, angiography, and MIBG scintigraphy. The combination of biochemical and radiologic tests and, in one case, negative histologic findings was considered adequate to exclude pheochromocytoma in these 38 patients.

False-positive results of measurement of urinary metanephrine levels probably reflected a urine collection period that exceeded 24 hours. In many cases, false-positive results of measurement of the metanephrine-to-creatinine ratio reflected an acute event with hyperadrenergia. Of the 23 patients with a high ratio, 13 had a disease previously described as being associated with increased catecholamine secretion: heart failure [12], diabetic ketoacidosis [13], neurovascular disorders [14], psychiatric disease [15], and adrenocortical adenoma [16]. Another 4 patients had conditions that are believed to increase neuronal catecholamine secretion: cancer during chemotherapy (2 patients) and severe respiratory failure (2 patients).

Measuring 24-hour urinary metanephrine excretion using liquid chromatography and normalizing this value to creatinine excretion is therefore a simple and sensitive test for pheochromocytoma. However, false-positive results of this test, which reflect true hyperadrenergia, may be found in hypertensive patients who have various acute events and complications.