Findings in the derivation set
The derivation set (from which the age-adjusted score was developed) consisted of 349 consecutive subjects (including 33 people with pheochromocytoma) who underwent fractionated plasma metanephrine measurements as well as 24-hour urinary total metanephrine measurements with or without 24-hour urinary catecholamine measurements in testing for pheochromocytoma at Mayo Clinic Rochester. In the derivation set, 8 of the 33 individuals had clinically-diagnosed genetic syndromes predisposing to pheochromocytoma (three familial malignant paraganglioma, two von Hippel-Lindau, one had multiple endocrine neoplasia 2a, one had multiple endocrine neoplasia 2b, and one had familial multiple benign paraganglioma). The 316 individuals in the derivation set who did not have pheochromocytoma underwent such testing the following reasons: refractory hypertension (174, 55%), spells (periodic episodes of symptoms such as palpitations, headache, or sweating, 124, 39%), adrenal mass (45, 14%), previous pheochromocytoma or known genetic predisposition to pheochromocytoma (24, 8%). The mean age of subjects with pheochromocytoma was 48 years (SD 18 years, range 16 to 60 years), whereas the mean age of subjects without pheochromocytoma was 52 years (SD 15 years, range 10 to 73 years). Six of the 316 subjects without pheochromocytoma in the derivation set did not have a plasma metanephrine fraction recorded secondary to "interfering substances" and therefore were excluded from the analyses. In the derivation set, the sensitivity of traditionally interpreted fractionated plasma metanephrine measurements (using 95% reference ranges) was 93.9% (95% CI, 80.4, 98.3) (31/33 subjects), with a specificity of 85.2% (95% CI, 80.8, 88.7) (264/310 subjects). Baseline characteristics of individuals in the derivation set without pheochromocytoma were compared for individuals who had true negative fractionated metanephrine measurements (n = 264) to those who had false positive results (n = 46) (Table ). The individuals with false positive fractionated plasma metanephrine measurements in the derivation set were significantly older than those with true negative measurements (p = 0.007), whereas blood pressure, antihypertensive medication use, and rates of obstructive sleep apnea were not significantly different between these groups. Thus, age was chosen as an important variable to adjust for in interpretation of fractionated plasma metanephrines and an age-adjusted metanephrine score was developed from the derivation set data using logistic regression (as described in the Methods). At a cut-off value of -1.4752, the sensitivity of the age-adjusted metanephrine score was 90.9% (30/33 patients, 95% CI, 76.4%, 96.9%), with a specificity of 96.8% (300/310 patients, 95% CI, 94.2%, 98.2%). In this derivation set, which included individuals genetically predisposed to pheochromocytoma, one individual with a dopamine-secreting paraganglioma, another patient with a von Hippel-Lindau disease (diagnosed clinically), and a third patient with sporadic pheochromocytoma had false-negative age-adjusted metanephrine scores. The efficacy of the age-adjusted metanephrine score was then validated in the validation set.
Clinical characteristics of subjects without pheochromocytoma from the derivation set
Findings in the validation set
In the validation set of 158 subjects, 23 patients had histologically-proven sporadic pheochromocytoma (17 adrenal, 6 extra-adrenal, 8 malignant). Of the patients with pheochromocytoma, none were known to be genetically predisposed to pheochromocytoma and 14/23 were women (61%). The mean age of subjects with pheochromocytoma was 50 years (SD 16 years, range 16 to 83 years), whereas the mean age of subjects without pheochromocytoma was 55 years (SD 16 years, range 7 to 86 years). Of the 135 subjects without pheochromocytoma, 83 (62%) were women. Reasons for measurement of fractionated plasma metanephrines in the subjects without pheochromocytoma were as follows: hypertension (55, 41%), spells with or without hypertension (44, 33%), an incidentally discovered adrenal mass (20, 15%), and previously surgically cured pheochromocytoma (16, 12%).
In the validation set, the sensitivity of the age-adjusted metanephrine score was the same as the traditional interpretation of fractionated plasma metanephrine measurements at 100% (23/23, 95% CI, 85.7%, 100%). The specificity of the traditional interpretation of fractionated plasma metanephrine measurements was 83.7% (113/135 patients, 95% CI, 76.6%, 89.0%) and the specificity of the age-adjusted plasma metanephrine score was 97.0% (131/135 patients, 95% CI, 92.6%, 98.8%). Thus, the false positive rate with traditional interpretation of fractionated plasma metanephrine measurements was 16.3% (22/135, 95% CI, 11.0%, 23.4%) and with the age-adjusted score it was significantly lower at 3.0% (4/135, 95% CI, 1.2%, 7.4%) (Figure ) (p < 0.001 using McNemar's test).
Figure 1 Percentage of false positive test results (and 95% confidence interval) at 100% sensitivity in using a traditional interpretation of fractionated plasma metanephrine measurements or an age-adjusted metanephrine score. Legend – Pmet(s), fractionated (more ...)
Imaging cost implications of screening strategies for pheochromocytoma
In the decision analysis, biochemical testing by measurement of fractionated plasma metanephrines (traditional versus age-adjusted interpretation) was followed by CT imaging for all positive biochemical tests and if CT imaging was negative, then MIBG (I-123 or I-131) would be performed (Figure ). For the purpose of the economic evaluation, in all three screening strategies, a 0.5% prevalence of pheochromocytoma was assumed in a target hypertensive population, so 500 patients with pheochromocytoma would be expected in a sample of 100,000 hypertensive subjects (Figure ). Mayo Clinic Rochester charges for diagnostic studies were used: CT scan of the abdomen (with and without contrast) $1460, I-123 MIBG scan (with and without spect) $1875.
Decision analysis: Testing algorithm for pheochromocytoma in 100,000 hypothetical hypertensive subjects (including 500 individuals with pheochromocytoma)
If 100,000 subjects with hypertension would be screened using algorithm "A" (beginning with biochemical testing by measurement of fractionated plasma metanephrines, traditional interpretation), 499/500 patients with pheochromocytoma (overall sensitivity 99.8%) would be expected to be detected (1 patient expected to have false negative CT and MIBG imaging); furthermore 94,510/99,500 of subjects without pheochromocytoma would be reassured with a negative diagnosis (overall specificity 95.0%). In algorithm "A", 16,718 individuals would undergo CT scanning and 11,363 individuals would undergo I-123 or I-131 MIBG imaging. The total cost of imaging for algorithm "A" would be estimated to be 45.7 million dollars.
If algorithm "B" (biochemical testing using the age-adjusted metanephrine score) would be used in 100,000 subjects with hypertension, 499/500 patients with pheochromocytoma (99.8%) would be expected to be detected and 98592/99,500 individuals without pheochromocytoma would be reassured with a negative test result (overall specificity of 99.1%). In algorithm "B", 3,485 individuals would undergo CT scanning and 2,100 individuals would undergo I-123 or I-131 MIBG imaging. The cost of imaging for algorithm "B" would be approximately 9.0 million dollars. Thus, use of the age-adjusted plasma metanephrine score for biochemical testing for sporadic pheochromocytoma in a hypothetical population of 100,000 tertiary care hypertensive patients could result in a cost savings of 36.7 million dollars with equal detection of pheochromocytoma cases, relative to using the same biochemical testing but interpreting fractionated plasma metanephrine measurements in a traditional fashion.