https://orcid.org/0000-0002-8129-9360
Abstract
False-positive results are common for pheochromocytoma/paraganglioma (PPGL) real-world screening.
Determine the correlation between screening urine and seated plasma metanephrines in outpatients where PPGL was absent, compared to meticulously prepared and supine-collected plasma metanephrines with age-adjusted references.
Retrospective cohort study.
Databases from a single-provider provincial laboratory (2012-2018), a validated PPGL registry, and a manual chart review from a specialized endocrine testing unit.
PPGL registry data excluded known PPGL cases from the laboratory database. Outpatients having both urine and plasma metanephrines <90 days apart.
The correlation between urine and seated plasma measures along with the total positivity rate. All cases of plasma metanephrines drawn in the endocrine unit were reviewed for test indication and test positivity rate.
There were 810 non-PPGL pairs of urine and plasma metanephrines in the laboratory database; 46.1% of urine metanephrines were reported high. Of seated outpatient plasma metanephrines drawn a median of 5.9 days later, 19.2% were also high (r = 0.33 and 0.50 for normetanephrine and metanephrine, respectively). In contrast, the meticulously prepared and supine collected patients (n = 139, 51% prior high urine metanephrines) had <3% rate of abnormal high results in patients without known PPGL/adrenal mass.
There was a poor-to-moderate correlation between urine and seated plasma metanephrines. Up to 20% of those with high urine measures also had high seated plasma metanephrines in the absence of PPGL. Properly prepared and collected supine plasma metanephrines had a false-positive rate of <3% in the absence of known PPGL/adrenal mass.
Pheochromocytoma and paraganglioma (PPGL) are endocrine tumors that often secrete catecholamines. This may produce a variety of clinical presentations that may or may not contain the classic triad of paroxysmal hypertension, headache, and palpitations (1,2). We have previously shown that screening for suspected PPGL is widespread despite the rarity of its actual incidence (3). The issue of frequent screening for a rare disease in a population of patients with wide-ranging indications for testing highlights the problem of low positive predictive values, even for tests with high sensitivity and specificity (4).
International guidelines recognize the tension between sensitivity and specificity of a first-line test for all of the hyperfunctional adrenal hypertension syndromes including primary aldosteronism, Cushing syndrome, and PPGL; for each case, it has been recommended that an initial abnormal screening test be followed up (where clinical uncertainty remains) with a so-called confirmatory test (5-7). This paradigm is intended to employ a second test with high specificity, and in the case of possible PPGL, this may include clonidine suppression (8) or positron emission tomography-computed tomography/metaiodobenzylguanidine imaging (9,10) if an indeterminate candidate lesion is already known.
However, we and others (3,11) have previously highlighted the hazards of starting with a PPGL screening test that has a high false-positive rate such as the 24-h urine metanephrines or seated plasma metanephrines (12). Without due attention to the consequences of widespread use of a test with poor specificity, there may be a need for extensive resource use in follow-up confirmatory testing. At our institution, we have employed a carefully controlled, supine, resting, and age-adjusted plasma metanephrines test as a follow-up step to the investigation of uncertain clinical or lab findings in patients suspected of PPGL. In the present study, we sought to examine the agreement between first-line urine metanephrines screening tests, validated against our published PPGL provincial registry data, and a properly collected and reported plasma metanephrines.
Methods
This study was approved by the institutional Conjoint Health Research Ethics Board under the protocols REB18-1124- and REB19-0787. Measurements of urine metanephrines and plasma metanephrines are provided by the Calgary site of Alberta Precision Laboratories, which performs these tests for the whole of southern Alberta, Canada. The specific laboratory procedures have been described elsewhere (3); liquid chromatography-mass spectroscopy is used in accordance with standard procedures for an Agilent 1100 series liquid chromatograph coupled to an Agilent 6410 triple quadrupole mass spectrometer in electrospray ionization mode. The plasma specimens initially undergo clean-up using Oasis HLB solid phase extraction cartridges. Analyte separation is conducted via a Kintex 2.6 U HILIC 100A column (Phenomenex; Torrence, CA, USA) with a 10-min run time. Calibrators and internal standard materials are prepared using commercial standards from Cerillant (Round Rock, TX, USA) with quality control material sourced from Recipe (Munich, Germany). The majority of the routine plasma metanephrines were measured in the seated position by virtue of collection in multiple different outpatient patient phlebotomy centers. For seated plasma metanephrines, a published reference range (13) of ≤0.49 pmol/L (metanephrines) and ≤0.89 pmol/L (normetanephrine) was applied. The coefficient of variance percentage of the plasma normetanephrine is 5% at a level of 0.6 nmol/L and 1.5% at 2.0 nmol/L. The reference range for urine determinations varied according to patient age and has been previously reported (12).
A database of all patients having had both urine and plasma metanephrines tests between April 1, 2012 and June 30, 2018 was constructed; patients with any history of known PPGL were excluded using the provincial PPGL registry data, which has been shown to capture nearly 100% of all known and expected PPGL cases in our province from April 2012 to March 2019 (14). Cases with renal impairment (estimated glomerular filtration rate < 50 mL/min/m2) or inpatient collection of the urine metanephrines were excluded, as were any cases where the urine and plasma metanephrines measurements were separated by more than 90 days. For patients who had duplicate measures of either urine or plasma metanephrines, each urine test was matched 1:1 with the nearest-time plasma measure, and all remaining un-matched urine tests excluded. The matched pairs of urine and plasma metanephrines were examined for their degree of overall correlation as well as according to diagnostic categorization (normal or high) with respect to their reference ranges to determine the clinical agreement between the 2 tests, either of which could be considered a first-line outpatient screening test.
A second independent database was then constructed from an audit of the supine plasma metanephrines drawn in the division of endocrinology’s Endocrine Testing Unit (ETU) spanning 12 months from January 1, 2019 to December 31, 2019. All patient charts were reviewed to determine the reason for referral and any other clinical features reported by the ordering endocrinologist. In the ETU, patients are contacted by the nurse clinician in advance to ensure that the patients arrived fasting and having refrained from any caffeine or alcohol for 24 h. Patients are advised to avoid taking any health supplements, vitamins, or any nasal decongestants for 24 h. In addition, the patients are asked to refrain from any exercise or smoking for 4 h prior to the test. The nurses review the medication profile and any potentially interfering medications (eg, tricyclic antidepressants, catecholaminergics, or stimulants) are brought to the attention of the ordering physician for directions on discontinuation. Upon arrival in the ETU, patients have an antecubital IV access line established after which they lie supine in a quiet room for 30 min. The nurse then draws the blood sample, and patients are discharged home. Results are interpreted using the supine, age-adjustment of Eisenhofer (15) after conducting a validation of our methods and results with that of the Dresden laboratory as part of an internal quality assurance initiative.
The population rates of abnormal results according to either urine or seated metanephrines were compared to the rates of abnormal results seen within the population who underwent the meticulous preanalytical process and interpretation of measures from the ETU to determine the extent to which each of the various approaches might result in patients needing further confirmatory or other PPGL-directed testing. There was no direct linkage between individual patients seen in the ETU vs those in the administrative data populations, and the database inclusion dates were not identically matched, hence the simple population rate comparisons rather than individual case reviews.
Demographics were presented using standard descriptive measures, and laboratory results examined for normality using the Shapiro-Wilkes’s test. Correlations were performed using Spearman’s r test for nonparametrically distributed data. All statistics were performed using GraphPad Prism 6.0 (La Jolla, CA, USA), and a P-value of 0.05 taken as the level of significance.
Results
Outpatient Paired Urine and Plasma Metanephrines From the General Population
The laboratory urine metanephrines database comprised 11 697 tests in 10 071 unique patients. The stepwise application of the exclusion criteria to reach the final cohort for analysis is shown in Figure 1; this resulted in 810 sets of matched, outpatient urine and plasma metanephrines from 770 unique patients where both measures occurred within 90 days of each other and excluding all samples from known PPGL patients. None of the age or biochemical results were normally distributed. The median [interquartile range (IQR)] age of the patients was 52 (39-62) years, and 56.5% were women. The median (IQR) urine normetanephrine was 2.2 (1.5-3.1) umol/d, and urine metanephrine was 0.6 (0.4-0.8) umol/d. Of these, 40.0% were reported as having supranormal urine normetanephrine, 9.8% had supranormal urine metanephrine, and 3.7% had elevations in both. A total of 46.1% had at least 1 high result. The median (IQR) interval between the urine metanephrines and the seated plasma metanephrines was 5.9 (1.1-29.0) days. The median (IQR) seated plasma normetanephrine was 0.51 (0.37-0.73) umol/L and plasma metanephrine was 0.19 (0.19-0.23) umol/L.
Laboratory database construction for paired urine and plasma metanephrines, general population outpatient collections. Abbreviations: eGFR, estimated glomerular filtration rate; PNM, plasma normetanephrine; PPGL, pheochromocytoma/paraganglioma; UNM, urine normetanephrines.
In Figure 2, the correlations between urine and plasma measures are displayed. From this figure it may be seen that the correlations are poor to moderate with a Spearman r = 0.33 (0.27-0.39) for normetanephrine and Spearman r = 0.50 (0.45-0.55) for metanephrine (both P < 0.0001). Among those patients with a high urine normetanephrine, 262/324 (80.8%) had a normal plasma normetanephrine result, and among those with a high urine metanephrine, 72/79 (91.1%) had a normal plasma metanephrine, usually less than 1 to 4 weeks apart. Therefore, after having both urine and seated plasma measurements, between 8.9% and 19.2% of patients would still potentially require further investigation due to apparently concordant high readings in both measures.
Correlation of paired plasma and urine measurements in patients undergoing PPGL screening but without known disease. (A) Normetanephrine and (B) metanephrine. Note that all data in red represents patients in whom the urine age-adjusted (normetanephrine) or sex-adjusted (metanephrine) results were high. When plasma measures were drawn in the seated position, a single upper reference limit was applied. Abbreviations: PNM, plasma normetanephrine; PPGL, pheochromocytoma/paraganglioma; UNM, urine normetanephrine.
Outpatient Supine Plasma Metanephrines From the Endocrine Testing Unit
Table 1 demonstrates the results of the ETU audit of 139 unique patients having supine, resting plasma metanephrines measured. From this group, 54% were women and the mean age was 49.6 years (range 15-89). There were 61% of patients with a history of hypertension, and 63% were using at least one anti-hypertensive medication (mean of 1.9 anti-hypertensives per treated patient). Regarding psycho-active medications, 15% of patients were using such medications, and half of these cases included tricyclic antidepressants or serotonin-norepinephrine reuptake inhibitors that the ordering physician felt could not be stopped. The 3 most common reasons for referral for ETU testing were (1) high urine metanephrines (51.1%), (2) adrenal mass (18.0%), and (3) known PPGL patient for testing/follow-up (17.2%). In total, 73% of patients had at least 1 of PPGL symptoms, adrenal mass, or hypertension cited in the referral. Among those for whom the urine metanephrines result was attached (n = 49), the mean urine normetanephrine was 3.73 nmol/d (range 2.3-7.1 nmol/d), and mean urine metanephrine was 1.0 nmol/d (range 0.2-3.3 nmol/d). The distribution of properly collected, supine plasma normetanephrine according to primary presenting feature is shown in Figure 3A. After excluding known PPGL cases identified by the referring physician (n = 24), only 11/115 patients (9.5%) had a high, age-adjusted plasma normetanephrine and/or metanephrines, and only 10/115 (8.6%) after considering the assay coefficient of variance percentage overlap with normal. One was using amitriptyline and therefore a false elevation so that out of the 115 patients undergoing the test who were not already known to have PPGL, only 9 (7.8%) would have abnormal results requiring further testing.
Distribution and prevalence of specific clinical data included on the referral form for supine collected plasma metanephrines in the Endocrine Testing Unit
| Primary stated indication for testing, n (%) | |||||||
|---|---|---|---|---|---|---|---|
| Symptoms compatible with PPGL n = 5 | Adrenal mass n = 25 | Prior high PNM n = 12 | High UNM n = 71 | Known PPGL n = 24 | Family PPGL screen n = 2 | HTN n = 4 | |
| Symptoms | 3 (12) | 4(33) | 14 (20) | ||||
| Adrenal mass | 3 (25) | 16 (23) | |||||
| Known PPGL | 3 (12) | 3 (4) | |||||
| HTN | 1 (8) | 19 (27) | |||||
| Genetics | 2 (16) | 1 (1) | 1 (4) | ||||
| Other adrenal HTN | 3 (12) | ||||||
| No other information provided | 5 (100) | 16 (64) | 2 (16) | 19 (27) | 23 (96) | 2 (100) | 4 (100) |
| Primary stated indication for testing, n (%) | |||||||
|---|---|---|---|---|---|---|---|
| Symptoms compatible with PPGL n = 5 | Adrenal mass n = 25 | Prior high PNM n = 12 | High UNM n = 71 | Known PPGL n = 24 | Family PPGL screen n = 2 | HTN n = 4 | |
| Symptoms | 3 (12) | 4(33) | 14 (20) | ||||
| Adrenal mass | 3 (25) | 16 (23) | |||||
| Known PPGL | 3 (12) | 3 (4) | |||||
| HTN | 1 (8) | 19 (27) | |||||
| Genetics | 2 (16) | 1 (1) | 1 (4) | ||||
| Other adrenal HTN | 3 (12) | ||||||
| No other information provided | 5 (100) | 16 (64) | 2 (16) | 19 (27) | 23 (96) | 2 (100) | 4 (100) |
Abbreviations: HTN, hypertension; PPGL, pheochromocytoma/paraganglioma; PNM, seated plasma normetanephrine; UNM, urine normetanephrine.
Distribution and prevalence of specific clinical data included on the referral form for supine collected plasma metanephrines in the Endocrine Testing Unit
| Primary stated indication for testing, n (%) | |||||||
|---|---|---|---|---|---|---|---|
| Symptoms compatible with PPGL n = 5 | Adrenal mass n = 25 | Prior high PNM n = 12 | High UNM n = 71 | Known PPGL n = 24 | Family PPGL screen n = 2 | HTN n = 4 | |
| Symptoms | 3 (12) | 4(33) | 14 (20) | ||||
| Adrenal mass | 3 (25) | 16 (23) | |||||
| Known PPGL | 3 (12) | 3 (4) | |||||
| HTN | 1 (8) | 19 (27) | |||||
| Genetics | 2 (16) | 1 (1) | 1 (4) | ||||
| Other adrenal HTN | 3 (12) | ||||||
| No other information provided | 5 (100) | 16 (64) | 2 (16) | 19 (27) | 23 (96) | 2 (100) | 4 (100) |
| Primary stated indication for testing, n (%) | |||||||
|---|---|---|---|---|---|---|---|
| Symptoms compatible with PPGL n = 5 | Adrenal mass n = 25 | Prior high PNM n = 12 | High UNM n = 71 | Known PPGL n = 24 | Family PPGL screen n = 2 | HTN n = 4 | |
| Symptoms | 3 (12) | 4(33) | 14 (20) | ||||
| Adrenal mass | 3 (25) | 16 (23) | |||||
| Known PPGL | 3 (12) | 3 (4) | |||||
| HTN | 1 (8) | 19 (27) | |||||
| Genetics | 2 (16) | 1 (1) | 1 (4) | ||||
| Other adrenal HTN | 3 (12) | ||||||
| No other information provided | 5 (100) | 16 (64) | 2 (16) | 19 (27) | 23 (96) | 2 (100) | 4 (100) |
Abbreviations: HTN, hypertension; PPGL, pheochromocytoma/paraganglioma; PNM, seated plasma normetanephrine; UNM, urine normetanephrine.
Endocrine Testing Unit data. (A) distribution of properly collected, supine, resting plasma normetanephrine with age-adjusted reference ranges and (B) correlation between outpatient urine normetanephrine values (all high for age-category) with subsequent supine resting plasma normetanephrine using age-adjusted reference ranges. Note that “Known PPGL” includes postoperative measurements.
Endocrine Testing Unit Patient Subgroups
Among the ETU patients with reportedly high urine metanephrines but no known PPGL (n = 71), 20% described typical PPGL symptoms, 23% also had a known adrenal mass, 27% had hypertension, and 27% had no reason stated. Among patients with high urine metanephrines but no known adrenal mass (n = 38), only 1 (2.5%) was found to have elevated supine plasma metanephrines under strictly controlled collection conditions. Using cases with specific urine normetanephrine results available, the correlation between prior outpatient urine normetanephrine and subsequent supine resting plasma normetanephrine is shown in Figure 3B; according to age based reference ranges, all 49 of the urine normetanephrines were high for age, but only 4 (8.1%) of the supine plasma normetanephrines were also elevated, potentially requiring further investigation.
Interpretation
Screening for PPGL is a common practice in our health system and when that screening approach employs a 24-h urine metanephrines or seated plasma metanephrines, there is a high rate of abnormal results (15). Physicians may be aware of the potential for false-positive results from the urine measurement and may already be using a seated plasma measurement as a second step; this may explain why nearly half of the plasma measures were ordered in the context of an already high urine test result. In patients without PPGL yet with frequently high urine metanephrines, there was a poor correlation between urine and plasma metanephrines in unique patients suggesting that neither test was accurate for diagnosis when used in a population with variable indications for testing. In addition, between 12% and 20% of individuals with high urine metanephrines also had high seated plasma metanephrines, indicating a substantial proportion who might go on to more sophisticated and costly testing.
However, both urine and seated tests seemed to perform particularly poorly at the population level when compared to the strictly controlled supine measures of the ETU, despite the fact that the ETU population was highly selected by virtue of having many clinical features of PPGL in addition to already having either high urine metanephrines and/or adrenal masses and coming in referral from endocrinologists alone. In contrast to the laboratory results from the general population, the ETU data suggest that application of strict attention to preanalytical details can reduce the numbers of patients who do not have a known PPGL but are in need of additional testing to less than 8%, despite using a test that is thought to approach 100% sensitivity for true PPGL (7).
Tests with poor positive predictive value have the potential to cause significant confusion for physicians (16) and may consume resources from health systems (17) in addition to burdening patients with health concerns while awaiting clarification (18). Other investigators have highlighted the marked increase in false-positive rates when using seated plasma metanephrines (7,19), and our data confirm this. However, even supine collected plasma metanephrines have been reported to have a clinically significant rate of false-positive results (20) in contrast to our experience (where false-positive results were extremely low), suggesting these differences may be related to other preanalytical factors, such as whether the blood is drawn from an immediate needle stick vs previously placed indwelling catheter (21). Attention to patient preparation such as refraining from smoking, exercise, and any potentially interfering drugs where possible may be important, along with placement of the intravenous access 30 min prior to the actual blood draw. Based on these results, it would seem that this kind of carefully controlled plasma metanephrines testing, interpreted with supine age-adjusted limits could markedly reduce both patient anxiety and the need for expensive additional testing in cases where the primary concern is the biochemical result as opposed to compelling objective clinical data such as known genetic PPGL mutations or high risk target lesions (7).
There are several limitations to our study. Plasma metanephrines are a free and direct measurement whereas urinary metanephrines represent total measurement following acid hydrolysis, which means that they are sulfate-conjugated products. The slight differences in analytes may thus partially explain some of the poor correlation observed. We largely assumed that normal plasma metanephrines effectively rule out PPGL, which may have missed nonfunctional PPGL (such as head and neck) and occasional functional lesions with mild biochemical abnormalities. However, these latter conditions are rare, and we likely were able to account for most of them using our comprehensive, validated PPGL registry (14). It was not possible to determine the collection position of the plasma metanephrines from the large laboratory data; sheer numbers alone would suggest that most of these patients did not come through the ETU, and even if some of them did, giving slight overlap, that would only result in an underestimation of the already high (20%) positivity rate reported for seated plasma metanephrines. In the general population sample, it was not possible to determine the order of testing and whether the urine test was always first; the use of patients who had more than one test of any kind (plasma and urine) necessarily introduces a bias in the correlations. The correlation between plasma and urine metanephrines may be better than that observed, if dual measures had been done in all patients. The ETU patients represent a highly selected population compared to those from the general population such that test positivity rates should not be directly compared. Nonetheless, by virtue of the ETU patients coming from endocrinologists, a higher test-abnormality prevalence might be expected, and the fact that the observed prevalence of abnormal PNM was much less than in the general population does still suggest overall fewer false positives from the ETU collections. The ETU chart audit was prone to reporting bias in that there was no standardized clinical information set to draw upon; the information available was at the ordering endocrinologist’s discretion and could not be linked to the larger administrative database. However, if anything, this might have resulted in an underappreciation of the factors making up the pretest probability, and we have already shown this cohort to have a very high prevalence of multiple clinical factors, proving this to be a group in whom PPGL could be a very legitimate consideration. It is important to emphasize that the present analysis was restricted to patients in whom the ordering physician had reason to want more than 1 test; a selection bias is therefore present, and so the results cannot be used to estimate the overall test sensitivity and specificities for either urine or plasma metanephrines. Rather, the main point is that properly collected and reported supine plasma metanephrines have an excellent true-negative rate among those for whom suspected false positives is the working concern.
In conclusion, in general practice, 24-h urine metanephrines and seated plasma metanephrines are likely to be poor screening tests for PPGL owing to a very high rate of false positives. However, primary screening or next-step testing with strictly controlled preanalytical conditions and supine, age-adjusted plasma measurements will show that even in a high-suspicion population, less than 8% of patients will have results that still require further testing, thus obviating the need for most PPGL confirmatory tests. In the absence of known adrenal mass, <3% of mild to moderately elevated urine metanephrines would require any additional testing after careful supine plasma metanephrines testing. Although properly prepared and supine-collected, age-adjusted plasma metanephrines may be more time- and resource-consuming, given the much lower false-positive rate, this should become the preferred confirmatory step for any patient with suspected false-positive urine or seated plasma metanephrines.
Acknowledgments
The authors gratefully acknowledge the contributions of Dr. Andrew Tang who assisted with study concept creation and planning. Dr. Tang tragically died in 2019.
Financial Support: This study was funded by the Canadian Institutes of Health Research Project Grant (159533). The funders had no role in study design, data collection, analysis, reporting, or the decision to submit for publication.
Author contributions: GK conceived the idea, conducted the analyses, and wrote the first draft. JB and HS assembled the laboratory database, performed the laboratory tests, and assisted with the first draft. AH and JP assisted with the assembly of the PPGL registry. AL created the PPGL registry and assisted with the primary analysis. All authors critically reviewed the manuscript for publication.
Additional Information
Disclosures: The authors have nothing to disclose.
Data Availability
Restrictions apply to the availability of some or all data generated or analyzed during this study to preserve patient confidentiality. The corresponding author will on request detail the restrictions and any conditions under which access to some data may be provided.
References
