Biochemical testing for neuroblastoma using plasma free 3‐O‐methyldopa, 3‐methoxytyramine, and normetanephrine

Neuroblastoma, the most common extracranial solid tumor of childhood, produces catecholamines that are metabolized within tumor cells. Homovanillic acid (HVA) and vanillylmandelic acid (VMA), the end products of catecholamine metabolism, have limited accuracy for testing of the tumors. This study assessed whether metabolites produced in earlier steps of catecholamine metabolism might offer improved diagnostic accuracy over urinary HVA and VMA.


INTRODUCTION
Neuroblastoma is an embryonal tumor of the autonomic nervous system that occurs near exclusively in children and is derived from developing and incompletely committed precursor cells of the neural crest. 1 Representing the most common extracranial solid tumor of childhood, responsible for 8.5% of all pediatric malignancies, neuroblastomas exhibit highly heterogeneous behavior. 2,3 Many run an aggressive course whereas others spontaneously and completely regress or mature into ganglioneuroblastomas or benign ganglioneuromas. Similar to pheochromocytomas, neuroblastomas produce catecholamines but have a limited capacity for catecholamine storage and secretion so that the catecholamines produced are largely metabolized within tumor cells. 4 Consequently, biochemical testing for neuroblastoma has relied on measurements of homovanillic acid (HVA) and vanillylmandelic acid (VMA). As the final end products of catecholamine metabolism, these metabolites are excreted in urine at micromolar concentrations, ensuring their easy measurement and continued use for biochemical testing. 5,6 Although HVA and VMA are produced in high abundance, this does not necessarily translate to high diagnostic accuracy. Diagnostic sensitivities of urinary HVA and VMA have been described at 73% to 92% at specificities of 96% to 100%. [6][7][8][9][10] Often such studies have not outlined adequate methods for exclusion of disease by alternatives to biochemical testing so that false-negative results may be missed and diagnostic sensitivity overestimated. The largest and most reliable analysis was provided by the prospective study of Schilling et al, 9 which included follow-up for exclusion or additional confirmation of disease. That study, involving 1.5 million infants screened for neuroblastoma at one year of age, indicated a diagnostic sensitivity of only 73%. Among the 27% of tumors that were initially missed many were aggressive, whereas among those detected many spontaneously regressed or ran a benign course so that screening did not improve outcomes and is no longer recommended.
The study of Schilling et al 9 was, however, a screening study and as shown for other catecholamine-producing tumors, 11 higher diagnostic sensitivities can be expected for patients in whom testing for disease is based on clinical presentation rather than screening. Clearly, however, more accurate biochemical tests for neuroblastoma would be useful, including tests that distinguish patients with aggressive versus more benign disease. To those ends, several studies have examined urinary normetanephrine and 3-methoxytyramine, the respective metabolites of norepinephrine and dopamine, as potential biomarkers for neuroblastoma. [12][13][14][15][16][17] Although one of those studies indicated utility of urinary 3-methoxytyramine for prognosis, 17 the overall diagnostic performance remained suboptimal. These findings are explained by indications that urinary 3-methoxytyramine, similar to urinary dopamine, is largely derived from renal uptake and metabolism of the amino acid precursor, 3,4-dihydroxyphenylanaline (L-dopa), 18,19 which in other studies has been associated with an unfavorable prognosis. 20 Two other studies examined use of plasma free 3-methoxytyramine and normetanephrine as tests for neuroblastoma, 21  Concentrations of plasma free 3-O-methyldopa, 3-methoxytyramine, normetanephrine, and metanephrine were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) as previously described. 24 Urinary VMA and HVA were measured at independent laboratories using LC-MS/MS, gas chromatography with mass spectrometry, or high-pressure liquid chromatography with electrochemical detection (see Supporting Information Table S2 for details) and normalized to creatinine, measured by photometry after applying the Jaffe reaction.

Data analysis
True-positive results for free 3-O-methlydopa, 3-methoxytyramine, normetanephrine, and metanephrine in the plasma panel for patients with neuroblastoma, or false-positive test results in patients without neuroblastoma were defined as values for any one of the metabolites above age-specific upper limits of reference intervals (Supporting Information Methods S1). 24

Clinical presentation of patients with neuroblastoma
Clinical data related to MYCN     As reflected by AUCs, measurements of plasma free 3methoxytyramine and normetanephrine showed superior performance for diagnosis of neuroblastoma than measurements of urinary HVA and VMA ( Figure 3 and Table 2). Higher AUCs for plasma compared with urinary metabolites were found irrespective of whether ROC curves were constructed using absolute values     Figure S1).

Diagnostic test performance
Higher AUCs for plasma than urinary metabolites were similarly found after data were restricted to complete sets of paired results for both plasma and urinary tests according to either absolute values (0.994 vs 0.946, P = 0.0105) or normalized to reference intervals (0.996 vs 0.880, P = 0.0003).
In addition to the above logistic regression analysis, discriminant analysis using random permutations of training and validation data sets confirmed the higher performance of the plasma than the urinary test (Supporting Information Analysis S1).

Correlations with clinical features
Among the 73 patients with available information on MYCN amplification, ratios of plasma 3-methoxytyramine to normetanephrine and urinary HVA to VMA were, respectively, 7.

DISCUSSION
This study establishes that measurements of plasma free 3methoxytyramine and normetanephrine provide a highly accurate diagnostic test for neuroblastoma. Preliminary evidence is also provided that addition of 3-O-methydopa to the panel might be useful for detecting occasional tumors characterized by limited production of downstream catecholamine metabolites, as previously indicated for L-dopa. 27,28 The panel also appears to offer opportunities for establishing prognostic risk.
Reasons for the improved diagnostic accuracy of plasma free Reflecting their suboptimal utility as biomarkers, measurements of urinary VMA and HVA have now been omitted from international neuroblastoma response criteria. 34 Furthermore, because use of the metabolites in screening is no longer recommended, diagnosis has come to depend on histopathology and imaging. 35 This in part reflects a common mode of discovery where a mass may be indicated by palpation or ultrasound; in such cases, a biopsy and radiographic and then nuclear medicine-based imaging represent the immediate options for diagnosis and staging. 36 Although differences in catecholamine metabolite profiles have been linked previously to high-risk versus low-risk disease and associated clinical outcomes, [15][16][17]38 the evidence of any relationships to MYCN amplification has until now been obscure. 39,40 Because catecholamine synthesis involves sequential conversion of L-dopa to dopamine and of dopamine to norepinephrine, the present findings of higher ratios of 3-methoxytyramine to normetanephrine or of HVA to VMA in neuroblastomas with than without MYCN amplification are in accordance with concepts that more aggressive disease is associated with more poorly differentiated catecholamine biosynthetic pathways. This is also in agreement with suggestions that measurements of L-dopa can provide additional utility for identifying high-risk cases of neuroblastoma in whom the usually measured catecholamine metabolites are not increased. 20 This in turn provides a rationale for inclusion of 3-O-methyldopa in the plasma panel.
Other efforts directed to the development of circulating molecular biomarkers for diagnosis, disease stratification, and management of patients with neuroblastoma, [41][42][43][44] while promising, will take time for validation and translation to the clinic. In the meantime, measurements of plasma-free O-methylated metabolites are already available at many centers as recommended tests for diagnosis of pheochromocytoma and paraganglioma. 45 Extending this application to neuroblastoma is therefore relatively straightforward. Nevertheless, even for pheochromocytoma and paraganglioma, there has been more than a 25-year delay from the initial introduction of plasma-free metanephrines 46 to final prospective validation for diagnosis. 11 The present report, along with others, 21,22,24 therefore only represents first steps in the diagnostic test development pipeline.
There are several limitations of the present study that must be addressed to bring the plasma test forward. Dependence on retrospectively collected patient samples mandates further prospective confirmatory studies. Another weakness involves the incomplete clinical and biochemical data that limited test interpretation. Thus, before the plasma test can be fully implemented into routine practice, there must be justification for inclusion of 3-O-methyldopa in the panel as well as validation of potential for disease monitoring and prognosis. It is, however, first crucial to prospectively establish utility for diagnosis.
For that, there must be inclusion of an appropriate comparison group of children in whom neuroblastoma is suspected and then appropriately excluded. In addition to urinary VMA and HVA measured by a single LC-MS/MS method, comparisons need to also include measurements of urinary free 3-methoxytyramine and normetanephrine. The considerable age-related falls in 3-O-methyldopa, 3-methoxytyramine, and normetanephrine mandate the use of age-specific reference intervals. 21,22 Considerations of preanalytical precautions and ageand sex-specific reference intervals should be considered similarly for comparisons of both plasma and urinary tests.
In summary, measurements of plasma free 3-methoxytyramine and normetanephrine provide a highly accurate test for neuroblastoma, with preliminary evidence indicating superior diagnostic performance compared with commonly used measurements of urinary VMA and HVA. With the availability of age-specific continuous reference intervals, a prerequisite for test interpretation, it is time to start the process of bringing laboratory testing for neuroblastoma in pediatric populations up to the same level already established for the diagnosis of chromaffin cell tumors in adults.

ACKNOWLEDGMENTS
Thanks are extended to Denise Kaden for technical assistance. Funding support from the Roland Ernst Stiftung für Gesundheitswesen (03/13, Germany) and from the Deutsche Forschungsgemeinschaft (CRC/Transregio 205/1) to MP, GE and AH is gratefully acknowledged.