Age-related medical decision limits for urinary free (unconjugated) metadrenalines,catecholamines and metabolites in random urine specimens from children

Introduction

Solid tumours presenting in childhood are rare when compared with findings in adults. Despite overall cure rates approaching 80%, cancer is the commonest cause of non-accidental death in Scottish children. ¹ Neuroblastoma is the most common extracranial solid tumour in childhood, and the most frequently diagnosed in infancy. ² It accounts for about 8% of all childhood cancers, has one of the highest death rates, and is characterized biochemically by elevated urinary catecholamine metabolites, particularly vanillylmandelic acid (VMA) and homovanillic acid (HVA). ³ It is the commonest cause of death from malignancy in this population.

By contrast, phaeochromocytoma and paraganglioma rarely present in childhood. ⁴ They stain positive for chromaffin and also produce catecholamines. ⁵ In adults, work from the Crosshouse laboratory has demonstrated that urinary free metadrenalines (also known as free metanephrines) exhibited superior test sensitivity for the diagnosis of phaeochromocytoma or paraganglioma when compared with either urinary VMA, catecholamines or plasma catecholamines. ⁶ Moreover, in view of their better sensitivity, recent recommendations for the diagnosis of chromaffin tumours presenting in childhood advocate the use of urinary metadrenalines as first-line biochemical tests for diagnosis. ⁷ Similarly, in the investigation of possible neuroblastoma, it has been noted that the additional combination of urinary normetadrenaline (also known as normetanephrine) with either VMA or HVA significantly enhanced the diagnostic power of testing, and inclusion of 3-methoxytyramine (a metabolic intermediate of dopamine [DA] catabolism) in the test profile could increase diagnostic sensitivity up to 100%. ^8,9

However, for ethical reasons, it is difficult, in general, to establish reference ranges for tests in children. Specimens with which to construct reference range data are often obtained from children in hospital and are not always based on large numbers of subjects. ¹⁰ In addition, in health, the daily urinary excretion of, for example, VMA, increases linearly with age with no difference between boys and girls. ¹¹ Also, it can be difficult to obtain accurately timed urine specimens from infants and children. ^8,12 However, throughout childhood, the urine creatinine excretion rate also increases in a relatively linear fashion with age, and it has long been suggested that measurement of substances in casual urine specimens may be corrected to some extent by expressing results per unit creatinine. ¹³ Therefore, biochemical diagnosis of catecholamine-secreting tumours presenting in infancy and childhood requires provision of age-appropriate reference intervals. ^7,14

In paediatrics, the diagnostic process is complex due to both the biochemical heterogeneity of these tumours and the scarcity of appropriate reference values for catecholamines and their derivatives, especially in the young child. ¹⁵ No age-related reference ranges for free metadrenalines in random urine specimens currently exist in the paediatric literature. In order to better inform the diagnostic process, the purpose of the present laboratory-based study was to provide up-to-date, age-related medical decision limits for free catecholamines and their metabolites, including their methylated derivatives, in untimed urine specimens obtained from a relatively large number of subjects.

Subjects and methods

From July 1994 to February 2010, the Crosshouse laboratory received 1658 random urines obtained from infants, children and young adults, whose specimens were referred from hospital laboratories throughout Scotland, for analysis of catecholamines and metabolites. Urines were received in universal containers pretreated, if required, by the addition of a drop of hydrochloric acid. On receipt in the laboratory, urine specimens were checked for adequate acidity (pH 2–4) and then analysed for creatinine by an automated kinetic Jaffe method (Roche Diagnostics Ltd, West Sussex, UK). In order to detect a common source of analytical interference, from June 2003 each specimen was also analysed for paracetamol as described previously. ¹⁶ Noradrenaline (NA), adrenaline (AD), DA-free normetadrenaline (fNMA), free metadrenaline (fMA), VMA, 5-hydroxyindoleacetic acid (HIAA) and HVA were measured by high performance liquid chromatography with electrochemical detection (HPLC-ECD) as described in more detail elsewhere. ¹⁷ For a limited period, i.e. during March 2004 to April 2005, and again from April 2009 to February 2010, a lesser number of results for urinary free 3-methoxytyramine (f3MT) were available by HPLC-ECD. Results for each of these analytes were expressed as mmol per mol creatinine, and together with date of collection, date of birth and any existing request form information were stored in a password-protected, computerized spreadsheet form.

In each case, follow-up from the laboratory was attempted by various approaches including communications by letter or telephone with clinicians and referring laboratories throughout Scotland; examination of patient case-notes where possible; recording of findings in repeat specimens; and recording of information supplied on request forms, as well as liaison with histopathology departments where appropriate. These ‘indirect approach’ procedures permitted identification of specimens from those individuals with conditions deemed inappropriate for inclusion in the reference group, e.g. those harbouring, or undergoing treatment for neuroblastic, chromaffin or other tumours; ¹⁸ those receiving administered catecholamines, e.g. DA or l-dopa; ^18–20 or those positive for the presence of paracetamol and whose specimens would be subject to in vitro interference. ¹⁶ In addition, since highly variable results for catecholamine metabolites are observed in relatively dilute urine specimens, all specimens exhibiting urine creatinine <1.0 mmol/L were also excluded from the reference group. ²¹ The remaining patient results were then stratified into each of seven age ranges, namely under 1, 1 or 2, 3 or 4, 5–7, 8–10, 11–13 and 14–19 y.

Data analysis

Statistical analysis was performed on the remaining specimens using the MedCalc statistical software package (version 11.3.3., MedCalc Software, Mariakerke, Belgium). For each analyte, the results of frequency distributions were plotted for each age range and examined by visual inspection, thereby permitting identification of obvious outliers. ^22,23 Following exclusion of any specimens displaying at least one outlying result, the remaining distributions were each examined for normality by the Kolmogorov–Smirnov test. Since the majority of analytes do not exhibit normal Gaussian distributions, putative reference ranges were calculated using the non-parametric method. The conventional 95th percentile reference limits were determined by calculating the rank numbers for the 2.5th and 97.5th percentiles. ²³ For each analyte, the differences between results from boys and girls were determined by the non-parametric Mann–Whitney U test for unpaired observations. In addition, for each analyte, the results for those aged up to six months were compared by the Mann–Whitney U test, with those obtained from individuals aged older than six months but under one year.

Validation of medical decision limits was achieved by comparison of the observed ranges with our pre-existing limits for NA, AD, DA, VMA and HVA, which were based on the work of Dr Fitzgibbon and her colleagues ^14,22 using random urine specimens collected from 305 well children divided into the same age ranges. Furthermore, for all the analytes, comparisons were made with the results from: (a) 55 patients harbouring neuroblastic tumours (neuroblastoma, ganglioneuroblastoma, ganglioneuroma) on specimens obtained at time of diagnosis; and (b) with the results from 11 patients presenting in childhood with phaeochromocytoma or paraganglioma again on specimens obtained at time of diagnosis. Application of the determined medical decision limits allowed calculation of test sensitivities for each of the analytes in each of the two patient groups (a) and (b) above.

Results

Following truncation of the plots of results versus patient age grouping for each analyte by application of the exclusion criteria as described, of the 1658 random urine specimens received in the laboratory and obtained from infants, children and young adults for analysis of catecholamines and metabolites, 786 were excluded from the determination of decision limits. The exclusion criteria applied; the numbers of specimens excluded; and the reasons for rejection (literature references) are given in Table 1. Thus there remained the results for 872 specimens which formed the basis for determination of suggested medical decision limits. Following this process, it was noted that there was no significant difference between the results for boys or girls. The medians for each analyte for both boys and girls, together with the probability that results are derived from the same population, as determined by the Mann–Whitney U test for unpaired observations, are given in Table 2.

For VMA, HVA, NA, AD and DA, the medians, the number of specimens, the observed range of values and our currently employed upper reference limits, ^14,22 at each of the assigned age ranges, are given in Table 3. For HIAA, fNMA, fMA and f3MT, the medians, observed ranges after truncation and suggested medical decision limits are shown in Table 4. In virtually every case there was an age-related downward trend in median values. However, both AD and its immediate metabolite, metadrenaline, exhibited a more marked degree of scattering of values within each age range such that arbitrary upper limits were selected. For AD, this limit was 0.08 mmol/mol creatinine as recommended previously by Fitzgibbon et al., ²² and for metadrenaline, a figure of 0.12 mmol/mol creatinine is similarly suggested.

Comparison of the results obtained from those aged up to six months, with those older than six months but less than one year, by the Mann–Whitney U test for unpaired observations, showed no significant difference for VMA, HIAA, HVA, NA, AD, DA or f3MT between the two age subgroups. However, a significant difference was observed for fNMA. The medians (range) up to age six months were 0.15 (0.001–0.529) mmol/mol creatinine, and 0.09 (0.01–0.238) mmol/mol creatinine for those older than six months but less than one year (P = 0.0003).

The individual results from specimens obtained at time of diagnosis for 55 patients each harbouring neuroblastic tumours, and for 11 patients with phaeochromocytoma or paraganglioma, are shown in Tables 5 and 6, respectively. Where more than one specimen was received at time of diagnosis, the mean result for that analyte is given. By adopting the existing upper reference limits for VMA, HVA, NA, AD and DA ¹⁴ (Table 3), and the suggested decision limits for HIAA, fNMA, fMA and f3MT (Table 4), it was possible to calculate the proportion of abnormally elevated results for each analyte in each of the two tumour groups. In the neuroblastic tumour group (Table 5), one or more of the analytes were elevated above the proposed limits in 54 of the 55 patients. In only one instance (case 16, Table 5, an 11-month infant at the time of first testing with a histologically poorly differentiated neuroblastoma) were all of the catecholamines and metabolites within the proposed limits. Relatively, on considering the whole of the neuroblastic tumour group, HVA was the most frequently elevated analyte (54 of 55) followed by f3MT (14 of 16), VMA (45 of 53) and DA (43 of 53). However, among those presenting at age one year or more, all of the available DA results (37 of 37) and all but one of the available f3MT results (13 of 14) were elevated. Moreover, of the eight patients who exhibited normal VMA results, seven showed elevation of both DA and HVA, and only two (2 of 7 available) had elevated fNMA. Nevertheless, the fNMA results were elevated in more than three quarters of all available specimens from patients harbouring neuroblastic tumours (36 of 47). Whereas, by contrast, the results for fMA (9 of 46), NA (7 of 50), AD (6 of 42) or HIAA (0 of 54) were much less frequently elevated, and hence these last four analytes would be of little value as tests in the diagnosis of neuroblastoma and related neuroblastic tumours.

In the second tumour group which comprised 11 cases of phaeochromocytoma or paraganglioma presenting in childhood (Table 6), the pattern of analyte elevation was quite different. At time of diagnosis, the most frequently elevated analyte was fNMA (11 of 11) followed by NA (10 of 11). VMA was elevated in approximately half the patients (6 of 11) while the remaining analytes were of little value in diagnosis. In addition, in this series, the age at presentation for phaeochromocytoma or paraganglioma ranged from 5 to 18 y, whereas for neuroblastic tumours, 69% (38 of 55) were under five years.

Discussion

The reasons for occasional outlying values in the reference group, which required truncation prior to assessment of medical decision limits, are difficult to determine from a laboratory-based study due largely to lack of adequate information. Dietary factors such as consumption of catecholamine-rich, fruit-containing foodstuffs, e.g. bananas or pineapple, may lead to substantial increases in the urinary excretion of catecholamines and their metabolites. ²⁶ Such interferences, which may lead to production of false-positive results, is a particular problem if biochemical measurements include the conjugated moieties (often termed total fractionated metanephrines). ^27,28 Measurement of only the urinary free species, as employed in the present study, tends to reduce, though not eliminate these unwanted dietary effects. ^27,28 Interestingly, DA is not the only precursor of HVA. For example, hydroxytyrosol, a component of olive oil, may be metabolized to HVA. ²⁹ In addition, Professor Crozier and his colleagues ³⁰ at the University of Glasgow have been able to demonstrate that the flavonoid, quercetin, a component of many common foodstuffs such as tomato juice or onions, is acted upon by colonic microflora, leading to production and urinary excretion of HVA. In adult subjects, dietary flavonol intake can significantly increase urinary HVA excretion and exceed the upper normal limit, leading to false-positive results. ³¹ Moreover, in a case report of a child undergoing investigation for the presence of a possible catecholamine-secreting lesion, serum and urine HVA concentrations of approximately 10 times upper normal were obtained while he was receiving quercetin-containing supplements, thereby falsely mimicking DA production by an occult tumour. Analysis of a second blood sample obtained 10 days later yielded a normal result and the child remained free of neoplasm for at least a year after his laboratory tests, indicating that the previously elevated result was likely to have been caused by a transient interferent. ³²

It is noteworthy that results for all of the analytes were not available for every specimen. These missing values were due to chromatographic interference, possibly by dietary or drug-related factors. Some analytes were more prone to interference than others. For example, in Table 5, which lists the available results obtained at time of diagnosis for 55 infants and children harbouring neuroblastic tumours, there were 13 and 9 missing values for urinary free AD and metadrenaline, respectively. Urinary fNMA and NA results were unavailable in eight and five cases, respectively, whereas results for VMA and DA each exhibited two missing values and HVA was available in every case. These figures illustrate the susceptibility of different analytes to interference in the analytical systems employed in this study. It is encouraging that those analytes manifesting the least susceptibility to interference (HVA, VMA, DA) also displayed the greatest usefulness in terms of clinical sensitivity in the identification of neuroblastoma and related tumours. Also, in Table 5, it is interesting that fNMA was positive in 76% of cases. However, given the limited clinical sensitivity of fNMA when compared with HVA, and the susceptibility of fNMA analysis to interference by largely unknown factors, it would be difficult to recommend fNMA as a first-line test in the detection of neuroblastoma. Although there was only a limited amount of data for urinary f3MT, this latter test exhibited a very high sensitivity for detection of neuroblastoma as suggested previously. ⁹

The real advantage of the fNMA test was in the detection of phaeochromocytoma or paraganglioma since this was positive in all 11 cases presenting in childhood (Table 6). Urinary NA showed marginally less sensitivity than fNMA in this small cohort. A further advantage of fNMA is its superior, in vitro stability during specimen storage compared with that of the parent catecholamines. ^33,34 The other free metadrenalines (fMA and f3MT) and VMA, HIAA and HVA were of little or no value in the identification of phaeochromocytoma or paraganglioma presenting in childhood (Table 6).

The differences in urinary biochemical excretion patterns between those harbouring either neuroblastic or chromaffin tumours have long been observed and are both striking and intriguing. For example, in neuroblastoma, the biochemical picture is dominated by elevated urinary excretion of DA and its metabolites, f3MT and HVA. However, despite very few instances showing elevated NA or AD (∼14%) in the present study, their metabolic products, fNMA and VMA, are very often increased to abnormal amounts. On the other hand, the pattern of urinary excretion in those with chromaffin tumours is dominated by NA and its metabolic products, fNMA and VMA. DA and its final metabolic product, HVA, were rarely, if ever, elevated, yet both tumour types are capable of producing and excreting the parent catecholamines. One explanation is provided by an electron microscopy study on tissue from both tumour types which showed many dense-cored granules contained in phaeochromocytomas, but few in neuroblastomas. ³⁵ The interpretation is that neuroblastomas lack a catecholamine storing mechanism so that catecholamine is released from the tumour cells and inactivated (metabolized) soon after its formation. ³⁵

The results in this study are expressed as ratios of analyte to creatinine concentrations and not authentic analyte excretion rates. However, it has been noted by the work of Professor Craft and his colleagues ^21,36 from the Newcastle School of Medicine, in a very large cohort of 10,000 babies aged 20–40 weeks, that, when results are expressed in this way, the upper limits of normal are almost entirely dependent upon the absolute sample creatinine concentration. In this age group, any effect due to age was relatively small by comparison. ³⁶ For example, at a sample creatinine concentration of 0.1 mmol/L, the upper limits of normal for HVA in boys, HVA in girls and VMA for both sexes were 104.1, 45.6 and 35.5 mmol/mol creatinine, respectively. These limits then decreased markedly with increasing sample creatinine concentration to produce much lower and more stable respective upper normal values of only 21, 22 and 12 mmol/mol creatinine at sample creatinine concentrations of around 1.0 mmol/L or greater. ²¹ The authors concluded that, whatever the explanation for this phenomenon, sample creatinine concentration must be taken into account when constructing apparent age-related reference ranges. ^21,36 It is for these reasons that only urine samples exhibiting a creatinine concentration of 1.0 mmol/L or greater were selected for inclusion in the present study (Table 1).

In this study, the results for fNMA, expressed in mmol/mol creatinine, appeared to be significantly higher at age up to six months (n = 64) than for those older than six months but under one year, although the latter age subgroup comprised significantly fewer subjects (n = 29). Therefore, on this basis, for fNMA, the suggested decision limit at age under one year could be further subdivided. However, application of this statistical manoeuvre using the present data would produce an apparently anomalous upper normal limit of only approximately 0.25 mmol/mol creatinine. Clearly, this age group (under one year) needs to be re-visited when more reference data for fNMA becomes available.

Note

There is, possibly, some confusion with regard to the terminology associated with catecholamines and their metabolites. The terms for the parent catecholamines used in the English scientific and medical literature tend to be based on latin roots, i.e. adrenaline and noradrenaline. In American-English, greek roots are more commonly used, i.e. epinephrine and norepinephrine. Similarly, their 3-methoxy metabolic intermediates tend to be called metadrenaline and normetadrenaline in the English scientific literature, whereas metanephrine and normetanephrine (collectively the metanephrines) predominate in American-English articles.

To add to the confusion, catecholamines, and their 3-methoxy intermediates, exist in biological fluids as both free (unconjugated) and conjugated species. The latter term describes a form of catecholamine or metabolite which has undergone (usually) a sulphate transfer reaction at the fourth carbon of the benzene ring of its catechol skeleton, in the cells of the gastrointestinal tract. For a more comprehensive description of these features, please see the editorial by Professor Eisenhofer. ³⁷

DECLARATIONS

Competing interests: None.

Funding: None.

Ethical approval: Not required.

Guarantor: DFD.

Contributorship: DFD conceived the study, researched literature and wrote the manuscript. PJH, DM and RC investigated and treated patients, discussed the study and reviewed and edited the manuscript.

Acknowledgements: The authors are grateful to clinical and laboratory colleagues throughout Scotland for providing follow-up information when available; to Robert James and the staff at the Biochemistry Department, Crosshouse Hospital for their skilled technical assistance; and to Rob Jones of Chromsystems Instruments & Chemicals GmbH, Munich, Germany.