Steroid Profiling for Adrenocortical Disorders: A Pathway for Omics-Based Diagnostics

During the past 15 years, a considerable investment of effort and funds has been directed to the use of omics-based technologies for biomarker identification and development. The outcome in terms of laboratory tests applicable to the routine environment has been slow to materialize, in some cases corrupted by poorly conceived and conducted work (1, 2). This disappointing return is in part attributable to the time and expertise required to meet the many steps and processes required from biomarker discovery to validation and acceptance for clinical use. Compounding the problem, application of an omics-based panel of multiple measurements requires considerably more complex approaches for test validation and interpretation than those required for more commonly used single biomarkers. In this issue of Clinical Chemistry, two groups, one at King's College (3) and a second at the Mayo Clinic (4), highlight the potential applications of mass spectrometry-based steroid profiling (“steroidomics”) for disorders of the adrenal cortex. With respective panels of 13 and 26 steroids, the outlined methods illustrate pathways for validation of omics-based signatures for diagnostics and stratification of disease for therapeutic intervention.

Application of steroid profiling for the diagnosis of adrenocortical disorders has a history spanning >50 years, first using laborious chromatographic separation and colorimetric or immunoassay procedures (5–7) subsequently replaced by methods using gas chromatography or liquid chromatography combined with mass spectrometry (8, 9). GC-MS methods, in particular, have proven extremely useful for steroid profiling, but because of the need for complex sample preparation, they are not well suited for the routine laboratory environment. Consequently, methods using liquid chromatography at the front end of mass spectrometry instruments have come to the forefront for routine laboratories. Whereas the King's College panel of 13 serum steroids uses an LC-MS/MS instrument common to many routine laboratories, the Mayo Clinic panel of 26 urinary steroids uses liquid chromatography in conjunction with Orbitrap mass spectrometry technology, more common to the research environment.

Today's new generation of hybrid Orbitrap or time-of-flight instruments combines the high mass resolving power of those technologies with accurate quantification and selectivity of quadrupole instruments. These hybrid instruments thereby allow both untargeted discovery and targeted quantitative omics applications. The result for targeted applications, such as that from the Mayo Clinic, is substantially improved resolution of ions with similar masses. Nevertheless, the instruments do not allow discrimination of isobaric isomers, which, as shown in Table 1 of the report from the Mayo Clinic, are common to numerous steroids in the panel. Presumably there are many other steroids not in the panel that may further contribute to potential problems related to separation of isobaric isomers. Therefore, chromatographic separation remains crucial so that both applications involve total run times of 20 to 35 min. As outlined in the report from the Mayo Clinic, some reduction in run times may be achieved by multiplexing. A potential alternative approach could be to include ion mobility, either alone or as a second dimension of rapid separation following a shortened liquid chromatography step. Either way, in terms of run times, the high mass resolution method does not appear to offer any advantage over standard LC-MS/MS methods. Perhaps of more importance for the high mass resolution method are the more easily interpretable ion spectra in a highly complex sample matrix present even after specimen cleanup.

Another difference between the 2 methods concerns choice of sample matrix. Because of highly dynamic diurnal changes, measurements of steroids in serum or plasma should be determined at a specific time during the day, with reference intervals matching to that period (10). The ideal time for sampling is at the nadir in the diurnal rhythm, typically close to midnight, which is impractical for outpatient blood sampling. The more practical time for patients is in the morning, but this is when plasma steroids are at their peak and when it can be more difficult to distinguish patients with adrenocortical hyperfunction from those with normal function. Twenty-four–hour urine collections have the advantage of providing an integrated measure of steroid output but also involve practical difficulties for patients. Completeness of collections is always a concern. Spot first or early morning urine collections may offer an alternative approach with the potential added advantage of collections over the nadir in the diurnal rhythm.

As outlined in the report from the Mayo Clinic, another disadvantage of urine collections is the need to consider sulfate and glucuronide conjugates, which, as major urinary end products of steroid metabolism, require a deconjugation step before sample cleanup. Without conjugated steroids for both calibration and quality control purposes, there is potential for error and uncertainty in final measurements. As pointed out by the King's College group and apparent in the report from the Mayo Clinic, the need to consider a wide array of largely obscure urinary metabolites rather than better known steroids that comprise the backbone of the biosynthetic pathway is a further factor that might be limiting to clinical acceptance.

In terms of clinical applications, mass spectrometry-based diagnostic methods have the potential to markedly improve the treatment of patients with adrenocortical disorders. Adrenal tumors, in particular, with a prevalence of at least 3% in individuals >50 years old, are the most frequently encountered neoplasms in humans and a substantial problem for diagnosis and management. Although the majority of these tumors are benign hormonally inactive adenomas, a small proportion of up to 8%, dependent on size, are extremely aggressive malignant adrenocortical carcinomas (11). The clinical challenge is that there is no single diagnostic tool that can discriminate benign from malignant tumors. Adrenocortical carcinoma has a 5-year survival rate of <50%, and only early detection at a localized stage can lead to surgical cure (12, 13). Otherwise, the median survival period is <15 months (14). Thus, improved diagnostic tools are urgently needed.

Both reports in this issue of Clinical Chemistry indicate distinct patterns of steroids potentially useful for distinguishing patients with adrenocortical carcinoma from those with other adrenocortical lesions. Among steroids of the serum panel, 11-deoxycortisol, 17-hydroxypregnenolone, and other intermediates in steroid metabolism were the most prominently increased in patients with adrenocortical carcinoma. These findings were mirrored by increases in respective urinary metabolites reported by the Mayo Clinic group. Although the consistent results from both studies are encouraging and in line with a previous European multicenter study that used GC-MS (8), the small patient numbers for both studies limit enthusiasm. Nevertheless, the heterogeneous nature of serum steroid patterns among patients with adrenocortical carcinoma also raises the possibility that differences in signatures may have prognostic significance that might impact treatment decisions.

Clearly further larger series are required, ideally involving carefully planned prospective trials with detailed clinical annotations and follow-up that can establish not only utility for diagnosis but also stratification of patients for therapy. With rare diseases such as adrenocortical carcinoma, any prospective trial would best take advantage of established multicenter networks, such as the European Network for the Study of Adrenal Tumors, to enable recruitment of sufficient patients and collection of data using already established procedures and registries.

Suitably large-sized reference populations are also required to fully characterize both sex differences and the highly dynamic age-related changes in steroids (15). The Mayo Clinic group, to their credit, adjusted results for individual patients according to sex and age. Nevertheless, the reference populations outlined in both reports are insufficiently large to provide truly useful age- and sex-specific reference data. Adding to the complexity of test interpretation and validation of biomarker panels are not only requirements to consider age- and sex-specific reference intervals but also the need in omics-based analyses to consider patterns of test results.

For interpretation of patterns of results in test panels, it is necessary to develop computational models and tools to deal with large sets of generated data. Appropriate and easily understandable reporting of results for clinicians represents another challenge to the development and clinical implementation of omics-based laboratory tests. For all this it is important to have the involvement of multidisciplinary teams well versed in the development and validation of tests for subsequent translation to the clinic. Both the teams at King's College and the Mayo Clinic are suitably experienced and well situated for bringing their methods into the routine clinical environment. The Mayo Clinic laboratory group headed by Ravinder Singh, a pioneer in the field of clinical mass spectrometry, is supported by the Bancos team with complementary expertise in clinical endocrinology and information technology. The Taylor laboratory team is also well supported by strong clinical partnerships at King's College. Such teams promise a pathway model for others in the omics-based biomarker field to follow. More specifically for adrenocortical disorders, the outlined methods and data from the 2 laboratories represent the first steps of important progress in bringing mass spectrometry-based steroidomics closer to hand for the routine clinical laboratory.

References