Crossing the Chasm of Death for Biomarkers

An interview with Christiane Auray-Blais, LL.M., Ph.D.

Biomarkers, or biological markers, have intuitive appeal, since chemists and biochemists are trained to relate activity to molecular structure and metabolism. I attend over 20 scientific meetings per year, where technical reports often claim discovery of a potential biomarker. Most frequently, this is the end of it. The biomarker is unable to cross the “chasm of death.” Reporting authors seem to have little incentive and capability to follow up by converting their discovery to a useful assay or therapy.

Let’s look at one success story to see what is involved. Perhaps if we understand the critical success factors, we can see a solution to the problem.

Despite the attention that goes with discovering a “biomarker” as an apparent correlate and thus an indicator of disease, the concept is not new. For example, the Quebec Neonatal Mass Urinary Screening Program was started in 1971 at the Centre Hospitalier Universitaire de Sherbrooke (CHUS)/Université de Sherbrooke, supported by the Ministry of Health and Social Services in the province of Quebec.1 Over the last four decades, this lab has received and processed almost three million samples2 looking for molecular indicators (now called biomarkers) of hereditary metabolic diseases in urine samples collected by parents from neonates at 21 days of age.1

Voluntary cooperation of the parents averages 90%. Disorders diagnosed include methylmalonic aciduria, argininosuccinic aciduria, citrullinemia, and several others. Today’s assay panel detects an abnormal sample in 1 in 1500 babies. Apparent abnormals are further analyzed with more specific techniques such as GC (organic acids) and ion exchange resin methods (amino acids). The screening program is headed by Christiane Auray-Blais and her dedicated team.

Assay development using LC/MS/MS methodology

Dr. Auray-Blais also has a research program to develop assays to extend the disease panel for high-risk screening. She chose to focus on lysosomal storage disorders (LSDs) such as mucopolysaccharidoses (MPS) and Fabry disease. These disorders are caused by inherited enzymatic deficiencies leading to accumulation of storage products in various tissues, blood, and urine. Several types of MPS have been reported.

Since existing assay methodologies were not very useful for some MPS, she decided to explore the development of assays using LC/MS/MS. She developed an efficient method for the analysis of some glycosaminoglycans, heparan sulfate, and dermatan sulfate, as biomarkers for MPS I, II, and VI, using urine samples collected on filter paper.3

Dr. Auray-Blais’s lab has reported a series of validated LC/MS/MS biomarker assays that are being used for diagnosis of LSDs as well as high-risk screening. One of the successful methodologies was the assay of globotriaosylceramide (Gb3) for Fabry disease using a urine sample dried on filter paper.4

In contrast to other LC/MS/MS reports, her team has validated the assay, which is now used routinely in the clinical as well as research fields. Further, the assay has been successfully transferred to other labs.

Recently, I had an opportunity to interview Christiane Auray-Blais on the development of the LC/MS/MS assay for biomarkers with a particular focus on LSDs in urine.

RLS: Christiane, how did you become interested in applying LC/MS/MS to Fabry disease and other LSDs?

CAB: I’ve been interested in lysosomal storage disorders for some time because they are diseases where often an early detection and intervention will change the outcomes for these affected patients who can present with severe clinical manifestations. LSDs are complex and heterogeneous diseases. Methods developed for analysis of biomarkers were time-consuming and labor-intensive, with many false-positives (and possibly false-negatives), such as in the case of mucopolysaccharidoses, where a spectrophotometric method is not reliable, but still widely used in clinical labs. Also, the analytics were not very useful in guiding therapy.

Treatment is now possible with enzyme replacement therapy for some of these LSDs. We thus felt that better analytics might be useful in detecting LSDs in patients as early as possible and for guiding therapy. As it turned out, some LSDs differ in gender and individual response. It was one of these things that, as we improved the analytics, we started to better understand the biochemistry in order to improve therapy, monitoring of patients, and their outcomes.

RLS: Many research reports claim a biomarker based upon an observed correlation. Yet few of these proposed biomarkers turn up in practice. Why?

CAB: There can be many reasons. The correlation may be coincidence unsupported by the underlying biochemistry. Or the relationship is not direct, but due to downstream events. Even if the candidate biomarker is well supported as being a key player in the biochemistry of the disease, developing and validating the assay to be eventually transferred to the clinic requires lots of careful work and tedious documentation.

RLS: So what are your key goals in developing and validating an assay?

CAB: At a general level, the assay must consistently provide useful information that can be relied upon to support decisions. The assay must be robust! This means that we need to investigate the effects of common laboratory variables—chemical, instrumental, and human. Weak spots must be described and mitigated.

On a personal level, the assay should be designed to fit into existing laboratory operation. At the end of the day, I want my staff to feel that they have accomplished something and that their results are worth the effort. We need to have confidence and pride in our work. This philosophy has paid off in dedication: Key members of the Newborn Screening Lab have several decades of dedicated service.

Biomarker discovery and analysis method

RLS: Human factors are important, and I’m not dismissing them, but what makes a robust assay?

CAB: Let’s start from the beginning. First, the biomarker discovery: We used a Synapt™ UPLC-QTOF MS (Waters Corp., Milford, MA). We detected novel lyso-Gb3-related analogs for Fabry disease by using a metabolomic approach.5 We confirmed that these analogs were modifications on the sphingosine moiety of lyso-Gb3. These biomarkers had not been reported previously. More importantly, some appear to be diagnostically significant for Fabry patients having a cardiac variant mutation. We therefore decided to devise a relative LC/MS/MS quantification for these analogs as a multiplex assay.

Robust assays require careful design combined with investigation of potential causes of imprecision and bias. Some of this is based upon years of laboratory experience. Choosing the right instrumentation is also essential since it affects the detection. For the multiplexed assay of lyso-Gb3 and related analogs for Fabry disease,6 we selected an Alliance 2795 XE and a Quattro micro tandem quadrupole MS/MS from Waters Corp. We operated in positive electrospray ionization (ESI) mode, and the signals were acquired during a multiple reaction monitoring (MRM) experiment. The column was a Waters 3-μm Atlantis Silica HILIC (hydrophilic interaction liquid chromatography) (2.1 × 50 mm). Different columns were evaluated before selecting the right one.

These analytical systems gave cycle times of 11 min or less. We avoided running at the extreme limits. We chose good separation and resolution parameters, which favor consistency and robustness at the expense of speed. Plus, we expected to use MRM for the relative quantitative analysis. The MS needs time for multiple MRM protocols, and we also need multiple scans per peak to estimate peak shape for precise quantitative analysis.

RLS: Robust assays often use internal standards. Did you use one or more? How were they selected?

CAB: Yes, internal standards improve assay robustness and often solve the problem of the “matrix effect” (enhancement or suppression of ionization). For this assay, we selected lyso-Gb3-Gly, which has a primary amine function, as do the analogs, but a unique molecular weight. Because of the similar structure, the internal standard (IS) was expected to show similar extraction and ionization behavior compared to the lyso-Gb3 analogs. The IS was effective in reducing the %CV from greater than 30% to single digits for lyso-Gb3 and analogs. This was a key success factor.

We prepared the IS in our lab and purified it with HPLC. Various tests confirmed purity and molecular weight. We then prepared a stock solution that was stable for at least nine months when stored at –80 °C. An aliquot of the stock solution was diluted and stored at 4 °C. This appears to be stable for at least three months.

RLS: Did you normalize to creatinine to compensate for urine dilution?

CAB: Yes, we normalize all urine assays to creatinine.7 This is essential, since earlier work in our lab showed a tremendous variability in Gb3 excretion in normal children who were less than six months old.8 Thus, we recommend that the reference healthy controls of infants or adults be age-matched to patients and, for some diseases, also gender-matched.

Sample prep

RLS: What about sample prep?

CAB: Sample preparation with either liquid/liquid extraction or solid-phase extraction cartridges is often useful to develop a robust method. For example, we used mixed-mode cation exchange cartridges for the analysis of lyso-Gb3 and related analogs in order to remove interferences such as salts, acids, and other unwanted compounds that can cause variable ionization and contaminate the ESI interface.

RLS: What other items did you investigate to assure a robust method?

CAB: Based upon our decades of experience with diagnostic assays, we routinely investigate stability at different temperatures (room temperature, 4 °C, –20 °C, and –80 °C) and freeze/thaw cycles. We also evaluate the adhesion to plastic and glassware. Solvents are thus chosen to minimize this issue and favor recovery and reproducibility.

Method validation and method transfer

RLS: How was the method validated?

CAB: Intraday and interday assay method precision was measured with urine samples from high- and low-excretor Fabry patients. Accuracy was evaluated with spiked urine samples at high, medium, and low concentrations. For lyso-Gb3, linearity was measured with a nine-point calibration curve covering the range of 0–28,000 pmol/L. Limits of detection and quantification were also evaluated.

RLS: You mentioned that the lyso-Gb3 peak from HILIC also contained seven novel analogs. Are these diagnostically significant?

CAB: Yes, some of these analogs appear to be significant and useful, especially for male patients, who show much higher urinary excretion. Within the male cohort, the patterns are complex. At the individual level, we were able to see significant changes in the pattern in response to enzyme replacement therapy for males and females. We also noticed that some of these analogs are excreted in biological fluids of patients having cardiac variant mutations of Fabry disease for whom no reliable biomarkers existed.

RLS: Method transfer to other labs is a sensitive metric of a method. How many labs have adopted your method? What problems did they encounter?

CAB: I know of labs in Europe and in the U.S. that have adopted our methods. Since we are careful to provide detailed information about the LC as well as the MS/MS method parameters and validation process when we publish our results, usually labs who want to adopt our methods have no difficulty reproducing our results.

Conclusion

RLS: You and your colleagues have worked hard to develop these assays. What advice would you give other labs that might be interested in developing assays from discovery of a probable biomarker?

CAB: On the most basic level, health is due to optimum operation of the various biochemical pathways in an organism. Diminishing or activating normal biochemical pathways leads to modification of normal performance and clinical manifestations.

One needs to be able to identify abnormalities with confidence, even if these are very rare. For example, in our Mass Newborn Urine Screening Laboratory, which is preventive genetic medicine offered to the population of Quebec and the Nunavut, we have processed almost three million specimens to detect inborn errors of amino acids and organic acids. This is an important base for us to expand the focus to include other rare diseases. In fact, we are in the evaluation process to add creatine synthesis and transport defects using LC/MS/MS to the existing urine screening program to detect these affected babies early and begin appropriate treatment.

Specifically, we at the laboratory in Sherbrooke, Quebec, enjoy several important advantages:

  • We have an infrastructure in place for population as well as high-risk screening of genetic diseases.
  • Assay development and validation fall within the charter of the lab.
  • Ethics has always been a priority in developing translational research projects.
  • Dedicated and knowledgeable personnel are essential, as well as a passion to help others.
  • The ability to obtain research funds and to have good partnerships with the industry (pharmaceutical and instrument technology companies) are a must.

Good instruments and technical and scientific support from the mass spectrometry industry are a major advantage. Being up-to-date with high-performing instruments is also important. For low-abundance molecules for different diseases and to obtain high sensitivity, we have recently acquired a Xevo™ TQ-S (Waters) for quantification of complex biomarkers.

Our staff is trained, experienced, and proficient in running critical screening assays in a high-throughput setting. They take pride in their work. They use their experience to critically review their work flow to minimize variance associated with their work.

I recognize that our laboratory is ideally suited by our charter and funding to develop new assays that may be useful in treating disorders in young children and also adults. We are located in a medical university/hospital center that favors close contact between clinicians and researchers and translational research.

I think that our route for developing assays of biomarkers is simple and at lower cost, and by publishing these results, we offer the scientific and clinical communities the opportunity to use these tests for their patients. Indeed, I see a general trend of offering new assays as a service to the population.

RLS: I understand! Congratulations on developing a program that makes a significant contribution to the knowledge of mankind and also to human welfare.

References

  1. Auray-Blais, C.; Cyr, D. et al. Quebec neonatal mass urinary screening programme: from micromolecules to macromolecules. Inhert. Metab. Dis.  2007, 30(1), 515–21.
  2. Auray-Blais, C. Private communication, 2013.
  3. Auray-Blais, C.; Bherer, P. et al. An improved method for glycosaminoglycan analysis by LC–MS/MS of urine samples collected on filter paper. Clin. Chim. Acta  2012, 413, 771–8.
  4. Auray-Blais, C.; Cyr, D. et al. Urinary globotriaosylceramide excretion correlates with the genotype in children and adults with Fabry disease. Mol. Genet. Metab.  2008, 93(3), 331–40.
  5. Auray-Blais, C.; Boutin, M. et al. Urinary globotriaosylsphingosine-related biomarkers for Fabry disease targeted by metabolomics. Anal. Chem.  2012, 84(6), 2745–53.
  6. Lavoie, P.; Boutin, M. et al. Multiplex analysis of novel urinary lyso-Gb3-related biomarkers for Fabry disease by tandem mass spectrometry. Anal. Chem.  2013, 85(3), 1743–52.
  7. Auray-Blais, C.; Millington, D.S. et al. Gb3/creatinine biomarkers for Fabry disease: issues to consider. Mol. Genet. Metab.  2009, 97(23).
  8. Barr, C.; Clarke, J.T. et al. Fabry disease urinary globotriaosylceramide/creatinine biomarker evaluation by liquid chromatography-tandem mass spectrometry in healthy infants from birth to 6 months. Mol. Genet. Metab.  2009, 97, 278–83.

Robert L. Stevenson, Ph.D., is a Consultant and Editor of Separation Science for American Laboratory/Labcompare; e-mail: [email protected].