Clinical Chemistry: The Road to N=1
Why we need more chemistry in the clinic if we’re going to reach the goal of individualized medicine
Peter Kissinger |
At a Glance
- A lot of medicine focuses on physical statistics and not enough on chemistry statistics
- Chemistry has been a part of medicine for a long time, but ihas only recently become part of critical care
- It could help us further personalize care to better serve patients
- Combining as many variables and statistics as possible will build a better personalized picture
Sixty years ago, as an elementary school student, I was required to complete a physical examination in order to join an athletic team or participate in summer camp. At the time, such exams were fully in the domain of physics. The available tools measured height and weight, and included a chilly stethoscope, a blood pressure cuff, a rubber hammer, and a mercury thermometer. There was a little device with a bright light used to peer into my ears, nose, and throat. Virtually no chemical measurements were made beyond looking at the clarity of urine and a semi-quantitative test for sugar therein.
Following a recent morning encounter with my physician, I told a class of premeds that I’d just had a “pchem” exam. I related how a “physical” had become a “physical chemistry” exam, with the doctor showing me tables of numbers on a tablet computer, enabling comparisons with reference ranges and my own longitudinal data. Those same data are now available to me anywhere on planet Earth. Clinical chemistry has come a long way in my lifetime, and it is advances in instrumentation that have had the biggest impact on medicine. The microscope and the thermometer got us started, but even these are recent advances considering our history of several hundred millennia.
Where it all began
Clinical chemistry is a relatively new component of critical care and the community hospital setting, and even newer in routine diagnostics. The history of the field began with some fabulous pioneers, such as Donald Dexter Van Slyke (1883-1971), Joseph J. Kleiner (1897–1974), Arnold Orville Beckman (1900-2004), Wallace H. Coulter (1913-1998), Leland C. Clark (1918-2005), Solomon Aaron Berson (1918-1972), Lenard Tucker Skeggs, Jr. (1918-2002), Rosalind Sussman Yalow (1921-2011), and John Wendell Severinghaus (b. 1922). These great minds were tinkerers – they did not follow a strategic plan, create PowerPoint slides or speak of reimbursement codes or third party payers. There is no room here to dig deep into the history of clinical chemistry, but a great place to start learning more is a short review by Larry Kricka and John Savory, published in 2011 (1). My point: clinical chemistry is largely a post-WWII phenomenon which in many respects did not accelerate until the 1970s. Diabetics had no means to monitor glucose at home even modestly well until 1980, 50 years after insulin became a drug. The American Association of Clinical Chemists began in 1948 and about thirty years later, just as I joined, the name was changed to the American Association for Clinical Chemistry, implying advocacy and welcoming a wider demographic.
The age of complexity
With the human genome project, we were thought to be on the cusp of a great advance in diagnostics, but we now know that knowledge of genes alone are not enough. Next, at the turn of the millennium, we thought that the proteome would be the answer. The terms “biomarker” and “molecular diagnostics” were invented, but once again, the new dawn of diagnostics failed to materialize. Now, we are moving onto metabolomics – will it deliver? Only time will tell. All of these areas have potential to develop further, but it will require more investigative effort than was initially thought.
Each person is unique and defined by much more than their genome, which itself is less stable than we thought. Our proteins are in constant post-translational flux, depending on the time of day, the time we last ate, and the time a drug entered the body. What we consume, the variability of our microbiome, and the state of various organs are not reliably programmed at birth. Yet we largely operate with the tyranny of averages – we chase p-values (2), and find more correlations and probabilities than we do mechanisms. When a physician is confronted with a unique patient, averages aren’t much help. While some analytes are reliably fixed in a homeostatic fashion, these appear to be very scarce. When one doctor sits with one patient, more often than not, intuition based on experience matters most.
Clearly, we need to explore the virtue of more chemistry measures versus time. In the ICU, displays still mainly focus on physical metrics. The only routine chemical measure is oxygen saturation. But physical measures of temperature, blood pressure and heart rate are all responding to chemistry. When they wander too far, we take a blood sample, but could the problem have been anticipated? We dose a drug based on such crude notions as 10 mg for all or mg/kg or mg/m2. Shouldn’t we be dosing to achieve a measured exposure? Isn’t concentration in circulation a better concept of dose than a pill swallowed or a bolus infused? Shouldn’t drug monitoring be the most common companion diagnostic, especially in critical care where drug–drug interactions and organ system deficiencies are likely? Getting the right drug at the right dose at the right time is not often a genomics problem. Likewise, every child matures biochemically and physiologically in a way that does not follow a consistent timeline – shouldn’t we be measuring more? Is it not odd that a bioanalytical chemist who has lived seven decades has never had a single measurement of the circulating concentration of a prescribed drug? I’ve never even been tested for glucose tolerance. Pianos get tuned more often. My doctor tells me my hemoglobin A1C is average, but averages can come from an infinite number of data sets. I want to know my variance, the method variance, and a subpopulation variance (old men, in my case) (3).
So much for venting. Things are improving – we are doing more point-of-care testing, although it is still limited. We are getting closer to N=1 personal reference ranges and we have access to our own electronic health data. We can make measurements in smaller volumes of blood than ever before and can now do a lot of tests with 0.1 mL, a few with 0.01 mL, and some with less than 0.001 mL. However, we still frequently take far more blood than we need. There have been several reports of anemia resulting from too many blood draws with cardiac patients (4)(5) and we’ve all heard of excessive ordering of diagnostic tests. I suspect that most of the volume of those blood draws was thrown away, and this can be confirmed by a visit to your local clinical chemistry laboratory – more than one major lab has told me “all of our automation is based on sample tubes large enough to hold a bar code”. The patients are waiting for bioanalytical chemists, clinical chemists, and pathologists to improve this situation. The tools are getting better, and among them are mass spectrometers, which in the clinical world are now at the stage where the Skeggs’ AutoAnalyzers were in the 1960s.
Mass spectrometry as an analytical resource is older than pH meters, oxygen electrodes, and immunoassays, but is relatively new to diagnostics. Performance is good, but there remain significant challenges for quantitative work in clinical chemistry, including many nonlinearities whereby variable matrix components influence the response for the desired analyte(s). Many do not fully understand this matrix effect and its impact on method validation. Mass spectrometry technology is not yet economically competitive for random access, allowing for rapid examination of different analytes in each of a series of samples using a single instrument. This is especially impactful for intensive care clinical applications where rapid turnaround time can be critical. On the other hand, when samples are numerous for a single analyte or panel, and time is not critical, there is no better performance for the price.
A major worry in clinical chemistry today is the difficulty in finding properly collected and characterized samples from carefully controlled biology. Sampling matters – every bio sample comes with a set of attributes, which too often are incomplete, with time (chronobiology), nutrition, polypharmacy, and comorbidities rarely available in any detail. Understanding of the problem, the will to do better, and the money to improve are generally in short supply, but the time has come to fix these deficiencies. In the age of “big data” it is clear that a lot of those data are not as good as they need to be – too often, there is no sorting out the biology inferences from sampling errors and analytical variances, and a reproducibility crisis has been
Some suggest that the traditional annual physical examination is not very helpful (6). I’d certainly prefer a quarterly chemical examination, but I want reliable numbers. Some have advocated building facilities for chemical examination at local pharmacies or even grocery stores (7), but the recent scandal involving Theranos and their founder suggests that the proposed enabling technology is not what was promised (8). The resulting book and movie will bring bioanalysis into view for a wider audience than ever before (9) – we can only hope that the negative publicity will not derail the efforts of the wider clinical chemistry community to make blood tests more comfortable, affordable, and efficient.
Peter Kissinger is Professor, Brown Laboratory of Chemistry, Purdue University, and a founder of Bioanalytical Systems, Inc. (BASi), Prosolia, Inc., and Phlebotics, Inc. Indiana, USA.
- L Kricka, J Savory, “A guide to the history of clinical chemistry”, Clin Chem, 57,1118–1126 (2011).
- TM Annesley, JC Boyd, “The P value: probable does not mean practical”, Clin Chem, 60, 1021–1023 (2014).
- E Lenters-Westra et al., “Biological variation of hemoglobin A1c: consequences for diagnosing diabetes mellitus”, Clin Chem, 60, 1570–1572 (2014).
- A Salisbury et al., “Diagnostic blood loss from phlebotomy and hospital-acquired anemia during acute myocardial infarction”, Arch Intern Med, 171, 1646–1653 (2011).
- C Koch et al., “Contemporary bloodletting in cardiac surgical care”, Ann Thorac Surg, 99, 779–784 (2015).
- Medscape Multispecialty. WebMD Health Professional Network. Ritual, not science, keeps the annual physical alive. Available at: wb.md/2r0WybU . Accessed April 26, 2018.
- Theranos Inc., “Theranos”, (2018). Available at: bit.ly/1Gwh5DO. Accessed April 26, 2018.
- DDN News, “Theranos: lessons learned”, (2018). Available at: bit.ly/2r0LPPG. Accessed April 26, 2018.
- J Carreyrou. “Bad Blood: Secrets and Lies in Silicon Valley”, Knopf: 2018.