The Challenges and Opportunities of Prospective Longitudinal Studies
Is it possible to accelerate the future?
What is exciting about working in the field of clinical molecular diagnostics is its promise of real impact on human health. Yet the path from biomarker discovery to an FDA-approved test in general use is a long and painstaking process: not only the technology and all of its underpinnings have to be proven reliable and reproducible, they also have to be proven to provide a clinical benefit.
In the parlance of the molecular diagnostic industry, it has to not only have analytical validity, it has to also have clinical validity, and more so clinical utility.
Analytical validity, clinical validity, and clinical utility
Some definitions are in order: analytical validity is the ability of a test to detect and measure the marker of interest – it could be DNA, RNA, protein or even a metabolite. How repeatable is this test when you perform it a number of times on the same sample? How reproducible is this test under different conditions, such as different equipment and different operators? How sensitive is this test under a variety of samples that have been through all the pre-analytic variability that naturally arise?
For clinical validity, is the biomarker being tested clearly associated with a disease or a response to treatment? The answer to this question involves clinical performance, including sensitivity (‘rule-in’ those who test positive for the disease or condition), specificity (‘rule-out’ those who test negative), and positive and negative predictive values (PPV, NPV) are derived from a combination of sensitivity, specificity, and prevalence of the condition. (For more details and examples see Wikipedia.)
It should be noted here that for clinical validity measurements, sensitivity and specificity can be estimated from case-control studies, while PPV and NPV values need a cross-sectional, population-based study in order to include prevalence in a population at a given point in time.
For clinical utility, that is where the true difference of a diagnostic test is measured against metrics of patient health. Does the test improve patient outcome or improve quality of life? Will the test confirm or change the diagnosis of a patient’s condition? Will it change clinical intervention? And for genetic tests, the diagnostic test may have familial and social impacts regarding the identification of at-risk family members or ethnicities.
You can think of this as concentric circles of time and expense; the technical side is the first major step that takes time, effort and resources; the clinical validity takes even more time, effort and resources, and for clinical utility that is the final goal.
A public health debate on low-dose aspirin and cholesterol-lowering statins
It was a few years ago a debate began whether low-dose aspirin and cholesterol-lowering statins should be administered to the general public as a ‘wonder-drug cocktail’ to extend health and lifespan. For example, in 2015 a US Federal Task Force recommended daily low-dose aspirin for individuals at high-risk for disease with an additional benefit of potential benefit of reducing risk of colorectal cancer if taken for more than 10 years.
And for statins, way back in 2005 the results of large-scale clinical tests for cardiovascular health and cholesterol-lowering effects of statins measured an interesting positive side-effect: improving bone strength, reducing incidents of stroke, cancer and arthritis.
How can these two hypotheses, that low-dose aspirin combined with statins over a long period of time in humans, be tested?
Defining a Prospective Double-Blind Clinical Trial
Enter the prospective, double-blind clinical trial. A clinical trial is a study done in humans; double-blind means neither the researcher nor the patient-subject knows whether they are giving or receiving a placebo or an active agent being tested. In the context of diagnostics, the laboratory doing the testing is given blinded samples – that is, the laboratory testers do not know beforehand whether a given sample is ‘positive’ or ‘negative’ regarding the condition being tested.
And by prospective, this test is monitoring that patient-subject from one point in time to a given end-point, which in many cases is many years, and sometimes decades. The prominent example of a prospective study is the Framingham Heart Study, started in 1948 with 5,209 adults and continues with its third generation of participants.
Why clinical trials are expensive
Clinical trials to measure clinical utility, that is its effect on patient health outcomes, take many years of work and involve thousands or tens of thousands of patient research subjects. At this scale, the work and time and expense all add up: individual clinical trial participants need to enroll, be selected, interviewed, examined by a clinician, clinical samples need to be taken; samples need to be handled and stored, samples need to be tested; participants re-examined over time, likely additional samples collected, stored and tested; several years later the data collated, processed, analyzed and summarized.
Therefore the hypothesis that needs to be tested has to be a strong one. In the example of low-dose aspirin, such a study has been initiated, called ASPREE (ASPirin in Reducing Events in the Elderly). Partially funded by the US National Institutes of Health (NIH), the study set out to examine the effects of low-dose aspirin on healthy individuals age 70 and older, begun in 2010 and ended in 2014, and its results published in September 2018 in the New England Journal of Medicine. Enrolling a total of 19,114 individuals, half randomly assigned aspirin and half a placebo, and followed up after five years to look for a primary end-point of death, dementia or physical disability.
The results are in, and the conclusions were surprising; from the abstract, “Aspirin use in healthy elderly persons did not prolong disability-free survival over a period of 5 years but led to a higher rate of major hemorrhage than placebo.”
As a clinical trial, the clinical utility of administering low-dose aspirin to healthy individuals 70 years old and greater are clear, and its impact on clinical care and physicians’ advice to the general public affected. Although it took many years and many millions of dollars, this is important work with impact on public health, even if the results were somewhat contrary to the original hypothesis.
The power of biobanks and molecular technology
One of the surprising resources in China has been the establishment of the Taizhou Longitudinal Study. Initiated in 2008 by Dr. Jin Li of Fudan University in Shanghai, this study set out to explore the environmental and genetic risk factors for common non-communicable diseases, setting out to enroll 100,000 healthy individuals 30 to 80 years of age drawn from the general population.
One feature of the Taizhou population was its high rate of cancer mortality; according to a survey of causes of death from 1988-2002, the rate at 288/100,000 person-years is almost three times the national rate in China of 108/100,000 person-years. Esophageal, stomach and liver cancer are the highest in prevalence.
The primary diseases for this study are cancer, cardiovascular disease, and cerebral vascular disease (stroke). The particulars of the study design and rationale were published in 2009 and of the 120,000 enrollees, about 1% (a total of 1,200) received a diagnosis of cancer.
Through access to blood samples drawn every two years throughout the study, biobanked samples can go back to those individuals who were diagnosed with cancer and then go back retrospectively to examine those samples before any symptoms appeared.
Singlera Genomics had access to the samples of 575 individuals diagnosed with cancer from the top five types by prevalence in the Taizhou cohort: esophageal, stomach, liver, lung and colon. The 575 individuals also had another unique feature: blood samples that were collected at least four years before conventional cancer diagnosis.
You can access the PDF of the PanSeer assay as applied to these Taizhou Longitudinal Study samples here (link to PanSeer Technical Note PDF). If you’d like additional detail, you can access the ASHG 2018 poster “Circulating tumor DNA methylation haplotypes in plasma can detect cancer four years prior to conventional diagnosis” by registering here.
The future cannot come quickly enough
Naturally a case-controlled prospective study is in order with the Singlera Genomics technology that looks at circulating tumor methylated DNA (ctmDNA). There are many interesting hypotheses to test using this innovative and exciting technology, which may bear out the promise of liquid biopsy.
It will take time, effort and an investment in the future to verify these retrospective clinical results in other populations, but this data speaks for itself as a blinded retrospective study. This kind of data is unusual for the novel analyte of circulating free methylated DNA.
Do you have access to an interesting set of samples and are looking for a novel method for a new type of biomarker? Singlera is interested in partnership and collaboration opportunities – contact us today.